Patent Application
20220396289
At least one embodiment pertains to processing resources used to calculate a path through an environment using neural networks. For example, at least one embodiment, pertains to processors or computing resources used to calculate distances to a location in an environment using neural networks to calculate a path according to various novel techniques described herein.
Calculating a path through an environment is an important task in many contexts. In various cases, calculating a path to a goal through an environment can be difficult, such as when the goal and the environment undergo various changes. Calculating a path to a goal through an environment can also require large amounts of computing resources. Techniques for calculating a path through an environment may therefore be improved.
In at least one embodiment, a neural network, based on an environment, a position, and a goal position, calculates a distance between said position and said goal position along a feasible trajectory through said environment. In at least one embodiment, a position refers to any suitable location, area, and/or region of an environment. In at least one embodiment, a goal position refers to a position in an environment that one or more agents are to navigate to. In at least one embodiment, an environment, also referred to as a scene, refers to any suitable two-dimensional (2D) or three-dimensional (3D) environment that can comprise various obstacles. In at least one embodiment, an environment can correspond to any suitable real-world environment or simulated environment. In at least one embodiment, given an environment, a first position, and a second position, a neural network calculates a reaching distance from said first position to said second position, which refers to a distance from said first position to said second position along a feasible trajectory through said environment. In at least one embodiment, a feasible trajectory refers to a trajectory that is geometrically feasible and/or semantically plausible, as described in further detail below. In at least one embodiment, a feasible trajectory refers to any suitable trajectory through an environment that does not collide or otherwise intersect with any obstacles and/or inaccessible regions of said environment.
In at least one embodiment, a neural network represents an environment through an environment field that encodes reaching distances of various positions in an environment to a goal position. In at least one embodiment, a neural network represents an environment internally through an environment field. In at least one embodiment, an environment field is a representation that comprises a collection of values indicating a reaching distance for each position in an environment to a goal position. In at least one embodiment, an environment field represents an environment, in which each value of said environment field corresponds to a position of said environment. In at least one embodiment, a particular value in an environment field corresponds to a particular position in an environment, in which said particular value is a reaching distance value from said particular position in said environment to a goal position in said environment.
In at least one embodiment, an environment field can be represented through a visual representation, such as an image, in which different reaching distance values are assigned different color values. In at least one embodiment, one or more systems visualize an environment field by at least using a neural network to calculate a reaching distance value for each position in an environment to a goal position, and utilizing calculated reaching distance values to visualize said environment field. In at least one embodiment, an environment field visualization is continuous. In at least one embodiment, while various environment field visualizations described herein may comprise discrete sections, environment field visualizations can comprise continuous color gradients, continuous shading, and/or any suitable continuous visualizations that indicate reaching distance values.
In at least one embodiment, an environment field is utilized to guide dynamic behavior of agents inside an environment. In at least one embodiment, an environment is represented by an image, in which each position corresponds to a pixel of said image. In at least one embodiment, a position of an environment corresponds to any suitable location, region, or area of said environment. In at least one embodiment, an environment is structured as a grid, in which a position corresponds to a set of pixels that correspond to a grid cell. In at least one embodiment, a position corresponds to any suitable set of pixels. In at least one embodiment, an image is any suitable image, such as a red-green-blue (RGB) image, black/white (B/W) image, grayscale image, RGB-depth (RGB-D) image, and/or variations thereof. In at least one embodiment, an agent refers to any suitable entity, such as an autonomous device, robot, human, or other entity, that is navigating or otherwise moving through an environment. In at least one embodiment, a neural implicit function represents an environment internally through an environment field. In at least one embodiment, a neural implicit function is a neural network that approximates or otherwise models a function, such as a function that outputs a reaching distance to a goal position for a position in an environment. In at least one embodiment, a neural implicit function that approximates or otherwise models a function that outputs a reaching distance to a goal position for a position in an environment is referred to as an implicit environment function.
In at least one embodiment, one or more systems utilize a neural implicit function that represents an environment through an environment field to guide an agent navigating a scene, such as navigating an agent to reach a given goal in a physically plausible and/or feasible manner. In at least one embodiment, for each position in a scene, an environment field captures a distance from a position to a given goal position along a geometrically feasible and/or semantically plausible trajectory. In at least one embodiment, a geometrically feasible trajectory refers to a trajectory through an environment that does not collide with any obstacles of said environment. In at least one embodiment, a semantically plausible trajectory refers to a trajectory through an environment that is physically appropriate to follow by one or more agents. In at least one embodiment, for example, a semantically implausible trajectory can be a trajectory through an environment that requires a human agent to crawl under an obstacle; in this example, crawling may not be a physically appropriate action for said human agent. In at least one embodiment, one or more systems use a neural implicit function to calculate a plurality of paths through which an agent is to traverse or otherwise navigate. In at least one embodiment, one or more systems navigate an agent utilizing a neural implicit function by repeatedly moving said agent to a next location with a smallest reaching distance, until said agent reaches a goal position.
In at least one embodiment, neural implicit functions can be generalized to arbitrary goal positions and arbitrary scenes/environments, allowing generalization to different environment fields given different goal positions and/or scenes/environments using a single neural implicit function. In at least one embodiment, a neural implicit function is able to learn to determine or otherwise calculate continuous environment fields using discretely sampled training data. In at least one embodiment, querying a reaching distance between any position and a goal position only requires a fast network forward pass, enabling efficient trajectory prediction.
In at least one embodiment, to enforce semantic plausibility of an environment field in indoor environments, one or more systems define accessible regions, which refer to plausible regions for agents to appear in. In at least one embodiment, one or more systems determine accessible regions for various environments using one or more neural network models, such as a variational auto-encoder, and designate locations outside of said accessible regions as obstacles, also referred to as inaccessible regions. In at least one embodiment, one or more systems utilize a neural implicit function that represents an environment through an environment field in various human trajectory modeling in 3D environments, such as various indoor environments, outdoor environments, or other suitable environments, and agent navigation in 2D environments, such as various mazes or other suitable environments.
In at least one embodiment, a neural implicit function is trained using training data generated based on one or more environments. In at least one embodiment, accessible regions and inaccessible regions are defined for environments by one or more systems using various neural network models, and/or other appropriate systems. In at least one embodiment, training data is generated by one or more systems using any suitable path planning algorithm, method, system, and/or variations thereof. In at least one embodiment, one or more systems generate training data by processing an environment with defined accessible regions and inaccessible regions using one or more path planning algorithms to calculate one or more reaching distances from one or more accessible positions in said environment to one or more goal positions in said environment, in which said training data comprises one or more of said one or more reaching distances, said one or more accessible positions, and/or said one or more goal positions, and train a neural network using said training data.
In preceding and following description, numerous specific details are set forth to provide a more thorough understanding of at least one embodiment. However, it will be apparent to one skilled in art that inventive concepts may be practiced without one or more of these specific details.
In at least one embodiment, techniques described herein achieve various technical advantages, including but not limited to: an ability to use a neural network to represent an environment to facilitate guidance for agent behavior/navigation in said environment; an ability to use a neural network to represent an environment through an environment field that ensures geometric and semantic feasibility; an ability to use a neural network to calculate one or more reaching distances for one or more positions to a goal in an environment in a single forward pass; and various other technical advantages.
In at least one embodiment, one or more systems train a neural implicit function to learn to represent an environment through a continuous environment field, in which each value of each position of said continuous environment field indicates a reaching distance from a particular position of said environment to a goal position. In at least one embodiment, one or more systems model an environment field as a wave propagation to compute reaching distance. In at least one embodiment, a wave propagation refers to a phenomena in which a distance between an initial position and a goal is equivalent to a minimal amount of time required by a wave, starting from said initial position, to propagate its front boundary to reach said goal.
In at least one embodiment, one or more systems denote all positions that can be reached by an agent as accessible regions, and all other positions as obstacles, also referred to as inaccessible regions. In at least one embodiment, for a spreading wave denoted by τ (e.g., a closed curve in an environment), said wave spreads based on a speed function, denoted by f(x), defined at each location. In at least one embodiment, a current position of a spreading wave, denoted by τ, is modeled by an arrival time function, denoted by u(x), with respect to a location, denoted by x, starting from a goal position. In at least one embodiment, when f(x)>0, one or more systems formulate an arrival time function as a continuous shortest-path problem through an equation such as an Eikonal equation, or any suitable equation, that is represented through a following equation, although any variation thereof can be utilized:
∥∇u(x)∥f(x)=1,x∈Ω
in which Ω denotes a feasible region (e.g., accessible region), ∇ denotes a gradient, ∥⋅∥ denotes a Euclidean norm, and u(xe)=0 at a goal location denoted by xe. In at least one embodiment, one or more systems specify an environment layout in which, at obstacles, a wave expansion speed, denoted by f(x), is set to an infinitesimally small positive value, or any suitable value, as a wave cannot go through obstacles, and within an accessible region, a wave expansion speed is set to a constant value of 1, or any suitable value.
In at least one embodiment, one or more systems solve an Eikonal equation, such as those described herein, through a neural implicit function that models a continuous energy surface. In at least one embodiment, a neural implicit function, also referred to as an implicit environment function, neural network model, neural network function, machine learning function, implicit neural representation, and/or variations thereof, is a neural network that approximates one or more functions. In at least one embodiment, one or more systems train a neural implicit function to represent an environment through an environment field as a mapping from location coordinates, denoted by x∈R2, to an arrival time, denoted by u(x)∈
. In at least one embodiment, one or more systems train a neural implicit function using training data comprising discretely sampled pairs of {x, u(x)} to represent an environment through a continuous field.
In at least one embodiment, one or more systems utilize one or more path planning algorithms, methods, and/or systems, such as Breath First Searching (BFS), Dijkstra's algorithm, Fast Marching Method (FMM), Rapidly Exploring Random Tree (RRT), Probabilistic Roadmaps (PRM), and/or variations thereof, to generate training data for a neural implicit function. In at least one embodiment, one or more path planning algorithms, methods, and/or systems, based on an input environment and a goal position, determine a reaching distance from each accessible position of said input environment to said goal position. In at least one embodiment, training data comprises an environment, a goal position, and a reaching distance for each accessible position in said environment. In at least one embodiment, an accessible position refers to a position in an environment that resides in an accessible region.
In at least one embodiment, one or more systems utilize a path planning method that solves u(x) in one or more environments, such as an environment structured as a grid. In at least one embodiment, one or more path planning methods initiate a discrete map starting from a goal position, and iteratively expand and update neighboring feasible grid cells according to a speed using Eikonal updating criteria, or any suitable criteria, until reaching a starting position. In at least one embodiment, one or more path planning methods include a method, such as an FMM implementation, that obtains a discrete grid (e.g., grid corresponding to an environment) with a given goal position as inputs and outputs a reaching distance from each accessible position (e.g., position of an environment) to said goal position. In at least one embodiment, one or more systems train a neural implicit function to regress a reaching distance at each grid cell given a cell's coordinates (e.g., normalized to [−1, 1]) as inputs. In at least one embodiment,
In at least one embodiment, a training framework 102 is a collection of one or more hardware and/or software computing resources with instructions that, when executed, performs one or more neural network training processes, functions, and/or operations. In at least one embodiment, a training framework 102 is in accordance with those described in connection with
In at least one embodiment, training data 104 is generated by one or more path planning algorithms, methods, and/or systems based on one or more environments. In at least one embodiment, training data 104 is a collection of data that comprises one or more reaching distances from one or more accessible positions in an environment to a particular goal position in said environment. In at least one embodiment, training data 104 is implemented through one or more data structures, such as an array or list, or any suitable collection of data, that encodes reaching distances and positions corresponding to said reaching distances. In at least one embodiment, a position of an environment corresponds to any suitable location, region, or area of said environment. In at least one embodiment, an environment is represented by an image, in which each position corresponds to a pixel and/or collection of pixels of said image. In at least one embodiment, an environment is represented by a set of points (e.g., point cloud data), in which each position corresponds to a point and/or collection of points of said set of points. In at least one embodiment, training data 104 is generated by one or more path planning algorithms, methods, and/or systems that, based on an input environment (e.g., through an image, set of points, and/or variations thereof) and a goal position, outputs a reaching distance for each accessible position in said environment to said goal position.
In at least one embodiment, a training framework 102 trains an implicit environment function 106 using training data 104 to represent an environment with a particular goal position (e.g., an environment and a goal position used to generate training data 104) through an environment field. In at least one embodiment, an implicit environment function 106 is any suitable neural network model, algorithm, representation, function, and/or variations thereof. In at least one embodiment, an implicit environment function comprises various neural network models such as a perceptron model, a radial basis network (RBN), an auto encoder (AE), Boltzmann Machine (BM), Restricted Boltzmann Machine (RBM), deep belief network (DBN), deep convolutional network (DCN), extreme learning machine (ELM), deep residual network (DRN), support vector machines (SVM), and/or variations thereof.
In at least one embodiment, a training framework 102 trains an implicit environment function 106 by inputting a position of training data 104 to said implicit environment function 106, causing said implicit environment function 106 to calculate a reaching distance for said position, comparing said calculated reaching distance for said position to a reaching distance of training data 104, also referred to as a ground truth reaching distance, corresponding to said position, calculating loss using one or more loss functions based on said comparison, and updating said implicit environment function 106 based on calculated loss. In at least one embodiment, a training framework 102 updates one or more weights, biases, and/or structural connections (e.g., architecture(s) and/or configuration(s) of one or more components) of an implicit environment function 106 such that calculated loss is minimized. In at least one embodiment, a training framework 102 calculates loss using any suitable loss function, such as cross-entropy loss, binary cross-entropy loss, softmax loss, logistic loss, focal loss, and/or variations thereof.
In at least one embodiment, an implicit environment function 106 is trained when calculated loss for said implicit environment function 106 is below a defined threshold, which can be any suitable value. In at least one embodiment, a training framework 102 trains an implicit environment function 106 to represent reaching distances for positions to a particular goal position in an environment used to generate training data 104. In at least one embodiment, a training framework 102 trains an implicit environment function 106 to determine an implicit environment function 108, which is said implicit environment function 106 after one or more training processes. In at least one embodiment, an implicit environment function 108 is a trained implicit environment function 106. In at least one embodiment, an implicit environment function 108 is specific to an environment and a goal position utilized to generate training data 104.
In at least one embodiment, an implicit environment function 108 represents an environment (e.g., an environment used to generate training data 104) through an environment field. In at least one embodiment, an implicit environment function 108 represents an environment internally through an environment field representation. In at least one embodiment, an environment field is a representation of an environment in which each position of said environment field corresponds to a position of said environment, and each position of said environment field comprises or otherwise indicates a reaching distance value to a particular goal position. In at least one embodiment, a reaching distance value for a particular position of an environment field indicates a reaching distance from a corresponding position of an environment to a goal position in said environment.
In at least one embodiment, an implicit environment function 108 represents an environment via an environment field through one or more data structures and/or collections of data that encode positions of said environment and reaching distances for said positions to a particular goal position in said environment. In at least one embodiment, an environment field is visualized through an environment field visualization 112. In at least one embodiment, an environment field visualization 112 is an image in which each position of said environment field visualization 112 corresponds to a position of an environment field. In at least one embodiment, different colors, shades, patterns, and/or variations thereof of an environment field visualization 112 correspond to different reaching distance values. In at least one embodiment, referring to
In at least one embodiment, one or more systems input a query position 110 to an implicit environment function 108 to determine a reaching distance 114. In at least one embodiment, a query position 110 is a position in an environment, and is implemented through one or more data objects, types, and/or variations thereof, that encode said position. In at least one embodiment, a query position 110 comprises coordinate data indicating a position. In at least one embodiment, an implicit environment function 108 processes a query position 110 to calculate a reaching distance 114. In at least one embodiment, a reaching distance 114 is a reaching distance value and is implemented through one or more data objects, types, and/or variations thereof, that encode said reaching distance value. In at least one embodiment, a reaching distance 114 comprises a reaching distance value from a position in an environment indicated by a query position 110 to a particular goal position in said environment (e.g., an environment and a particular goal position utilized to generate training data 104). In at least one embodiment, one or more systems utilize a determined reaching distance 114 for various navigation tasks, as described in further detail with regards to
In at least one embodiment, one or more systems train an implicit function to predict different environment fields for different goal positions in a same environment, which can be referred to as a conditional implicit function.
In at least one embodiment, a training framework 202 is a collection of one or more hardware and/or software computing resources with instructions that, when executed, performs one or more neural network training processes, functions, and/or operations. In at least one embodiment, a training framework 202 trains an implicit environment function 206 using training data 204 to determine an implicit environment function 208. In at least one embodiment, training data 204 is generated by one or more path planning algorithms, methods, and/or systems based on one or more environments. In at least one embodiment, training data 204 is a collection of data that comprises one or more reaching distances for one or more accessible positions in an environment to one or more goal positions in said environment. In at least one embodiment, training data 204 is implemented through one or more data structures, such as an array or list, or any suitable collection of data, that encodes reaching distances, and positions and goal positions corresponding to said reaching distances.
In at least one embodiment, training data 204 is generated by one or more path planning algorithms, methods, and/or systems that, based on an input environment (e.g., through an image, set of points, and/or variations thereof) and one or more goal positions, output one or more reaching distances for each accessible position in said environment to said one or more goal positions. In at least one embodiment, training data 204 is generated by one or more systems by inputting an environment and one or more randomly sampled goal positions to one or more path planning algorithms, methods, and/or systems to determine one or more reaching distances from each accessible position in said environment to said one or more randomly sampled goal positions.
In at least one embodiment, a training framework 202 trains an implicit environment function 206 using training data 204 to represent an environment (e.g., an environment used to generate training data 204) with any suitable goal position through an environment field. In at least one embodiment, an implicit environment function 206 is any suitable neural network model, algorithm, representation, function, and/or variations thereof. In at least one embodiment, an implicit environment function 206 comprises various neural network models such as a perceptron model, a radial basis network (RBN), an auto encoder (AE), Boltzmann Machine (BM), Restricted Boltzmann Machine (RBM), deep belief network (DBN), deep convolutional network (DCN), extreme learning machine (ELM), deep residual network (DRN), support vector machines (SVM), and/or variations thereof.
In at least one embodiment, a training framework 202 trains an implicit environment function 206 by inputting a position and a goal position of training data 204 to said implicit environment function 206, causing said implicit environment function 206 to calculate a reaching distance for said position to said goal position, comparing said calculated reaching distance for said position to said goal position to a reaching distance of training data 204, also referred to as a ground truth reaching distance, corresponding to said position and said goal position, calculating loss using one or more loss functions based on said comparison, and updating said implicit environment function 206 based on calculated loss. In at least one embodiment, a training framework 202 updates one or more weights, biases, and/or structural connections (e.g., architecture(s) and/or configuration(s) of one or more components) of an implicit environment function 206 such that calculated loss is minimized. In at least one embodiment, a training framework 202 calculates loss using any suitable loss function, such as cross-entropy loss, binary cross-entropy loss, softmax loss, logistic loss, focal loss, and/or variations thereof.
In at least one embodiment, an implicit environment function 206 is trained when calculated loss for said implicit environment function 206 is below a defined threshold, which can be any suitable value. In at least one embodiment, a training framework 202 trains an implicit environment function 206 to represent reaching distances for positions to any suitable goal position in an environment used to generate training data 204. In at least one embodiment, a training framework 202 trains an implicit environment function 206 to determine an implicit environment function 208, which is said implicit environment function 206 after one or more training processes. In at least one embodiment, an implicit environment function 208 is a trained implicit environment function 206. In at least one embodiment, an implicit environment function 208 is specific to an environment utilized to generate training data 204.
In at least one embodiment, an implicit environment function 208 represents an environment (e.g., an environment used to generate training data 204) through an environment field. In at least one embodiment, an implicit environment function 208 represents an environment with any suitable goal position internally through an environment field representation. In at least one embodiment, an implicit environment function 208 determines an environment field representation based on an input goal position (e.g., goal position 212). In at least one embodiment, an implicit environment function 208 represents an environment via an environment field through one or more data structures and/or collections of data that encode positions of said environment and reaching distances for said positions to a particular goal position (e.g., input via a goal position 212) in said environment.
In at least one embodiment, an environment field is visualized through environment field visualizations 214A-214B. In at least one embodiment, an environment field visualization is an image in which each position of said environment field visualization corresponds to a position of an environment field. In at least one embodiment, different colors, shades, patterns, and/or variations thereof of an environment field visualization correspond to different reaching distance values. In at least one embodiment, referring to
In at least one embodiment, one or more systems input a query position 210 and a goal position 212 to an implicit environment function 208 to calculate a reaching distance 216. In at least one embodiment, one or more systems concatenate a query position 210 and a goal position 212. In at least one embodiment, a query position 210 is a position in an environment, and is implemented through one or more data objects, types, and/or variations thereof, that encode said position. In at least one embodiment, a query position 210 comprises coordinate data indicating a position. In at least one embodiment, a goal position 212 is a goal position in an environment, and is implemented through one or more data objects, types, and/or variations thereof, that encode said position. In at least one embodiment, a goal position 212 comprises coordinate data indicating a goal position.
In at least one embodiment, an implicit environment function 208 processes a query position 210 and a goal position 212 to determine a reaching distance 216 (e.g., an environment field value at a query position 210). In at least one embodiment, a reaching distance 216 is a reaching distance value and is implemented through one or more data objects, types, and/or variations thereof, that encode said reaching distance value. In at least one embodiment, a reaching distance 216 comprises a reaching distance value from a position in an environment indicated by a query position 210 to a particular goal position indicated by a goal position 212 in said environment (e.g., an environment utilized to generate training data 204). In at least one embodiment, one or more systems utilize a determined reaching distance 216 for various navigation tasks, as described in further detail with regards to
In at least one embodiment, one or more systems, to extend an implicit function to an arbitrary environment, train an implicit environment function to generalize to any environment and/or any goal position, also referred to as a context-aligned implicit function. In at least one embodiment, given a scene context (e.g., a representation of an environment, such as an image or set of points), one or more systems extract scene context features using a fully convolutional environment encoder. In at least one embodiment, one or more systems forward a concatenation of goal coordinates, query position coordinates, and scene context features aligned to said query position into an implicit function, and regress a reaching distance value for said query position. In at least one embodiment, a fully convolutional environment encoder enlarges a receptive field at a query position such that an implicit function has sufficient contextual information to predict a reaching distance to a goal. In at least one embodiment, scene environment feature alignment causes an implicit function to focus on a query position while ignoring other scene context information. In at least one embodiment, one or more systems utilize various neural network models, such as hypernetworks, to encode arbitrary environments and goals.
In at least one embodiment, one or more systems train a context-aligned implicit function using different environments (e.g., mazes), along with reaching distances computed by one or more path planning algorithms, methods, and/or systems as training data. In at least one embodiment, during inference, an implicit environment function predicts or otherwise calculates an environment field for an environment with a given goal position, and is utilized to search for a feasible trajectory from any start position to said goal position.
In at least one embodiment, a training framework 302 is a collection of one or more hardware and/or software computing resources with instructions that, when executed, performs one or more neural network training processes, functions, and/or operations. In at least one embodiment, a training framework 302 trains an implicit environment function 306 using training data 304 to determine an implicit environment function 308. In at least one embodiment, training data 304 is a collection of data that comprises one or more reaching distances for one or more accessible positions in one or more environments to one or more goal positions in said one or more environments. In at least one embodiment, training data 304 is implemented through one or more data structures, such as an array or list, or any suitable collection of data, which encodes reaching distances, positions, goal positions, and/or environments.
In at least one embodiment, training data 304 is generated by one or more path planning algorithms, methods, and/or systems that, based on an input environment (e.g., through an image, set of points, features of an environment, and/or variations thereof) and one or more goal positions, output one or more reaching distances for each accessible position in said environment to said one or more goal positions. In at least one embodiment, training data 304 is generated by one or more systems by inputting an environment and one or more randomly sampled goal positions to one or more path planning algorithms, methods, and/or systems to determine one or more reaching distances for each accessible position in said environment to said one or more randomly sampled goal positions. In at least one embodiment, training data 304 is generated by one or more path planning algorithms, methods, and/or systems based on one or more environments. In at least one embodiment, for example, training data 304 comprises one or more reaching distances for each accessible position of a first environment to one or more goal positions in said first environment, one or more reaching distances for each accessible position of a second environment to one or more goal positions in said second environment, and so on for any number of environments. In at least one embodiment, one or more systems use various neural network models, and/or other appropriate systems, to define accessible and/or inaccessible regions for an environment.
In at least one embodiment, a training framework 302 trains an implicit environment function 306 using training data 304 to represent any suitable environment with any suitable goal position through an environment field. In at least one embodiment, an implicit environment function 306 is any suitable neural network model, algorithm, representation, function, and/or variations thereof. In at least one embodiment, an implicit environment function 306 comprises various neural network models such as a perceptron model, a radial basis network (RBN), an auto encoder (AE), Boltzmann Machine (BM), Restricted Boltzmann Machine (RBM), deep belief network (DBN), deep convolutional network (DCN), extreme learning machine (ELM), deep residual network (DRN), support vector machines (SVM), and/or variations thereof.
In at least one embodiment, a training framework 302 trains an implicit environment function 306 by determining features of each environment used to generate training data 304. In at least one embodiment, a training framework 302 determines or otherwise generates features, also referred to as environment features, scene features, scene context features, and/or variations thereof, from an environment through an encoder (e.g., an encoder 312). In at least one embodiment, an encoder is a neural network that processes an input and outputs a representation of said input, also referred to as features of said input, that indicates various features of said input. In at least one embodiment, features of an input can comprise indications of various details or other characteristics of said input, such as edges, certain patterns, objects, obstacles, accessible/inaccessible regions, other features, and/or variations thereof. In at least one embodiment, features are represented through a feature map, a set of feature vectors, or any suitable representation. In at least one embodiment, an encoder is a convolutional neural network (CNN), recurrent neural network (RNN), and/or variations thereof. In at least one embodiment, an encoder is a convolutional-based encoder. In at least one embodiment, a training framework 302 inputs an environment (e.g., via an image, set of points, or other representation) to an encoder to determine features of said environment, which are represented through a feature map output by said encoder, or any suitable representation, such as one or more feature vectors.
In at least one embodiment, a training framework 302 performs various feature alignment processes on determined features based on a query position. In at least one embodiment, a training framework 302 performs one or more feature alignment processes on features determined from an environment based on a query position that attenuate or otherwise enhance portions of said features corresponding to said query position. In at least one embodiment, a feature map indicates features of each position of an environment, in which a training framework 302 aligns said feature map based on a query position in said environment by attenuating or otherwise enhancing particular aspects of said feature map that correspond to said query position (e.g., features of said query position).
In at least one embodiment, a training framework 302 trains an implicit environment function 306 by inputting a position, a goal position, and features of an environment (e.g., features which may be aligned based on a position) of training data 304 to said implicit environment function 306, causing said implicit environment function 306 to calculate a reaching distance for said position to said goal position, comparing said calculated reaching distance for said position to said goal position to a reaching distance of training data 304, also referred to as a ground truth reaching distance, corresponding to said position, said goal position, and said environment, calculating loss using one or more loss functions based on said comparison, and updating said implicit environment function 306 based on calculated loss. In at least one embodiment, a training framework 302 updates one or more weights, biases, and/or structural connections (e.g., architecture(s) and/or configuration(s) of one or more components) of an implicit environment function 306 such that calculated loss is minimized. In at least one embodiment, a training framework 302 calculates loss using any suitable loss function, such as cross-entropy loss, binary cross-entropy loss, softmax loss, logistic loss, focal loss, and/or variations thereof.
In at least one embodiment, an implicit environment function 306 is trained when calculated loss for said implicit environment function 306 is below a defined threshold, which can be any suitable value. In at least one embodiment, a training framework 302 trains an implicit environment function 306 to represent reaching distances for positions to any suitable goal position for any suitable environment. In at least one embodiment, a training framework 302 trains an implicit environment function 306 to determine an implicit environment function 308, which is said implicit environment function 306 after one or more training processes. In at least one embodiment, an implicit environment function 308 is a trained implicit environment function 306. In at least one embodiment, an implicit environment function 308 processes environments that may or may not have been utilized to generate training data 304.
In at least one embodiment, an implicit environment function 308 represents an environment through an environment field. In at least one embodiment, an implicit environment function 308 represents any suitable environment with any suitable goal position internally through an environment field representation. In at least one embodiment, an implicit environment function 308 determines an environment field representation based on an input goal position (e.g., goal position 318) and features of an environment (e.g., features 314). In at least one embodiment, an implicit environment function 308 represents an environment (e.g., an environment corresponding to features 314) via an environment field through one or more data structures and/or collections of data that encode positions of said environment and reaching distances for said positions to a particular goal position (e.g., input via a goal position 318) in said environment.
In at least one embodiment, an environment field is visualized through environment field visualizations 320A-320B. In at least one embodiment, an environment field visualization is an image in which each position of said environment field visualization corresponds to a position of an environment field. In at least one embodiment, different colors, shades, patterns, and/or variations thereof of an environment field visualization correspond to different reaching distance values. In at least one embodiment, referring to
In at least one embodiment, environment images 310 comprise images, or other representations, such as sets of points, of environments. In at least one embodiment, a representation of an environment is referred to as an environment context and/or a scene context. In at least one embodiment, one or more systems input an environment (e.g., via an image or other representation of environment images 310 representing an environment) to an encoder 312 to determine features of said environment, which are represented through a feature map output by encoder 312, or any suitable representation, such as one or more feature vectors. In at least one embodiment, one or more systems perform various feature alignment processes on determined features based on a query position (e.g., indicated by a query position 316). In at least one embodiment, one or more systems perform one or more feature alignment processes on features determined from an environment (e.g., represented via an image or other representation) based on a query position that attenuate or otherwise enhance portions of said features corresponding to said query position. In at least one embodiment, a feature map indicates features of each position of an environment, in which one or more systems align said feature map based on a query position in said environment by attenuating or otherwise enhancing particular aspects of said feature map that correspond to said query position (e.g., features of said query position).
In at least one embodiment, features 314 comprise features output by an encoder 312 (e.g., a feature map) based on an environment (e.g., an environment represented by a representation of environment images 310). In at least one embodiment, features 314 comprise features output by an encoder 312 based on an environment (e.g., an environment represented by a representation of environment images 310) processed through one or more alignment processes based on a query position 316 (e.g., an aligned feature map).
In at least one embodiment, one or more systems input features 314, a query position 316, and a goal position 318 to an implicit environment function 308 to calculate a reaching distance 322. In at least one embodiment, one or more systems concatenate input features 314, a query position 316, and a goal position 318. In at least one embodiment, a query position 316 is a position in an environment, and is implemented through one or more data objects, types, and/or variations thereof, that encode said position. In at least one embodiment, a query position 316 comprises coordinate data indicating a position. In at least one embodiment, a goal position 318 is a goal position in an environment, and is implemented through one or more data objects, types, and/or variations thereof, that encode said position. In at least one embodiment, a goal position 318 comprises coordinate data indicating a goal position.
In at least one embodiment, an implicit environment function 308 processes features 314, a query position 316, and a goal position 318 to determine a reaching distance 322 (e.g., an environment field value at a query position 316). In at least one embodiment, a reaching distance 322 is a reaching distance value and is implemented through one or more data objects, types, and/or variations thereof, that encode said reaching distance value. In at least one embodiment, a reaching distance 322 comprises a reaching distance value for an environment corresponding to features 314 (e.g., an environment utilized to generate features 314) from a position in said environment indicated by a query position 316 to a particular goal position indicated by a goal position 318 in said environment.
In at least one embodiment, a query position 316 indicates one or more positions in an environment corresponding to features 314 (e.g., an environment utilized to generate features 314), in which a reaching distance 322 comprises a reaching distance value for each position of said one or more positions to a particular goal position indicated by a goal position 318 in said environment. In at least one embodiment, features 314 comprises one or more features each aligned to a particular position of a query position 316. In at least one embodiment, an implicit environment function 308 calculates any number of reaching distance values for any number of query positions to a particular goal position in a single forward pass. In at least one embodiment, one or more systems utilize a determined reaching distance 322 to determine a plurality of paths for various navigation tasks, as described in further detail with regards to
In at least one embodiment, an environment 402 is any suitable environment such as various indoor environments, outdoor environments, maze environments, and/or variations thereof. In at least one embodiment, an environment 402 is a physical environment. In at least one embodiment, an agent 404 is a virtual entity in which an environment 402 is a virtual or otherwise simulated environment. In at least one embodiment, an environment 402 comprises various inaccessible regions, depicted in
In at least one embodiment, a goal 408 indicates a position, also referred to as a location, in an environment 402. In at least one embodiment, a goal 408 is a position in an environment 402 where an agent 404 is to navigate to. In at least one embodiment, one or more systems define a navigation task indicating that an agent 404 is to navigate through an environment 402 to a goal 408. In at least one embodiment, one or more systems input features of an environment 402 to an implicit environment function. In at least one embodiment, one or more systems determine features of an environment 402 by processing a representation of said environment 402. In at least one embodiment, an environment 402 is represented in any suitable manner, such as through an image, set of points, point cloud, set of voxels, and/or variations thereof. In at least one embodiment, an environment is represented through any suitable 2D representation, such as an image, set of points, and/or variations thereof, or any suitable 3D representation, such as an image with depth information, point cloud, set of voxels, and/or variations thereof.
In at least one embodiment, one or more systems determine features of an environment 402 by inputting a representation of said environment 402 to one or more encoders which output said features. In at least one embodiment, one or more systems perform various feature alignment processes on features to determine one or more aligned features, each aligned to a particular position of said environment 402. In at least one embodiment, one or more systems input features of an environment and an indication of a position of a goal 408 to an implicit environment function. In at least one embodiment, one or more systems input an indication of one or more positions of an environment 402, one or more features (e.g., features aligned to said one or more positions), and an indication of a position of a goal 408 to an implicit environment function. In at least one embodiment, an indication of a position refers to any suitable indication of said position, such as coordinate data or other indications.
In at least one embodiment, an implicit environment function outputs a reaching distance value for each position (e.g., each accessible position) in an environment 402 to a goal 408. In at least one embodiment, an implicit environment function outputs a reaching distance value for each position input to said implicit environment function. In at least one embodiment, one or more systems visualize an environment field for an environment 402 using reaching distance values by visualizing each reaching distance value for each position of environment 402 through a particular color scheme, shading scheme, and/or variations thereof. In at least one embodiment, for example, one or more systems visualize an environment field by assigning different shades to particular reaching distance values, such as lighter shading for lower reaching distance values, darker shading for higher reaching distance values, and completely dark shading for inaccessible regions.
In at least one embodiment, one or more systems utilize reaching distance values to guide an agent 404 to a goal 408. In at least one embodiment, one or more systems determine a reaching distance value for each position accessible to an agent 404 from its current position. In at least one embodiment, one or more systems define a step size for an agent 404 that corresponds to a distance of each step of movement an agent 404 can take, in which said one or more systems determine a reaching distance value for each position accessible to an agent 404 through a step, size of said step corresponding to said defined step size, from its current position. In at least one embodiment, a step refers to a step of movement or navigation of an entity. In at least one embodiment, a step size refers to how far an entity travels per step. In at least one embodiment, a step size can be any suitable distance value. In at least one embodiment, one or more systems represent an environment 402 through a grid, in which each grid cell represents a position, and an agent 404 can move to any grid cell immediately accessible to a current grid cell of agent 404. In at least one embodiment, a size of a grid cell can be any suitable value. In at least one embodiment, an agent 404 can move from a center of a grid cell to a center of an adjacent accessible grid cell. In at least one embodiment, one or more systems determine a reaching distance value for each grid cell accessible to an agent 404 from its current grid cell.
In at least one embodiment, one or more systems determine a reaching distance value for each position accessible through a step from a current position of an agent 404. In at least one embodiment, a size of a step of an agent 404 corresponds to a distance travelled through one or more movements of agent 404, and can be any suitable distance value. In at least one embodiment, for example, an agent 404 is an autonomous robot comprising robotic legs, wheels, and/or other hardware for movement, in which, a size of a step corresponds to a distance travelled through one or more steps of said robot legs, distance travelled through one or more rotations of said wheels, and/or variations thereof. In at least one embodiment, one or more systems determine a position with a minimum reaching distance value of a set of positions accessible through a step from a current position of agent 404, determine a path to said position, and cause agent 404 to navigate using said path to said position. In at least one embodiment, one or more systems determine a position with a minimum reaching distance value of a set of positions by comparing each reaching distance value of each position of said set of positions, and determining which position of said set of positions has a minimum reaching distance value. In at least one embodiment, a position is determined with a minimum reaching distance value, a maximum reaching distance value, a reaching distance value under a threshold, or any suitable reaching distance value. In at least one embodiment, one or more systems utilize reaching distance values to calculate a plurality of paths (e.g., paths 406A-406E) through which an entity (e.g., an agent) is to traverse or otherwise navigate.
In at least one embodiment, one or more systems cause an agent to navigate to one or more positions until said agent is within a step size from a goal and can navigate to said goal, in which said one or more systems cause said agent to navigate to said goal. In at least one embodiment, for example, for a navigation task, one or more systems determine a first position with a minimum reaching distance value of a first set of positions accessible through a step from a current position of agent, determine a first path to said first position, and cause said agent to navigate through said first path to said first position, in which said one or more systems then determine a second position with a minimum reaching distance value of a second set of positions accessible through a step from a new current position of agent, determine a second path to said second position, and cause said agent to navigate through said second path to said second position and so on, until said agent is at a position within a step size from a goal position, in which said one or more systems can cause said agent to navigate to said goal position to complete said navigation task.
In at least one embodiment, referring to
In at least one embodiment, one or more systems navigate an agent through an environment through determining one or more reaching distance values for one or more positions accessible from an agent's current position, determining one or more reaching distance values for one or more positions accessible through a step, in which size of said step is pre-defined, from an agent's current position, representing an environment through a grid and determining one or more reaching distance values for one or more grid cells accessible from an agent's current grid cell, and/or variations thereof. In at least one embodiment, one or more systems use an implicit environment function to calculate a plurality of paths, also referred to as a plurality of trajectories and/or a trajectory, for an agent to traverse or otherwise navigate to a goal.
In at least one embodiment, a scene 502A and/or a scene 502B are top-down views of an environment. In at least one embodiment, a scene 502A and/or a scene 502B are top-down views of a 3D environment. In at least one embodiment, one or more systems capture a scene 502A and/or a scene 502B using a depth camera, or other suitable hardware that can capture depth information. In at least one embodiment, images of a scene 502A and/or a scene 502B comprise depth information associated with each pixel of said images.
In at least one embodiment, to model human walking, one or more systems define a floor area as an accessible region for humans in a bird's eye view of a 3D scene. In at least one embodiment, one or more systems determine depth of each pixel in a bird's eye view rendering and label pixels with maximum depth as a floor. In at least one embodiment, one or more systems denote all pixels belonging to a floor area as accessible and other pixels as obstacles, also referred to as inaccessible. In at least one embodiment, one or more systems utilize a bird's eye view image with denoted accessible and inaccessible regions to train an implicit environment function. In at least one embodiment, one or more systems utilize arbitrary goal positions as well as arbitrary environments to train an implicit environment function to determine an environment field of any bird's eye view image.
In at least one embodiment, one or more systems, during inference, given a start and a goal position, search for a feasible trajectory that leads a human to navigate from said start position to said goal while avoiding collision with obstacles in a room. In at least one embodiment, one or more systems define a human step size, which can be a value that approximates an average human step size, and move a human one step towards a direction that yields a largest reaching distance reduction. In at least one embodiment, one or more systems repeat this process until a human reaches a goal. In at least one embodiment, given a searched trajectory in a bird's eye view, one or more systems align an existing human walking sequence onto it by rotating a human pose towards a tangent direction of a trajectory at each time step.
In at least one embodiment, one or more systems utilize an implicit environment function to navigate multiple people in a multiple person environment. In at least one embodiment,
In at least one embodiment, a 3D environment 602 is any suitable environment such as various indoor environments, outdoor environments, maze environments, and/or variations thereof. In at least one embodiment, a 3D environment 602 is a physical environment. In at least one embodiment, a 3D environment 602 is an environment in which an entity is to navigate through to a goal as part of a navigation task.
In at least one embodiment, one or more systems define accessible regions for 3D environments, also referred to as 3D scenes. In at least one embodiment, one or more systems train a generative model to generate an accessible region for humans in a 3D environment. In at least one embodiment, one or more systems learn a distribution of human torso locations in 3D scenes, conditioned on a scene context, through a variational auto-encoder (VAE). In at least one embodiment, one or more systems generate features from a scene point cloud using one or more layers such as a pooling layer of a network such as Point-Net as a scene context feature. In at least one embodiment, one or more systems map a context feature together with human torso location observations to a normal distribution using an encoder. In at least one embodiment, one or more systems utilize a decoder that reconstructs a human torso location given a concatenation of a sampled noise from a normal distribution and a scene context feature. In at least one embodiment, one or more systems train a VAE using a reconstruction objective on human torso locations together with an objective such as a Kullback-Leibler (KL) divergence objective. In at least one embodiment, one or more systems, during inference, sample random noise from a standard normal distribution to generate feasible locations for human torsos to formulate an accessible region in a 3D scene. In at least one embodiment, one or more systems, to further suppress a noisy generation by a VAE model and filter out locations that potentially lead to collisions with furniture in a room, project all generated locations to a bird's eye view and remove locations that fall on furniture or walls. In at least one embodiment, an accessible region visualization 604 is a visualization of generated accessible regions, in which a VAE predicts plausible locations suitable for humans to sit, walk, etc. while avoiding collisions with furniture in a room. In at least one embodiment, a generated accessible region uniformly spreads out in a 3D room and is plausible for all kinds of human actions. In at least one embodiment, for example, points on an obstacle such as a chair or sofa are feasible locations for sitting human torsos, while points in midair are possible locations for walking human torsos.
In at least one embodiment, one or more systems train an implicit environment function to model a full 3D space. In at least one embodiment, one or more systems train an implicit environment function to model a particular 3D environment. In at least one embodiment, one or more systems, for training of an implicit environment function, utilize arbitrary goal positions in a fixed scene environment. In at least one embodiment, an input to an implicit function is a concatenation of goal coordinates and query position coordinates. In at least one embodiment, an output is a reaching distance from a query position to a goal. In at least one embodiment, one or more systems discretize a 3D scene to a voxel grid of any suitable size (e.g., a 64×64×64 voxel grid) and mark all voxel cells within a generated accessible region as accessible and other voxel cells as obstacles, also referred to as inaccessible. In at least one embodiment, one or more systems determine a trajectory to navigate an entity from a start location to a goal by at least querying an implicit function for reaching distances at all possible positions said entity can reach, and navigating said entity to a position with a smallest reaching distance until said entity reaches said goal. In at least one embodiment, one or more systems set a step size as average human step size and determine all possible directions a human can take in a 3D space. In at least one embodiment, an environment field visualization 606 is a visualization of an environment field determined by an implicit environment function trained to model a particular 3D environment. In at least one embodiment, an environment field visualization 606 depicts lighter shading for lower reaching distance values and darker shading for higher reaching distance values. In at least one embodiment, in an environment field, closer a point is to a goal, smaller its reaching distance value is.
In at least one embodiment, a trajectory visualization 608 is a visualization of a trajectory calculated by one or more systems for an entity to navigate to a goal. In at least one embodiment, one or more systems calculate a trajectory comprising a plurality of paths. In at least one embodiment, one or more systems optimize a trajectory based on a given pose sequence and guarantee that a human is navigated towards physically feasible locations; this yields a plausible trajectory aligned with a given human pose sequence. In at least one embodiment, one or more systems, at each step, use a step size computed from adjacent locations in a pose sequence instead of a predefined average step size. In at least one embodiment, one or more systems, for each possible location a human can reach within one step size, check if a human is well supported and a human is not colliding with other objects at these locations and move a human to a location with a smallest reaching distance value while utilizing these constraints and/or other constraints. In at least one embodiment, one or more systems, to determine if a pose is well supported, check if a torso, a left lap, and/or a right lap joints of sitting poses, as well as feet joints of standing poses, have non-positive signed distances to a scene surface. In at least one embodiment, one or more systems, to determine if a pose collides with other objects, check if all joints of a pose, except various support joints, have non-negative signed distances. In at least one embodiment, one or more systems, based on various entity poses, determine a trajectory, visualized in a trajectory visualization 608, for an entity (e.g., a human) to navigate to a goal. In at least one embodiment, a trajectory successfully avoids colliding with obstacles (e.g., furniture) in a room while leading an entity (e.g., a human) towards a goal. In at least one embodiment, entity poses (e.g., human poses) are also plausible at each location on a trajectory.
In at least one embodiment, one or more systems train a conditional VAE and all implicit functions using a solver such as an Adam solver with any suitable learning rate (e.g., a learning rate of 5×10−5). In at least one embodiment, one or more systems, to balance a KL-divergence and reconstruction objectives in a VAE, utilize a schedule such as a Cyclical Annealing Schedule.
In at least one embodiment, one or more systems evaluate an implicit environment function (IEF) on datasets such as a Minigrid and Gridworld maze datasets, which include 2D mazes with randomly placed obstacles. In at least one embodiment, an implicit environment function determines an environment field that accurately encodes a reaching distance from any accessible point to a goal while successfully detecting all obstacles in a maze. In at least one embodiment, one or more systems use a success rate (e.g., a percentage of paths that successfully lead an agent to a goal among all searched paths) as an evaluation metric.
In at least one embodiment, table 1 702 depicts results of an ablation study on network architectures. In at least one embodiment, 11×11 mazes are from a dataset such as a Minigrid dataset while all other mazes are from a dataset such as a Gridworld dataset. In at least one embodiment, “IF” denotes an implicit function. In at least one embodiment, a context-aligned implicit function performs comparably to hypernetworks on smaller-sized mazes (e.g., 8×8 and 11×11 mazes). In at least one embodiment, hypernetworks fail on larger-sized mazes (e.g., 16×16 and 28×28 mazes), in which a context-aligned implicit function is superior over hypernetworks.
In at least one embodiment, one or more systems evaluate an implicit environment function with a network such as a value iteration networks (VIN) on a dataset such as a Minigrid dataset. In at least one embodiment, one or more systems utilize a same train and test split as VIN, and learn different implicit functions for mazes of different sizes. In at least one embodiment, table 2 704 depicts a success rate and average time taken to search a path by different methods. In at least one embodiment, a batch size of 1 is utilized, or any suitable value. In at least one embodiment, an implicit environment function achieves a comparable success rate with much less time in larger-sized mazes compared to VIN. In at least one embodiment, an implicit environment function only needs one network forwarding pass to obtain reaching distance values of all locations in a maze and can guide an agent at any location to reach a goal.
In at least one embodiment, one or more systems utilize an implicit environment function to model long-term dynamic human motion on a dataset such as a Proximal Relationships with Object eXclusion (PROX) dataset. In at least one embodiment, an implicit environment function can predict long-term dynamic human motion across any number of video frames. In at least one embodiment, an implicit environment field represents an environment through an environment field that encodes a reaching distance between any point to a goal position. In at least one embodiment, areas that are closer to a goal have a lower reaching distance value and vice versa. In at least one embodiment, an environment field dynamically changes based on a positions of people in an environment. In at least one embodiment, an inaccessible region corresponds to a region with large reaching distance values (e.g., over a particular threshold). In at least one embodiment, an implicit environment function can model an environment field in a scene and generalize to a dynamically changing environment.
In at least one embodiment, table 3 706 depicts quantitative comparisons with sampling-based path planning methods, such as Rapidly-exploring random trees (RRT) and probabilistic roadmaps (PRM). In at least one embodiment, AFF indicates a pose dependent search. In at least one embodiment, one or more systems evaluate metrics such as a distance between an ending of a searched path and a goal, and an average time cost for a single trajectory search. In at least one embodiment, one or more systems set a number of sampling points and nearest neighbors in PRM to any suitable values, such as 500 and 5, respectively. In at least one embodiment, one or more systems set a maximum iteration for trajectory search in RRT to any suitable value, such as 500 iterations. In at least one embodiment, an implicit environment field is more efficient, as each step can be predicted by a single fast network forward pass.
In at least one embodiment, one or more systems quantitatively compare trajectory predictions against a network such as a PathNet using a distance metric as well as a ratio of valid poses to all poses on a trajectory. In at least one embodiment, one or more systems define a pose as valid if it is both well supported and not collision-free in a scene. In at least one embodiment, one or more systems utilize train and test trajectories such as those of a long-term human motion prediction with scene context (HMP) method, with each trajectory including 30 frames. In at least one embodiment, a PathNet in HMP requires torso locations in a first 10 frames of each trajectory as input and predicts torso locations for following 20 frames. In at least one embodiment, one or more systems predict any number of locations (e.g., 30 locations) that lead a human from a start position in a first frame to a goal in a last frame. In at least one embodiment, table 4 708 depicts quantitative evaluations. In at least one embodiment, an implicit environment function is more effective at navigating humans towards goal positions. In at least one embodiment, after using a pose-dependent trajectory search process, an implicit environment function is able to better fit human poses to a scene, as shown in a last column in table 4 708.
In at least one embodiment, an implicit environment field represents an environment through an environment field that encodes a reaching distance between a pair of points in either 2D or 3D space. In at least one embodiment, a learned environment field is a continuous energy surface that can navigate agents in 2D mazes in dynamically changing scene environments. In at least one embodiment, an environment field is extended to 3D scenes to model dynamic human motion in indoor environments.
In at least one embodiment, to encode different maze environments as well as goal positions, one or more systems utilize networks such as a hypernetwork. In at least one embodiment, a hypernetwork refers a neural network that generates a network for a main network, also referred to as a hyponetwork. In at least one embodiment, for each maze map, one or more systems use a hypernetwork to predict a parameters of a hyponetwork (e.g., an implicit environment function). In at least one embodiment, a hyponetwork obtains as input a concatenation of goal coordinates as well as query position coordinates and outputs a reaching distance between them. In at least one embodiment, a hypernetwork includes any suitable layers, such as 11 convolutional layers followed by 7 fully-connected layers. In at least one embodiment, a hyponetwork includes any suitable layers, such as 7 fully-connected layers, each of which is followed by a periodic activation.
In at least one embodiment, instead of using a reaching distance computed by FMM as supervision to train an implicit environment function, one or more systems utilize a reciprocal of a reaching distance from a feasible point to a goal and normalize it to a range of [0, 1], and train an implicit environment function to predict negative values instead of extremely small values for locations occupied by obstacles. In at least one embodiment, an implicit environment function can differentiate feasible locations and obstacles.
In at least one embodiment, during training of a VAE model, besides points on training human trajectories, one or more systems also augment more points in a scene. In at least one embodiment, one or more systems utilize a random standing human pose and include all locations that can afford this pose (e.g., a pose can be supported by a floor and does not collide with other objects). In at least one embodiment, one or more systems train a conditional VAE model for all training trajectories in various different scenes, while a separate implicit function is trained for each 3D indoor scene. In at least one embodiment, one or more models are trained using a solver such as an Adam solver using a learning rate of 5×10−5. In at least one embodiment, one or more systems implement one or more models using a framework such as a PyTorch framework, or any suitable framework.
In at least one embodiment, one or more systems utilize a given human pose to avoid moving humans to locations that violate physical constraints in a room (e.g., locations that cannot well support a human or lead to collision with other objects). In at least one embodiment, one or more systems, if it is determined that a location can afford a human body mesh, or other suitable representation, such as a skeleton, skeleton mesh, and/or variations thereof, consider vertices belonging to a body part instead of joints. In at least one embodiment, one or more systems utilize a human body mesh, a skeleton, a skeleton mesh, and/or any suitable representation of a human. In at least one embodiment, for example, to check if a sitting human is well supported, one or more systems check if vertices belonging to a gluteus part have non-positive signed distance values to object surfaces, and to check if a human is colliding with other objects, one or more systems check if vertices belonging to legs and thighs have non-negative signed distance values.
In at least one embodiment,
In at least one embodiment, process 1100 is performed by one or more systems such as those described in this present disclosure. In at least one embodiment, one or more systems include any suitable system with a collection of one or more hardware and/or software resources with instructions that, when executed, performs various implicit environment function training operations, implicit environment function processing operations, neural network functions, navigation operations (e.g., causing an entity to navigate to a location), environment processing functions, and/or various other operations such as those described herein. In at least one embodiment, process 1100 is performed by one or more systems associated with an autonomous device.
In at least one embodiment, a system performing at least a part of process 1100 includes executable code to at least obtain 1102 a first location, a set of locations, and a final location. In at least one embodiment, a location is a location in an environment and can be referred to as a position. In at least one embodiment, an environment is any suitable environment, such as a 2D environment, 3D environment, indoor environment, outdoor environment, simulated environment, maze, and/or variations thereof. In at least one embodiment, an environment comprises an autonomous device. In at least one embodiment, an autonomous device is a system such as an autonomous car, autonomous robot, and/or variations thereof. In at least one embodiment, an autonomous device is to navigate from a first location to a final location, also referred to as a goal position or location, in an environment as part of a navigation task. In at least one embodiment, a first location is a location of an autonomous device. In at least one embodiment, a set of locations are one or more locations in an environment. In at least one embodiment, a set of locations include one or more locations in accessible regions of an environment.
In at least one embodiment, a system performing at least a part of process 1100 includes executable code to at least cause 1104 one or more neural networks to calculate a set of distances based at least in part on set of locations and final location. In at least one embodiment, one or more neural networks include an implicit environment function. In at least one embodiment, a system inputs a set of locations and a final location to an implicit environment function. In at least one embodiment, a system inputs features of an environment to an implicit environment function. In at least one embodiment, a system generates features of an environment by inputting a representation of said environment to one or more neural networks, such as an encoder. In at least one embodiment, a representation of an environment is any suitable representation, such as an image, point cloud data, set of points, and/or variations thereof. In at least one embodiment, a representation can be captured using various image capturing hardware, such as a camera, depth camera, sensor device, and/or variations thereof. In at least one embodiment, a system performs various alignment processes on generated features.
In at least one embodiment, one or more neural networks output a set of distances corresponding to a set of locations. In at least one embodiment, a set of distances are reaching distances corresponding to a set of locations. In at least one embodiment, a reaching distance is also referred to as a distance value, distance, reaching distance value, and/or variations thereof. In at least one embodiment, a first distance of a set of distances corresponds to a first location of a set of locations, and indicates a distance of a feasible path from said first location to a final location. In at least one embodiment, a set of locations are relative to a final location. In at least one embodiment, a set of distances for a set of locations are relative to a final location. In at least one embodiment, a feasible path refers to a path that is semantically and/or geometrically feasible. In at least one embodiment, a feasible path refers to any suitable path that an autonomous device can navigate through. In at least one embodiment, an implicit environment function outputs a set of distances in a single forward pass.
In at least one embodiment, a system performing at least a part of process 1100 includes executable code to at least calculate 1106 a plurality of paths based at least in part on set of distances, wherein plurality of paths form a path from first location to final location. In at least one embodiment, a system determines a subset of locations accessible from a first location by an autonomous device. In at least one embodiment, a location is accessible to an autonomous device if said autonomous device can navigate to said location from a current location of said autonomous device. In at least one embodiment, an accessible location is a location that an autonomous device can navigate to within a single step. In at least one embodiment, a system calculates a size for a step of an autonomous device, which corresponds to a distance value of how far said autonomous device travels per step. In at least one embodiment, a system defines a step size as any suitable value.
In at least one embodiment, a system determines a subset of distances of a set of distances corresponding to a subset of locations. In at least one embodiment, a system determines a second location of a subset of locations corresponding to a minimum distance of a subset of distances. In at least one embodiment, a second location corresponds to a minimum distance, maximum distance, or any suitable distance of a subset of distances. In at least one embodiment, a second location corresponds to a location in a particular direction from a first location of an autonomous device. In at least one embodiment, a system calculates a first path comprising a path from a first location to a second location. In at least one embodiment, a system causes an autonomous device to navigate to a second location using a first path.
In at least one embodiment, a system continuously determines paths for an autonomous device until said autonomous device can navigate to a final location. In at least one embodiment, for example, a system determines a second subset of locations accessible from a second location of an autonomous device, obtains a second subset of distances corresponding to said second subset of locations, determines a third location of said second subset of locations (e.g., corresponding to a minimum distance, or any suitable distance), calculates a second path from said second location to said third location, and causes said autonomous device to navigate to said third location using said second path, and so on until a final location is accessible from a current location of said autonomous device, in which said system causes said autonomous device to navigate to said final location.
In at least one embodiment, a system trains one or more neural networks to calculate a plurality of paths, through which an autonomous device is to traverse. In at least one embodiment, a system obtains an environment and a location, causes one or more algorithms to determine one or more reaching distance values for one or more locations in said environment to said location, and trains one or more neural networks at least using said one or more reaching distance values. In at least one embodiment, a system trains one or more neural networks by causing said one or more neural networks to process one or more locations to calculate one or more predicted reaching distance values, and updating said one or more neural networks based on differences between said one or more predicted reaching distance values and one or more reaching distance values output by one or more algorithms. In at least one embodiment, a system updates or otherwise trains one or more neural networks to minimize differences (e.g., update such that differences are under a pre-defined threshold) between predicted reaching distance values and reaching distance values output by one or more algorithms. In at least one embodiment, one or more algorithms include various path planning algorithms, various FMM algorithms, or any suitable algorithm. In at least one embodiment, a system trains one or more neural networks such that trained one or more neural networks can predict reaching distance values that are similar or same as reaching distance values output by one or more algorithms. In at least one embodiment, a system utilizes predicted reaching distance values to determine a plurality of paths through which an autonomous device is to traverse.
In at least one embodiment, one or more processes of process 1100, as well as those described in connection with process 1100, can be performed in any suitable order, including sequential, parallel, and/or variations thereof. In at least one embodiment, process 1100 can include various processes described elsewhere in this disclosure.
In at least one embodiment, inference and/or training logic 1215 may include, without limitation, code and/or data storage 1201 to store forward and/or output weight and/or input/output data, and/or other parameters to configure neurons or layers of a neural network trained and/or used for inferencing in aspects of one or more embodiments. In at least one embodiment, training logic 1215 may include, or be coupled to code and/or data storage 1201 to store graph code or other software to control timing and/or order, in which weight and/or other parameter information is to be loaded to configure, logic, including integer and/or floating point units (collectively, arithmetic logic units (ALUs)). In at least one embodiment, code, such as graph code, loads weight or other parameter information into processor ALUs based on an architecture of a neural network to which such code corresponds. In at least one embodiment, code and/or data storage 1201 stores weight parameters and/or input/output data of each layer of a neural network trained or used in conjunction with one or more embodiments during forward propagation of input/output data and/or weight parameters during training and/or inferencing using aspects of one or more embodiments. In at least one embodiment, any portion of code and/or data storage 1201 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory.
In at least one embodiment, any portion of code and/or data storage 1201 may be internal or external to one or more processors or other hardware logic devices or circuits. In at least one embodiment, code and/or code and/or data storage 1201 may be cache memory, dynamic randomly addressable memory (“DRAM”), static randomly addressable memory (“SRAM”), non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, a choice of whether code and/or code and/or data storage 1201 is internal or external to a processor, for example, or comprising DRAM, SRAM, flash or some other storage type may depend on available storage on-chip versus off-chip, latency requirements of training and/or inferencing functions being performed, batch size of data used in inferencing and/or training of a neural network, or some combination of these factors.
In at least one embodiment, inference and/or training logic 1215 may include, without limitation, a code and/or data storage 1205 to store backward and/or output weight and/or input/output data corresponding to neurons or layers of a neural network trained and/or used for inferencing in aspects of one or more embodiments. In at least one embodiment, code and/or data storage 1205 stores weight parameters and/or input/output data of each layer of a neural network trained or used in conjunction with one or more embodiments during backward propagation of input/output data and/or weight parameters during training and/or inferencing using aspects of one or more embodiments. In at least one embodiment, training logic 1215 may include, or be coupled to code and/or data storage 1205 to store graph code or other software to control timing and/or order, in which weight and/or other parameter information is to be loaded to configure, logic, including integer and/or floating point units (collectively, arithmetic logic units (ALUs)).
In at least one embodiment, code, such as graph code, causes loading of weight or other parameter information into processor ALUs based on an architecture of a neural network to which such code corresponds. In at least one embodiment, any portion of code and/or data storage 1205 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory. In at least one embodiment, any portion of code and/or data storage 1205 may be internal or external to one or more processors or other hardware logic devices or circuits. In at least one embodiment, code and/or data storage 1205 may be cache memory, DRAM, SRAM, non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, a choice of whether code and/or data storage 1205 is internal or external to a processor, for example, or comprising DRAM, SRAM, flash memory or some other storage type may depend on available storage on-chip versus off-chip, latency requirements of training and/or inferencing functions being performed, batch size of data used in inferencing and/or training of a neural network, or some combination of these factors.
In at least one embodiment, code and/or data storage 1201 and code and/or data storage 1205 may be separate storage structures. In at least one embodiment, code and/or data storage 1201 and code and/or data storage 1205 may be a combined storage structure. In at least one embodiment, code and/or data storage 1201 and code and/or data storage 1205 may be partially combined and partially separate. In at least one embodiment, any portion of code and/or data storage 1201 and code and/or data storage 1205 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory.
In at least one embodiment, inference and/or training logic 1215 may include, without limitation, one or more arithmetic logic unit(s) (“ALU(s)”) 1210, including integer and/or floating point units, to perform logical and/or mathematical operations based, at least in part on, or indicated by, training and/or inference code (e.g., graph code), a result of which may produce activations (e.g., output values from layers or neurons within a neural network) stored in an activation storage 1220 that are functions of input/output and/or weight parameter data stored in code and/or data storage 1201 and/or code and/or data storage 1205. In at least one embodiment, activations stored in activation storage 1220 are generated according to linear algebraic and or matrix-based mathematics performed by ALU(s) 1210 in response to performing instructions or other code, wherein weight values stored in code and/or data storage 1205 and/or data storage 1201 are used as operands along with other values, such as bias values, gradient information, momentum values, or other parameters or hyperparameters, any or all of which may be stored in code and/or data storage 1205 or code and/or data storage 1201 or another storage on or off-chip.
In at least one embodiment, ALU(s) 1210 are included within one or more processors or other hardware logic devices or circuits, whereas in another embodiment, ALU(s) 1210 may be external to a processor or other hardware logic device or circuit that uses them (e.g., a co-processor). In at least one embodiment, ALUs 1210 may be included within a processor's execution units or otherwise within a bank of ALUs accessible by a processor's execution units either within same processor or distributed between different processors of different types (e.g., central processing units, graphics processing units, fixed function units, etc.). In at least one embodiment, code and/or data storage 1201, code and/or data storage 1205, and activation storage 1220 may share a processor or other hardware logic device or circuit, whereas in another embodiment, they may be in different processors or other hardware logic devices or circuits, or some combination of same and different processors or other hardware logic devices or circuits. In at least one embodiment, any portion of activation storage 1220 may be included with other on-chip or off-chip data storage, including a processor's L1, L2, or L3 cache or system memory. Furthermore, inferencing and/or training code may be stored with other code accessible to a processor or other hardware logic or circuit and fetched and/or processed using a processor's fetch, decode, scheduling, execution, retirement and/or other logical circuits.
In at least one embodiment, activation storage 1220 may be cache memory, DRAM, SRAM, non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, activation storage 1220 may be completely or partially within or external to one or more processors or other logical circuits. In at least one embodiment, a choice of whether activation storage 1220 is internal or external to a processor, for example, or comprising DRAM, SRAM, flash memory or some other storage type may depend on available storage on-chip versus off-chip, latency requirements of training and/or inferencing functions being performed, batch size of data used in inferencing and/or training of a neural network, or some combination of these factors.
In at least one embodiment, inference and/or training logic 1215 illustrated in
In at least one embodiment, each of code and/or data storage 1201 and 1205 and corresponding computational hardware 1202 and 1206, respectively, correspond to different layers of a neural network, such that resulting activation from one storage/computational pair 1201/1202 of code and/or data storage 1201 and computational hardware 1202 is provided as an input to a next storage/computational pair 1205/1206 of code and/or data storage 1205 and computational hardware 1206, in order to mirror a conceptual organization of a neural network. In at least one embodiment, each of storage/computational pairs 1201/1202 and 1205/1206 may correspond to more than one neural network layer. In at least one embodiment, additional storage/computation pairs (not shown) subsequent to or in parallel with storage/computation pairs 1201/1202 and 1205/1206 may be included in inference and/or training logic 1215.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, untrained neural network 1306 is trained using supervised learning, wherein training dataset 1302 includes an input paired with a desired output for an input, or where training dataset 1302 includes input having a known output and an output of neural network 1306 is manually graded. In at least one embodiment, untrained neural network 1306 is trained in a supervised manner and processes inputs from training dataset 1302 and compares resulting outputs against a set of expected or desired outputs. In at least one embodiment, errors are then propagated back through untrained neural network 1306. In at least one embodiment, training framework 1304 adjusts weights that control untrained neural network 1306. In at least one embodiment, training framework 1304 includes tools to monitor how well untrained neural network 1306 is converging towards a model, such as trained neural network 1308, suitable to generating correct answers, such as in result 1314, based on input data such as a new dataset 1312. In at least one embodiment, training framework 1304 trains untrained neural network 1306 repeatedly while adjust weights to refine an output of untrained neural network 1306 using a loss function and adjustment algorithm, such as stochastic gradient descent. In at least one embodiment, training framework 1304 trains untrained neural network 1306 until untrained neural network 1306 achieves a desired accuracy. In at least one embodiment, trained neural network 1308 can then be deployed to implement any number of machine learning operations.
In at least one embodiment, untrained neural network 1306 is trained using unsupervised learning, wherein untrained neural network 1306 attempts to train itself using unlabeled data. In at least one embodiment, unsupervised learning training dataset 1302 will include input data without any associated output data or “ground truth” data. In at least one embodiment, untrained neural network 1306 can learn groupings within training dataset 1302 and can determine how individual inputs are related to untrained dataset 1302. In at least one embodiment, unsupervised training can be used to generate a self-organizing map in trained neural network 1308 capable of performing operations useful in reducing dimensionality of new dataset 1312. In at least one embodiment, unsupervised training can also be used to perform anomaly detection, which allows identification of data points in new dataset 1312 that deviate from normal patterns of new dataset 1312.
In at least one embodiment, semi-supervised learning may be used, which is a technique in which in training dataset 1302 includes a mix of labeled and unlabeled data. In at least one embodiment, training framework 1304 may be used to perform incremental learning, such as through transferred learning techniques. In at least one embodiment, incremental learning enables trained neural network 1308 to adapt to new dataset 1312 without forgetting knowledge instilled within trained neural network 1308 during initial training.
In at least one embodiment, training framework 1304 is a framework processed in connection with a software development toolkit such as an OpenVINO (Open Visual Inference and Neural network Optimization) toolkit. In at least one embodiment, an OpenVINO toolkit is a toolkit such as those developed by Intel Corporation of Santa Clara, Calif.
In at least one embodiment, OpenVINO is a toolkit for facilitating development of applications, specifically neural network applications, for various tasks and operations, such as human vision emulation, speech recognition, natural language processing, recommendation systems, and/or variations thereof. In at least one embodiment, OpenVINO supports neural networks such as convolutional neural networks (CNNs), recurrent and/or attention-based neural networks, and/or various other neural network models. In at least one embodiment, OpenVINO supports various software libraries such as OpenCV, OpenCL, and/or variations thereof.
In at least one embodiment, OpenVINO supports neural network models for various tasks and operations, such as classification, segmentation, object detection, face recognition, speech recognition, pose estimation (e.g., humans and/or objects), monocular depth estimation, image inpainting, style transfer, action recognition, colorization, and/or variations thereof.
In at least one embodiment, OpenVINO comprises one or more software tools and/or modules for model optimization, also referred to as a model optimizer. In at least one embodiment, a model optimizer is a command line tool that facilitates transitions between training and deployment of neural network models. In at least one embodiment, a model optimizer optimizes neural network models for execution on various devices and/or processing units, such as a GPU, CPU, PPU, GPGPU, and/or variations thereof. In at least one embodiment, a model optimizer generates an internal representation of a model, and optimizes said model to generate an intermediate representation. In at least one embodiment, a model optimizer reduces a number of layers of a model. In at least one embodiment, a model optimizer removes layers of a model that are utilized for training. In at least one embodiment, a model optimizer performs various neural network operations, such as modifying inputs to a model (e.g., resizing inputs to a model), modifying a size of inputs of a model (e.g., modifying a batch size of a model), modifying a model structure (e.g., modifying layers of a model), normalization, standardization, quantization (e.g., converting weights of a model from a first representation, such as floating point, to a second representation, such as integer), and/or variations thereof.
In at least one embodiment, OpenVINO comprises one or more software libraries for inferencing, also referred to as an inference engine. In at least one embodiment, an inference engine is a C++ library, or any suitable programming language library. In at least one embodiment, an inference engine is utilized to infer input data. In at least one embodiment, an inference engine implements various classes to infer input data and generate one or more results. In at least one embodiment, an inference engine implements one or more API functions to process an intermediate representation, set input and/or output formats, and/or execute a model on one or more devices.
In at least one embodiment, OpenVINO provides various abilities for heterogeneous execution of one or more neural network models. In at least one embodiment, heterogeneous execution, or heterogeneous computing, refers to one or more computing processes and/or systems that utilize one or more types of processors and/or cores. In at least one embodiment, OpenVINO provides various software functions to execute a program on one or more devices. In at least one embodiment, OpenVINO provides various software functions to execute a program and/or portions of a program on different devices. In at least one embodiment, OpenVINO provides various software functions to, for example, run a first portion of code on a CPU and a second portion of code on a GPU and/or FPGA. In at least one embodiment, OpenVINO provides various software functions to execute one or more layers of a neural network on one or more devices (e.g., a first set of layers on a first device, such as a GPU, and a second set of layers on a second device, such as a CPU).
In at least one embodiment, OpenVINO includes various functionality similar to functionalities associated with a CUDA programming model, such as various neural network model operations associated with frameworks such as TensorFlow, PyTorch, and/or variations thereof. In at least one embodiment, one or more CUDA programming model operations are performed using OpenVINO. In at least one embodiment, various systems, methods, and/or techniques described herein are implemented using OpenVINO.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, as shown in
In at least one embodiment, grouped computing resources 1414 may include separate groupings of node C.R.s housed within one or more racks (not shown), or many racks housed in data centers at various geographical locations (also not shown). In at least one embodiment, separate groupings of node C.R.s within grouped computing resources 1414 may include grouped compute, network, memory or storage resources that may be configured or allocated to support one or more workloads. In at least one embodiment, several node C.R.s including CPUs or processors may grouped within one or more racks to provide compute resources to support one or more workloads. In at least one embodiment, one or more racks may also include any number of power modules, cooling modules, and network switches, in any combination.
In at least one embodiment, resource orchestrator 1412 may configure or otherwise control one or more node C.R.s 1416(1)-1416(N) and/or grouped computing resources 1414. In at least one embodiment, resource orchestrator 1412 may include a software design infrastructure (“SDI”) management entity for data center 1400. In at least one embodiment, resource orchestrator 1212 may include hardware, software or some combination thereof.
In at least one embodiment, as shown in
In at least one embodiment, software 1432 included in software layer 1430 may include software used by at least portions of node C.R.s 1416(1)-1416(N), grouped computing resources 1414, and/or distributed file system 1428 of framework layer 1420. In at least one embodiment, one or more types of software may include, but are not limited to, Internet web page search software, e-mail virus scan software, database software, and streaming video content software.
In at least one embodiment, application(s) 1442 included in application layer 1440 may include one or more types of applications used by at least portions of node C.R.s 1416(1)-1416(N), grouped computing resources 1414, and/or distributed file system 1428 of framework layer 1420. In at least one embodiment, one or more types of applications may include, but are not limited to, any number of a genomics application, a cognitive compute, application and a machine learning application, including training or inferencing software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.) or other machine learning applications used in conjunction with one or more embodiments.
In at least one embodiment, any of configuration manager 1424, resource manager 1426, and resource orchestrator 1412 may implement any number and type of self-modifying actions based on any amount and type of data acquired in any technically feasible fashion. In at least one embodiment, self-modifying actions may relieve a data center operator of data center 1400 from making possibly bad configuration decisions and possibly avoiding underutilized and/or poor performing portions of a data center.
In at least one embodiment, data center 1400 may include tools, services, software or other resources to train one or more machine learning models or predict or infer information using one or more machine learning models according to one or more embodiments described herein. For example, in at least one embodiment, a machine learning model may be trained by calculating weight parameters according to a neural network architecture using software and computing resources described above with respect to data center 1400. In at least one embodiment, trained machine learning models corresponding to one or more neural networks may be used to infer or predict information using resources described above with respect to data center 1400 by using weight parameters calculated through one or more training techniques described herein.
In at least one embodiment, data center may use CPUs, application-specific integrated circuits (ASICs), GPUs, FPGAs, or other hardware to perform training and/or inferencing using above-described resources. Moreover, one or more software and/or hardware resources described above may be configured as a service to allow users to train or performing inferencing of information, such as image recognition, speech recognition, or other artificial intelligence services.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
Autonomous vehicles may be described in terms of automation levels, defined by National Highway Traffic Safety Administration (“NHTSA”), a division of US Department of Transportation, and Society of Automotive Engineers (“SAE”) “Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles” (e.g., Standard No. J3016-201806, published on Jun. 15, 2018, Standard No. J3016-201609, published on Sep. 30, 2016, and previous and future versions of this standard). In at least one embodiment, vehicle 1500 may be capable of functionality in accordance with one or more of Level 1 through Level 5 of autonomous driving levels. For example, in at least one embodiment, vehicle 1500 may be capable of conditional automation (Level 3), high automation (Level 4), and/or full automation (Level 5), depending on embodiment.
In at least one embodiment, vehicle 1500 may include, without limitation, components such as a chassis, a vehicle body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other components of a vehicle. In at least one embodiment, vehicle 1500 may include, without limitation, a propulsion system 1550, such as an internal combustion engine, hybrid electric power plant, an all-electric engine, and/or another propulsion system type. In at least one embodiment, propulsion system 1550 may be connected to a drive train of vehicle 1500, which may include, without limitation, a transmission, to enable propulsion of vehicle 1500. In at least one embodiment, propulsion system 1550 may be controlled in response to receiving signals from a throttle/accelerator(s) 1552.
In at least one embodiment, a steering system 1554, which may include, without limitation, a steering wheel, is used to steer vehicle 1500 (e.g., along a desired path or route) when propulsion system 1550 is operating (e.g., when vehicle 1500 is in motion). In at least one embodiment, steering system 1554 may receive signals from steering actuator(s) 1556. In at least one embodiment, a steering wheel may be optional for full automation (Level 5) functionality. In at least one embodiment, a brake sensor system 1546 may be used to operate vehicle brakes in response to receiving signals from brake actuator(s) 1548 and/or brake sensors.
In at least one embodiment, controller(s) 1536, which may include, without limitation, one or more system on chips (“SoCs”) (not shown in
In at least one embodiment, controller(s) 1536 provide signals for controlling one or more components and/or systems of vehicle 1500 in response to sensor data received from one or more sensors (e.g., sensor inputs). In at least one embodiment, sensor data may be received from, for example and without limitation, global navigation satellite systems (“GNSS”) sensor(s) 1558 (e.g., Global Positioning System sensor(s)), RADAR sensor(s) 1560, ultrasonic sensor(s) 1562, LIDAR sensor(s) 1564, inertial measurement unit (“IMU”) sensor(s) 1566 (e.g., accelerometer(s), gyroscope(s), a magnetic compass or magnetic compasses, magnetometer(s), etc.), microphone(s) 1596, stereo camera(s) 1568, wide-view camera(s) 1570 (e.g., fisheye cameras), infrared camera(s) 1572, surround camera(s) 1574 (e.g., 360 degree cameras), long-range cameras (not shown in
In at least one embodiment, one or more of controller(s) 1536 may receive inputs (e.g., represented by input data) from an instrument cluster 1532 of vehicle 1500 and provide outputs (e.g., represented by output data, display data, etc.) via a human-machine interface (“HMI”) display 1534, an audible annunciator, a loudspeaker, and/or via other components of vehicle 1500. In at least one embodiment, outputs may include information such as vehicle velocity, speed, time, map data (e.g., a High Definition map (not shown in
In at least one embodiment, vehicle 1500 further includes a network interface 1524 which may use wireless antenna(s) 1526 and/or modem(s) to communicate over one or more networks. For example, in at least one embodiment, network interface 1524 may be capable of communication over Long-Term Evolution (“LTE”), Wideband Code Division Multiple Access (“WCDMA”), Universal Mobile Telecommunications System (“UMTS”), Global System for Mobile communication (“GSM”), IMT-CDMA Multi-Carrier (“CDMA2000”) networks, etc. In at least one embodiment, wireless antenna(s) 1526 may also enable communication between objects in environment (e.g., vehicles, mobile devices, etc.), using local area network(s), such as Bluetooth, Bluetooth Low Energy (“LE”), Z-Wave, ZigBee, etc., and/or low power wide-area network(s) (“LPWANs”), such as LoRaWAN, SigFox, etc. protocols.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, camera types for cameras may include, but are not limited to, digital cameras that may be adapted for use with components and/or systems of vehicle 1500. In at least one embodiment, camera(s) may operate at automotive safety integrity level (“ASIL”) B and/or at another ASIL. In at least one embodiment, camera types may be capable of any image capture rate, such as 60 frames per second (fps), 1220 fps, 240 fps, etc., depending on embodiment. In at least one embodiment, cameras may be capable of using rolling shutters, global shutters, another type of shutter, or a combination thereof. In at least one embodiment, color filter array may include a red clear clear clear (“RCCC”) color filter array, a red clear clear blue (“RCCB”) color filter array, a red blue green clear (“RBGC”) color filter array, a Foveon X3 color filter array, a Bayer sensors (“RGGB”) color filter array, a monochrome sensor color filter array, and/or another type of color filter array. In at least one embodiment, clear pixel cameras, such as cameras with an RCCC, an RCCB, and/or an RBGC color filter array, may be used in an effort to increase light sensitivity.
In at least one embodiment, one or more of camera(s) may be used to perform advanced driver assistance systems (“ADAS”) functions (e.g., as part of a redundant or fail-safe design). For example, in at least one embodiment, a Multi-Function Mono Camera may be installed to provide functions including lane departure warning, traffic sign assist and intelligent headlamp control. In at least one embodiment, one or more of camera(s) (e.g., all cameras) may record and provide image data (e.g., video) simultaneously.
In at least one embodiment, one or more camera may be mounted in a mounting assembly, such as a custom designed (three-dimensional (“3D”) printed) assembly, in order to cut out stray light and reflections from within vehicle 1500 (e.g., reflections from dashboard reflected in windshield mirrors) which may interfere with camera image data capture abilities. With reference to wing-mirror mounting assemblies, in at least one embodiment, wing-mirror assemblies may be custom 3D printed so that a camera mounting plate matches a shape of a wing-mirror. In at least one embodiment, camera(s) may be integrated into wing-mirrors. In at least one embodiment, for side-view cameras, camera(s) may also be integrated within four pillars at each corner of a cabin.
In at least one embodiment, cameras with a field of view that include portions of an environment in front of vehicle 1500 (e.g., front-facing cameras) may be used for surround view, to help identify forward facing paths and obstacles, as well as aid in, with help of one or more of controller(s) 1536 and/or control SoCs, providing information critical to generating an occupancy grid and/or determining preferred vehicle paths. In at least one embodiment, front-facing cameras may be used to perform many similar ADAS functions as LIDAR, including, without limitation, emergency braking, pedestrian detection, and collision avoidance. In at least one embodiment, front-facing cameras may also be used for ADAS functions and systems including, without limitation, Lane Departure Warnings (“LDW”), Autonomous Cruise Control (“ACC”), and/or other functions such as traffic sign recognition.
In at least one embodiment, a variety of cameras may be used in a front-facing configuration, including, for example, a monocular camera platform that includes a CMOS (“complementary metal oxide semiconductor”) color imager. In at least one embodiment, a wide-view camera 1570 may be used to perceive objects coming into view from a periphery (e.g., pedestrians, crossing traffic or bicycles). Although only one wide-view camera 1570 is illustrated in
In at least one embodiment, any number of stereo camera(s) 1568 may also be included in a front-facing configuration. In at least one embodiment, one or more of stereo camera(s) 1568 may include an integrated control unit comprising a scalable processing unit, which may provide a programmable logic (“FPGA”) and a multi-core micro-processor with an integrated Controller Area Network (“CAN”) or Ethernet interface on a single chip. In at least one embodiment, such a unit may be used to generate a 3D map of an environment of vehicle 1500, including a distance estimate for all points in an image. In at least one embodiment, one or more of stereo camera(s) 1568 may include, without limitation, compact stereo vision sensor(s) that may include, without limitation, two camera lenses (one each on left and right) and an image processing chip that may measure distance from vehicle 1500 to target object and use generated information (e.g., metadata) to activate autonomous emergency braking and lane departure warning functions. In at least one embodiment, other types of stereo camera(s) 1568 may be used in addition to, or alternatively from, those described herein.
In at least one embodiment, cameras with a field of view that include portions of environment to sides of vehicle 1500 (e.g., side-view cameras) may be used for surround view, providing information used to create and update an occupancy grid, as well as to generate side impact collision warnings. For example, in at least one embodiment, surround camera(s) 1574 (e.g., four surround cameras as illustrated in
In at least one embodiment, cameras with a field of view that include portions of an environment behind vehicle 1500 (e.g., rear-view cameras) may be used for parking assistance, surround view, rear collision warnings, and creating and updating an occupancy grid. In at least one embodiment, a wide variety of cameras may be used including, but not limited to, cameras that are also suitable as a front-facing camera(s) (e.g., long-range cameras 1598 and/or mid-range camera(s) 1576, stereo camera(s) 1568, infrared camera(s) 1572, etc.,) as described herein.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, in addition to, or alternatively from CAN, FlexRay and/or Ethernet protocols may be used. In at least one embodiment, there may be any number of busses forming bus 1502, which may include, without limitation, zero or more CAN busses, zero or more FlexRay busses, zero or more Ethernet busses, and/or zero or more other types of busses using different protocols. In at least one embodiment, two or more busses may be used to perform different functions, and/or may be used for redundancy. For example, a first bus may be used for collision avoidance functionality and a second bus may be used for actuation control. In at least one embodiment, each bus of bus 1502 may communicate with any of components of vehicle 1500, and two or more busses of bus 1502 may communicate with corresponding components. In at least one embodiment, each of any number of system(s) on chip(s) (“SoC(s)”) 1504 (such as SoC 1504(A) and SoC 1504(B)), each of controller(s) 1536, and/or each computer within vehicle may have access to same input data (e.g., inputs from sensors of vehicle 1500), and may be connected to a common bus, such CAN bus.
In at least one embodiment, vehicle 1500 may include one or more controller(s) 1536, such as those described herein with respect to
In at least one embodiment, vehicle 1500 may include any number of SoCs 1504. In at least one embodiment, each of SoCs 1504 may include, without limitation, central processing units (“CPU(s)”) 1506, graphics processing units (“GPU(s)”) 1508, processor(s) 1510, cache(s) 1512, accelerator(s) 1514, data store(s) 1516, and/or other components and features not illustrated. In at least one embodiment, SoC(s) 1504 may be used to control vehicle 1500 in a variety of platforms and systems. For example, in at least one embodiment, SoC(s) 1504 may be combined in a system (e.g., system of vehicle 1500) with a High Definition (“HD”) map 1522 which may obtain map refreshes and/or updates via network interface 1524 from one or more servers (not shown in
In at least one embodiment, CPU(s) 1506 may include a CPU cluster or CPU complex (alternatively referred to herein as a “CCPLEX”). In at least one embodiment, CPU(s) 1506 may include multiple cores and/or level two (“L2”) caches. For instance, in at least one embodiment, CPU(s) 1506 may include eight cores in a coherent multi-processor configuration. In at least one embodiment, CPU(s) 1506 may include four dual-core clusters where each cluster has a dedicated L2 cache (e.g., a 2 megabyte (MB) L2 cache). In at least one embodiment, CPU(s) 1506 (e.g., CCPLEX) may be configured to support simultaneous cluster operations enabling any combination of clusters of CPU(s) 1506 to be active at any given time.
In at least one embodiment, one or more of CPU(s) 1506 may implement power management capabilities that include, without limitation, one or more of following features: individual hardware blocks may be clock-gated automatically when idle to save dynamic power; each core clock may be gated when such core is not actively executing instructions due to execution of Wait for Interrupt (“WFI”)/Wait for Event (“WFE”) instructions; each core may be independently power-gated; each core cluster may be independently clock-gated when all cores are clock-gated or power-gated; and/or each core cluster may be independently power-gated when all cores are power-gated. In at least one embodiment, CPU(s) 1506 may further implement an enhanced algorithm for managing power states, where allowed power states and expected wakeup times are specified, and hardware/microcode determines which best power state to enter for core, cluster, and CCPLEX. In at least one embodiment, processing cores may support simplified power state entry sequences in software with work offloaded to microcode.
In at least one embodiment, GPU(s) 1508 may include an integrated GPU (alternatively referred to herein as an “iGPU”). In at least one embodiment, GPU(s) 1508 may be programmable and may be efficient for parallel workloads. In at least one embodiment, GPU(s) 1508 may use an enhanced tensor instruction set. In at least one embodiment, GPU(s) 1508 may include one or more streaming microprocessors, where each streaming microprocessor may include a level one (“L1”) cache (e.g., an L1 cache with at least 96 KB storage capacity), and two or more streaming microprocessors may share an L2 cache (e.g., an L2 cache with a 512 KB storage capacity). In at least one embodiment, GPU(s) 1508 may include at least eight streaming microprocessors. In at least one embodiment, GPU(s) 1508 may use compute application programming interface(s) (API(s)). In at least one embodiment, GPU(s) 1508 may use one or more parallel computing platforms and/or programming models (e.g., NVIDIA's CUDA model).
In at least one embodiment, one or more of GPU(s) 1508 may be power-optimized for best performance in automotive and embedded use cases. For example, in at least one embodiment, GPU(s) 1508 could be fabricated on Fin field-effect transistor (“FinFET”) circuitry. In at least one embodiment, each streaming microprocessor may incorporate a number of mixed-precision processing cores partitioned into multiple blocks. For example, and without limitation, 64 PF32 cores and 32 PF64 cores could be partitioned into four processing blocks. In at least one embodiment, each processing block could be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed-precision NVIDIA Tensor cores for deep learning matrix arithmetic, a level zero (“L0”) instruction cache, a warp scheduler, a dispatch unit, and/or a 64 KB register file. In at least one embodiment, streaming microprocessors may include independent parallel integer and floating-point data paths to provide for efficient execution of workloads with a mix of computation and addressing calculations. In at least one embodiment, streaming microprocessors may include independent thread scheduling capability to enable finer-grain synchronization and cooperation between parallel threads. In at least one embodiment, streaming microprocessors may include a combined L1 data cache and shared memory unit in order to improve performance while simplifying programming.
In at least one embodiment, one or more of GPU(s) 1508 may include a high bandwidth memory (“HBM”) and/or a 16 GB HBM2 memory subsystem to provide, in some examples, about 900 GB/second peak memory bandwidth. In at least one embodiment, in addition to, or alternatively from, HBM memory, a synchronous graphics random-access memory (“SGRAM”) may be used, such as a graphics double data rate type five synchronous random-access memory (“GDDR5”).
In at least one embodiment, GPU(s) 1508 may include unified memory technology. In at least one embodiment, address translation services (“ATS”) support may be used to allow GPU(s) 1508 to access CPU(s) 1506 page tables directly. In at least one embodiment, embodiment, when a GPU of GPU(s) 1508 memory management unit (“MMU”) experiences a miss, an address translation request may be transmitted to CPU(s) 1506. In response, 2 CPU of CPU(s) 1506 may look in its page tables for a virtual-to-physical mapping for an address and transmit translation back to GPU(s) 1508, in at least one embodiment. In at least one embodiment, unified memory technology may allow a single unified virtual address space for memory of both CPU(s) 1506 and GPU(s) 1508, thereby simplifying GPU(s) 1508 programming and porting of applications to GPU(s) 1508.
In at least one embodiment, GPU(s) 1508 may include any number of access counters that may keep track of frequency of access of GPU(s) 1508 to memory of other processors. In at least one embodiment, access counter(s) may help ensure that memory pages are moved to physical memory of a processor that is accessing pages most frequently, thereby improving efficiency for memory ranges shared between processors.
In at least one embodiment, one or more of SoC(s) 1504 may include any number of cache(s) 1512, including those described herein. For example, in at least one embodiment, cache(s) 1512 could include a level three (“L3”) cache that is available to both CPU(s) 1506 and GPU(s) 1508 (e.g., that is connected to CPU(s) 1506 and GPU(s) 1508). In at least one embodiment, cache(s) 1512 may include a write-back cache that may keep track of states of lines, such as by using a cache coherence protocol (e.g., MEI, MESI, MSI, etc.). In at least one embodiment, a L3 cache may include 4 MB of memory or more, depending on embodiment, although smaller cache sizes may be used.
In at least one embodiment, one or more of SoC(s) 1504 may include one or more accelerator(s) 1514 (e.g., hardware accelerators, software accelerators, or a combination thereof). In at least one embodiment, SoC(s) 1504 may include a hardware acceleration cluster that may include optimized hardware accelerators and/or large on-chip memory. In at least one embodiment, large on-chip memory (e.g., 4 MB of SRAM), may enable a hardware acceleration cluster to accelerate neural networks and other calculations. In at least one embodiment, a hardware acceleration cluster may be used to complement GPU(s) 1508 and to off-load some of tasks of GPU(s) 1508 (e.g., to free up more cycles of GPU(s) 1508 for performing other tasks). In at least one embodiment, accelerator(s) 1514 could be used for targeted workloads (e.g., perception, convolutional neural networks (“CNNs”), recurrent neural networks (“RNNs”), etc.) that are stable enough to be amenable to acceleration. In at least one embodiment, a CNN may include a region-based or regional convolutional neural networks (“RCNNs”) and Fast RCNNs (e.g., as used for object detection) or other type of CNN.
In at least one embodiment, accelerator(s) 1514 (e.g., hardware acceleration cluster) may include one or more deep learning accelerator (“DLA”). In at least one embodiment, DLA(s) may include, without limitation, one or more Tensor processing units (“TPUs”) that may be configured to provide an additional ten trillion operations per second for deep learning applications and inferencing. In at least one embodiment, TPUs may be accelerators configured to, and optimized for, performing image processing functions (e.g., for CNNs, RCNNs, etc.). In at least one embodiment, DLA(s) may further be optimized for a specific set of neural network types and floating point operations, as well as inferencing. In at least one embodiment, design of DLA(s) may provide more performance per millimeter than a typical general-purpose GPU, and typically vastly exceeds performance of a CPU. In at least one embodiment, TPU(s) may perform several functions, including a single-instance convolution function, supporting, for example, INT8, INT16, and FP16 data types for both features and weights, as well as post-processor functions. In at least one embodiment, DLA(s) may quickly and efficiently execute neural networks, especially CNNs, on processed or unprocessed data for any of a variety of functions, including, for example and without limitation: a CNN for object identification and detection using data from camera sensors; a CNN for distance estimation using data from camera sensors; a CNN for emergency vehicle detection and identification and detection using data from microphones; a CNN for facial recognition and vehicle owner identification using data from camera sensors; and/or a CNN for security and/or safety related events.
In at least one embodiment, DLA(s) may perform any function of GPU(s) 1508, and by using an inference accelerator, for example, a designer may target either DLA(s) or GPU(s) 1508 for any function. For example, in at least one embodiment, a designer may focus processing of CNNs and floating point operations on DLA(s) and leave other functions to GPU(s) 1508 and/or accelerator(s) 1514.
In at least one embodiment, accelerator(s) 1514 may include programmable vision accelerator (“PVA”), which may alternatively be referred to herein as a computer vision accelerator. In at least one embodiment, PVA may be designed and configured to accelerate computer vision algorithms for advanced driver assistance system (“ADAS”) 1538, autonomous driving, augmented reality (“AR”) applications, and/or virtual reality (“VR”) applications. In at least one embodiment, PVA may provide a balance between performance and flexibility. For example, in at least one embodiment, each PVA may include, for example and without limitation, any number of reduced instruction set computer (“RISC”) cores, direct memory access (“DMA”), and/or any number of vector processors.
In at least one embodiment, RISC cores may interact with image sensors (e.g., image sensors of any cameras described herein), image signal processor(s), etc. In at least one embodiment, each RISC core may include any amount of memory. In at least one embodiment, RISC cores may use any of a number of protocols, depending on embodiment. In at least one embodiment, RISC cores may execute a real-time operating system (“RTOS”). In at least one embodiment, RISC cores may be implemented using one or more integrated circuit devices, application specific integrated circuits (“ASICs”), and/or memory devices. For example, in at least one embodiment, RISC cores could include an instruction cache and/or a tightly coupled RAM.
In at least one embodiment, DMA may enable components of PVA to access system memory independently of CPU(s) 1506. In at least one embodiment, DMA may support any number of features used to provide optimization to a PVA including, but not limited to, supporting multi-dimensional addressing and/or circular addressing. In at least one embodiment, DMA may support up to six or more dimensions of addressing, which may include, without limitation, block width, block height, block depth, horizontal block stepping, vertical block stepping, and/or depth stepping.
In at least one embodiment, vector processors may be programmable processors that may be designed to efficiently and flexibly execute programming for computer vision algorithms and provide signal processing capabilities. In at least one embodiment, a PVA may include a PVA core and two vector processing subsystem partitions. In at least one embodiment, a PVA core may include a processor subsystem, DMA engine(s) (e.g., two DMA engines), and/or other peripherals. In at least one embodiment, a vector processing subsystem may operate as a primary processing engine of a PVA, and may include a vector processing unit (“VPU”), an instruction cache, and/or vector memory (e.g., “VMEM”). In at least one embodiment, VPU core may include a digital signal processor such as, for example, a single instruction, multiple data (“SIMD”), very long instruction word (“VLIW”) digital signal processor. In at least one embodiment, a combination of SIMD and VLIW may enhance throughput and speed.
In at least one embodiment, each of vector processors may include an instruction cache and may be coupled to dedicated memory. As a result, in at least one embodiment, each of vector processors may be configured to execute independently of other vector processors. In at least one embodiment, vector processors that are included in a particular PVA may be configured to employ data parallelism. For instance, in at least one embodiment, plurality of vector processors included in a single PVA may execute a common computer vision algorithm, but on different regions of an image. In at least one embodiment, vector processors included in a particular PVA may simultaneously execute different computer vision algorithms, on one image, or even execute different algorithms on sequential images or portions of an image. In at least one embodiment, among other things, any number of PVAs may be included in hardware acceleration cluster and any number of vector processors may be included in each PVA. In at least one embodiment, PVA may include additional error correcting code (“ECC”) memory, to enhance overall system safety.
In at least one embodiment, accelerator(s) 1514 may include a computer vision network on-chip and static random-access memory (“SRAM”), for providing a high-bandwidth, low latency SRAM for accelerator(s) 1514. In at least one embodiment, on-chip memory may include at least 4 MB SRAM, comprising, for example and without limitation, eight field-configurable memory blocks, that may be accessible by both a PVA and a DLA. In at least one embodiment, each pair of memory blocks may include an advanced peripheral bus (“APB”) interface, configuration circuitry, a controller, and a multiplexer. In at least one embodiment, any type of memory may be used. In at least one embodiment, a PVA and a DLA may access memory via a backbone that provides a PVA and a DLA with high-speed access to memory. In at least one embodiment, a backbone may include a computer vision network on-chip that interconnects a PVA and a DLA to memory (e.g., using APB).
In at least one embodiment, a computer vision network on-chip may include an interface that determines, before transmission of any control signal/address/data, that both a PVA and a DLA provide ready and valid signals. In at least one embodiment, an interface may provide for separate phases and separate channels for transmitting control signals/addresses/data, as well as burst-type communications for continuous data transfer. In at least one embodiment, an interface may comply with International Organization for Standardization (“ISO”) 26262 or International Electrotechnical Commission (“IEC”) 61508 standards, although other standards and protocols may be used.
In at least one embodiment, one or more of SoC(s) 1504 may include a real-time ray-tracing hardware accelerator. In at least one embodiment, real-time ray-tracing hardware accelerator may be used to quickly and efficiently determine positions and extents of objects (e.g., within a world model), to generate real-time visualization simulations, for RADAR signal interpretation, for sound propagation synthesis and/or analysis, for simulation of SONAR systems, for general wave propagation simulation, for comparison to LIDAR data for purposes of localization and/or other functions, and/or for other uses.
In at least one embodiment, accelerator(s) 1514 can have a wide array of uses for autonomous driving. In at least one embodiment, a PVA may be used for key processing stages in ADAS and autonomous vehicles. In at least one embodiment, a PVA's capabilities are a good match for algorithmic domains needing predictable processing, at low power and low latency. In other words, a PVA performs well on semi-dense or dense regular computation, even on small data sets, which might require predictable run-times with low latency and low power. In at least one embodiment, such as in vehicle 1500, PVAs might be designed to run classic computer vision algorithms, as they can be efficient at object detection and operating on integer math.
For example, according to at least one embodiment of technology, a PVA is used to perform computer stereo vision. In at least one embodiment, a semi-global matching-based algorithm may be used in some examples, although this is not intended to be limiting. In at least one embodiment, applications for Level 3-5 autonomous driving use motion estimation/stereo matching on-the-fly (e.g., structure from motion, pedestrian recognition, lane detection, etc.). In at least one embodiment, a PVA may perform computer stereo vision functions on inputs from two monocular cameras.
In at least one embodiment, a PVA may be used to perform dense optical flow. For example, in at least one embodiment, a PVA could process raw RADAR data (e.g., using a 4D Fast Fourier Transform) to provide processed RADAR data. In at least one embodiment, a PVA is used for time of flight depth processing, by processing raw time of flight data to provide processed time of flight data, for example.
In at least one embodiment, a DLA may be used to run any type of network to enhance control and driving safety, including for example and without limitation, a neural network that outputs a measure of confidence for each object detection. In at least one embodiment, confidence may be represented or interpreted as a probability, or as providing a relative “weight” of each detection compared to other detections. In at least one embodiment, a confidence measure enables a system to make further decisions regarding which detections should be considered as true positive detections rather than false positive detections. In at least one embodiment, a system may set a threshold value for confidence and consider only detections exceeding threshold value as true positive detections. In an embodiment in which an automatic emergency braking (“AEB”) system is used, false positive detections would cause vehicle to automatically perform emergency braking, which is obviously undesirable. In at least one embodiment, highly confident detections may be considered as triggers for AEB. In at least one embodiment, a DLA may run a neural network for regressing confidence value. In at least one embodiment, neural network may take as its input at least some subset of parameters, such as bounding box dimensions, ground plane estimate obtained (e.g., from another subsystem), output from IMU sensor(s) 1566 that correlates with vehicle 1500 orientation, distance, 3D location estimates of object obtained from neural network and/or other sensors (e.g., LIDAR sensor(s) 1564 or RADAR sensor(s) 1560), among others.
In at least one embodiment, one or more of SoC(s) 1504 may include data store(s) 1516 (e.g., memory). In at least one embodiment, data store(s) 1516 may be on-chip memory of SoC(s) 1504, which may store neural networks to be executed on GPU(s) 1508 and/or a DLA. In at least one embodiment, data store(s) 1516 may be large enough in capacity to store multiple instances of neural networks for redundancy and safety. In at least one embodiment, data store(s) 1516 may comprise L2 or L3 cache(s).
In at least one embodiment, one or more of SoC(s) 1504 may include any number of processor(s) 1510 (e.g., embedded processors). In at least one embodiment, processor(s) 1510 may include a boot and power management processor that may be a dedicated processor and subsystem to handle boot power and management functions and related security enforcement. In at least one embodiment, a boot and power management processor may be a part of a boot sequence of SoC(s) 1504 and may provide runtime power management services. In at least one embodiment, a boot power and management processor may provide clock and voltage programming, assistance in system low power state transitions, management of SoC(s) 1504 thermals and temperature sensors, and/or management of SoC(s) 1504 power states. In at least one embodiment, each temperature sensor may be implemented as a ring-oscillator whose output frequency is proportional to temperature, and SoC(s) 1504 may use ring-oscillators to detect temperatures of CPU(s) 1506, GPU(s) 1508, and/or accelerator(s) 1514. In at least one embodiment, if temperatures are determined to exceed a threshold, then a boot and power management processor may enter a temperature fault routine and put SoC(s) 1504 into a lower power state and/or put vehicle 1500 into a chauffeur to safe stop mode (e.g., bring vehicle 1500 to a safe stop).
In at least one embodiment, processor(s) 1510 may further include a set of embedded processors that may serve as an audio processing engine which may be an audio subsystem that enables full hardware support for multi-channel audio over multiple interfaces, and a broad and flexible range of audio I/O interfaces. In at least one embodiment, an audio processing engine is a dedicated processor core with a digital signal processor with dedicated RAM.
In at least one embodiment, processor(s) 1510 may further include an always-on processor engine that may provide necessary hardware features to support low power sensor management and wake use cases. In at least one embodiment, an always-on processor engine may include, without limitation, a processor core, a tightly coupled RAM, supporting peripherals (e.g., timers and interrupt controllers), various I/O controller peripherals, and routing logic.
In at least one embodiment, processor(s) 1510 may further include a safety cluster engine that includes, without limitation, a dedicated processor subsystem to handle safety management for automotive applications. In at least one embodiment, a safety cluster engine may include, without limitation, two or more processor cores, a tightly coupled RAM, support peripherals (e.g., timers, an interrupt controller, etc.), and/or routing logic. In a safety mode, two or more cores may operate, in at least one embodiment, in a lockstep mode and function as a single core with comparison logic to detect any differences between their operations. In at least one embodiment, processor(s) 1510 may further include a real-time camera engine that may include, without limitation, a dedicated processor subsystem for handling real-time camera management. In at least one embodiment, processor(s) 1510 may further include a high-dynamic range signal processor that may include, without limitation, an image signal processor that is a hardware engine that is part of a camera processing pipeline.
In at least one embodiment, processor(s) 1510 may include a video image compositor that may be a processing block (e.g., implemented on a microprocessor) that implements video post-processing functions needed by a video playback application to produce a final image for a player window. In at least one embodiment, a video image compositor may perform lens distortion correction on wide-view camera(s) 1570, surround camera(s) 1574, and/or on in-cabin monitoring camera sensor(s). In at least one embodiment, in-cabin monitoring camera sensor(s) are preferably monitored by a neural network running on another instance of SoC 1504, configured to identify in cabin events and respond accordingly. In at least one embodiment, an in-cabin system may perform, without limitation, lip reading to activate cellular service and place a phone call, dictate emails, change a vehicle's destination, activate or change a vehicle's infotainment system and settings, or provide voice-activated web surfing. In at least one embodiment, certain functions are available to a driver when a vehicle is operating in an autonomous mode and are disabled otherwise.
In at least one embodiment, a video image compositor may include enhanced temporal noise reduction for both spatial and temporal noise reduction. For example, in at least one embodiment, where motion occurs in a video, noise reduction weights spatial information appropriately, decreasing weights of information provided by adjacent frames. In at least one embodiment, where an image or portion of an image does not include motion, temporal noise reduction performed by video image compositor may use information from a previous image to reduce noise in a current image.
In at least one embodiment, a video image compositor may also be configured to perform stereo rectification on input stereo lens frames. In at least one embodiment, a video image compositor may further be used for user interface composition when an operating system desktop is in use, and GPU(s) 1508 are not required to continuously render new surfaces. In at least one embodiment, when GPU(s) 1508 are powered on and active doing 3D rendering, a video image compositor may be used to offload GPU(s) 1508 to improve performance and responsiveness.
In at least one embodiment, one or more SoC of SoC(s) 1504 may further include a mobile industry processor interface (“MIPI”) camera serial interface for receiving video and input from cameras, a high-speed interface, and/or a video input block that may be used for a camera and related pixel input functions. In at least one embodiment, one or more of SoC(s) 1504 may further include an input/output controller(s) that may be controlled by software and may be used for receiving I/O signals that are uncommitted to a specific role.
In at least one embodiment, one or more Soc of SoC(s) 1504 may further include a broad range of peripheral interfaces to enable communication with peripherals, audio encoders/decoders (“codecs”), power management, and/or other devices. In at least one embodiment, SoC(s) 1504 may be used to process data from cameras (e.g., connected over Gigabit Multimedia Serial Link and Ethernet channels), sensors (e.g., LIDAR sensor(s) 1564, RADAR sensor(s) 1560, etc. that may be connected over Ethernet channels), data from bus 1502 (e.g., speed of vehicle 1500, steering wheel position, etc.), data from GNSS sensor(s) 1558 (e.g., connected over an Ethernet bus or a CAN bus), etc. In at least one embodiment, one or more SoC of SoC(s) 1504 may further include dedicated high-performance mass storage controllers that may include their own DMA engines, and that may be used to free CPU(s) 1506 from routine data management tasks.
In at least one embodiment, SoC(s) 1504 may be an end-to-end platform with a flexible architecture that spans automation Levels 3-5, thereby providing a comprehensive functional safety architecture that leverages and makes efficient use of computer vision and ADAS techniques for diversity and redundancy, and provides a platform for a flexible, reliable driving software stack, along with deep learning tools. In at least one embodiment, SoC(s) 1504 may be faster, more reliable, and even more energy-efficient and space-efficient than conventional systems. For example, in at least one embodiment, accelerator(s) 1514, when combined with CPU(s) 1506, GPU(s) 1508, and data store(s) 1516, may provide for a fast, efficient platform for Level 3-5 autonomous vehicles.
In at least one embodiment, computer vision algorithms may be executed on CPUs, which may be configured using a high-level programming language, such as C, to execute a wide variety of processing algorithms across a wide variety of visual data. However, in at least one embodiment, CPUs are oftentimes unable to meet performance requirements of many computer vision applications, such as those related to execution time and power consumption, for example. In at least one embodiment, many CPUs are unable to execute complex object detection algorithms in real-time, which is used in in-vehicle ADAS applications and in practical Level 3-5 autonomous vehicles.
Embodiments described herein allow for multiple neural networks to be performed simultaneously and/or sequentially, and for results to be combined together to enable Level 3-5 autonomous driving functionality. For example, in at least one embodiment, a CNN executing on a DLA or a discrete GPU (e.g., GPU(s) 1520) may include text and word recognition, allowing reading and understanding of traffic signs, including signs for which a neural network has not been specifically trained. In at least one embodiment, a DLA may further include a neural network that is able to identify, interpret, and provide semantic understanding of a sign, and to pass that semantic understanding to path planning modules running on a CPU Complex.
In at least one embodiment, multiple neural networks may be run simultaneously, as for Level 3, 4, or 5 driving. For example, in at least one embodiment, a warning sign stating “Caution: flashing lights indicate icy conditions,” along with an electric light, may be independently or collectively interpreted by several neural networks. In at least one embodiment, such warning sign itself may be identified as a traffic sign by a first deployed neural network (e.g., a neural network that has been trained), text “flashing lights indicate icy conditions” may be interpreted by a second deployed neural network, which informs a vehicle's path planning software (preferably executing on a CPU Complex) that when flashing lights are detected, icy conditions exist. In at least one embodiment, a flashing light may be identified by operating a third deployed neural network over multiple frames, informing a vehicle's path-planning software of a presence (or an absence) of flashing lights. In at least one embodiment, all three neural networks may run simultaneously, such as within a DLA and/or on GPU(s) 1508.
In at least one embodiment, a CNN for facial recognition and vehicle owner identification may use data from camera sensors to identify presence of an authorized driver and/or owner of vehicle 1500. In at least one embodiment, an always-on sensor processing engine may be used to unlock a vehicle when an owner approaches a driver door and turns on lights, and, in a security mode, to disable such vehicle when an owner leaves such vehicle. In this way, SoC(s) 1504 provide for security against theft and/or carjacking.
In at least one embodiment, a CNN for emergency vehicle detection and identification may use data from microphones 1596 to detect and identify emergency vehicle sirens. In at least one embodiment, SoC(s) 1504 use a CNN for classifying environmental and urban sounds, as well as classifying visual data. In at least one embodiment, a CNN running on a DLA is trained to identify a relative closing speed of an emergency vehicle (e.g., by using a Doppler effect). In at least one embodiment, a CNN may also be trained to identify emergency vehicles specific to a local area in which a vehicle is operating, as identified by GNSS sensor(s) 1558. In at least one embodiment, when operating in Europe, a CNN will seek to detect European sirens, and when in North America, a CNN will seek to identify only North American sirens. In at least one embodiment, once an emergency vehicle is detected, a control program may be used to execute an emergency vehicle safety routine, slowing a vehicle, pulling over to a side of a road, parking a vehicle, and/or idling a vehicle, with assistance of ultrasonic sensor(s) 1562, until emergency vehicles pass.
In at least one embodiment, vehicle 1500 may include CPU(s) 1518 (e.g., discrete CPU(s), or dCPU(s)), that may be coupled to SoC(s) 1504 via a high-speed interconnect (e.g., PCIe). In at least one embodiment, CPU(s) 1518 may include an X86 processor, for example. CPU(s) 1518 may be used to perform any of a variety of functions, including arbitrating potentially inconsistent results between ADAS sensors and SoC(s) 1504, and/or monitoring status and health of controller(s) 1536 and/or an infotainment system on a chip (“infotainment SoC”) 1530, for example.
In at least one embodiment, vehicle 1500 may include GPU(s) 1520 (e.g., discrete GPU(s), or dGPU(s)), that may be coupled to SoC(s) 1504 via a high-speed interconnect (e.g., NVIDIA's NVLINK channel). In at least one embodiment, GPU(s) 1520 may provide additional artificial intelligence functionality, such as by executing redundant and/or different neural networks, and may be used to train and/or update neural networks based at least in part on input (e.g., sensor data) from sensors of a vehicle 1500.
In at least one embodiment, vehicle 1500 may further include network interface 1524 which may include, without limitation, wireless antenna(s) 1526 (e.g., one or more wireless antennas for different communication protocols, such as a cellular antenna, a Bluetooth antenna, etc.). In at least one embodiment, network interface 1524 may be used to enable wireless connectivity to Internet cloud services (e.g., with server(s) and/or other network devices), with other vehicles, and/or with computing devices (e.g., client devices of passengers). In at least one embodiment, to communicate with other vehicles, a direct link may be established between vehicle 150 and another vehicle and/or an indirect link may be established (e.g., across networks and over Internet). In at least one embodiment, direct links may be provided using a vehicle-to-vehicle communication link. In at least one embodiment, a vehicle-to-vehicle communication link may provide vehicle 1500 information about vehicles in proximity to vehicle 1500 (e.g., vehicles in front of, on a side of, and/or behind vehicle 1500). In at least one embodiment, such aforementioned functionality may be part of a cooperative adaptive cruise control functionality of vehicle 1500.
In at least one embodiment, network interface 1524 may include an SoC that provides modulation and demodulation functionality and enables controller(s) 1536 to communicate over wireless networks. In at least one embodiment, network interface 1524 may include a radio frequency front-end for up-conversion from baseband to radio frequency, and down conversion from radio frequency to baseband. In at least one embodiment, frequency conversions may be performed in any technically feasible fashion. For example, frequency conversions could be performed through well-known processes, and/or using super-heterodyne processes. In at least one embodiment, radio frequency front end functionality may be provided by a separate chip. In at least one embodiment, network interfaces may include wireless functionality for communicating over LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-Wave, ZigBee, LoRaWAN, and/or other wireless protocols.
In at least one embodiment, vehicle 1500 may further include data store(s) 1528 which may include, without limitation, off-chip (e.g., off SoC(s) 1504) storage. In at least one embodiment, data store(s) 1528 may include, without limitation, one or more storage elements including RAM, SRAM, dynamic random-access memory (“DRAM”), video random-access memory (“VRAM”), flash memory, hard disks, and/or other components and/or devices that may store at least one bit of data.
In at least one embodiment, vehicle 1500 may further include GNSS sensor(s) 1558 (e.g., GPS and/or assisted GPS sensors), to assist in mapping, perception, occupancy grid generation, and/or path planning functions. In at least one embodiment, any number of GNSS sensor(s) 1558 may be used, including, for example and without limitation, a GPS using a USB connector with an Ethernet-to-Serial (e.g., RS-232) bridge.
In at least one embodiment, vehicle 1500 may further include RADAR sensor(s) 1560. In at least one embodiment, RADAR sensor(s) 1560 may be used by vehicle 1500 for long-range vehicle detection, even in darkness and/or severe weather conditions. In at least one embodiment, RADAR functional safety levels may be ASIL B. In at least one embodiment, RADAR sensor(s) 1560 may use a CAN bus and/or bus 1502 (e.g., to transmit data generated by RADAR sensor(s) 1560) for control and to access object tracking data, with access to Ethernet channels to access raw data in some examples. In at least one embodiment, a wide variety of RADAR sensor types may be used. For example, and without limitation, RADAR sensor(s) 1560 may be suitable for front, rear, and side RADAR use. In at least one embodiment, one or more sensor of RADAR sensors(s) 1560 is a Pulse Doppler RADAR sensor.
In at least one embodiment, RADAR sensor(s) 1560 may include different configurations, such as long-range with narrow field of view, short-range with wide field of view, short-range side coverage, etc. In at least one embodiment, long-range RADAR may be used for adaptive cruise control functionality. In at least one embodiment, long-range RADAR systems may provide a broad field of view realized by two or more independent scans, such as within a 250 m (meter) range. In at least one embodiment, RADAR sensor(s) 1560 may help in distinguishing between static and moving objects, and may be used by ADAS system 1538 for emergency brake assist and forward collision warning. In at least one embodiment, sensors 1560(s) included in a long-range RADAR system may include, without limitation, monostatic multimodal RADAR with multiple (e.g., six or more) fixed RADAR antennae and a high-speed CAN and FlexRay interface. In at least one embodiment, with six antennae, a central four antennae may create a focused beam pattern, designed to record vehicle's 1500 surroundings at higher speeds with minimal interference from traffic in adjacent lanes. In at least one embodiment, another two antennae may expand field of view, making it possible to quickly detect vehicles entering or leaving a lane of vehicle 1500.
In at least one embodiment, mid-range RADAR systems may include, as an example, a range of up to 160 m (front) or 80 m (rear), and a field of view of up to 42 degrees (front) or 150 degrees (rear). In at least one embodiment, short-range RADAR systems may include, without limitation, any number of RADAR sensor(s) 1560 designed to be installed at both ends of a rear bumper. When installed at both ends of a rear bumper, in at least one embodiment, a RADAR sensor system may create two beams that constantly monitor blind spots in a rear direction and next to a vehicle. In at least one embodiment, short-range RADAR systems may be used in ADAS system 1538 for blind spot detection and/or lane change assist.
In at least one embodiment, vehicle 1500 may further include ultrasonic sensor(s) 1562. In at least one embodiment, ultrasonic sensor(s) 1562, which may be positioned at a front, a back, and/or side location of vehicle 1500, may be used for parking assist and/or to create and update an occupancy grid. In at least one embodiment, a wide variety of ultrasonic sensor(s) 1562 may be used, and different ultrasonic sensor(s) 1562 may be used for different ranges of detection (e.g., 2.5 m, 4 m). In at least one embodiment, ultrasonic sensor(s) 1562 may operate at functional safety levels of ASIL B.
In at least one embodiment, vehicle 1500 may include LIDAR sensor(s) 1564. In at least one embodiment, LIDAR sensor(s) 1564 may be used for object and pedestrian detection, emergency braking, collision avoidance, and/or other functions. In at least one embodiment, LIDAR sensor(s) 1564 may operate at functional safety level ASIL B. In at least one embodiment, vehicle 1500 may include multiple LIDAR sensors 1564 (e.g., two, four, six, etc.) that may use an Ethernet channel (e.g., to provide data to a Gigabit Ethernet switch).
In at least one embodiment, LIDAR sensor(s) 1564 may be capable of providing a list of objects and their distances for a 360-degree field of view. In at least one embodiment, commercially available LIDAR sensor(s) 1564 may have an advertised range of approximately 100 m, with an accuracy of 2 cm to 3 cm, and with support for a 100 Mbps Ethernet connection, for example. In at least one embodiment, one or more non-protruding LIDAR sensors may be used. In such an embodiment, LIDAR sensor(s) 1564 may include a small device that may be embedded into a front, a rear, a side, and/or a corner location of vehicle 1500. In at least one embodiment, LIDAR sensor(s) 1564, in such an embodiment, may provide up to a 120-degree horizontal and 35-degree vertical field-of-view, with a 200 m range even for low-reflectivity objects. In at least one embodiment, front-mounted LIDAR sensor(s) 1564 may be configured for a horizontal field of view between 45 degrees and 135 degrees.
In at least one embodiment, LIDAR technologies, such as 3D flash LIDAR, may also be used. In at least one embodiment, 3D flash LIDAR uses a flash of a laser as a transmission source, to illuminate surroundings of vehicle 1500 up to approximately 200 m. In at least one embodiment, a flash LIDAR unit includes, without limitation, a receptor, which records laser pulse transit time and reflected light on each pixel, which in turn corresponds to a range from vehicle 1500 to objects. In at least one embodiment, flash LIDAR may allow for highly accurate and distortion-free images of surroundings to be generated with every laser flash. In at least one embodiment, four flash LIDAR sensors may be deployed, one at each side of vehicle 1500. In at least one embodiment, 3D flash LIDAR systems include, without limitation, a solid-state 3D staring array LIDAR camera with no moving parts other than a fan (e.g., a non-scanning LIDAR device). In at least one embodiment, flash LIDAR device may use a 5 nanosecond class I (eye-safe) laser pulse per frame and may capture reflected laser light as a 3D range point cloud and co-registered intensity data.
In at least one embodiment, vehicle 1500 may further include IMU sensor(s) 1566. In at least one embodiment, IMU sensor(s) 1566 may be located at a center of a rear axle of vehicle 1500. In at least one embodiment, IMU sensor(s) 1566 may include, for example and without limitation, accelerometer(s), magnetometer(s), gyroscope(s), a magnetic compass, magnetic compasses, and/or other sensor types. In at least one embodiment, such as in six-axis applications, IMU sensor(s) 1566 may include, without limitation, accelerometers and gyroscopes. In at least one embodiment, such as in nine-axis applications, IMU sensor(s) 1566 may include, without limitation, accelerometers, gyroscopes, and magnetometers.
In at least one embodiment, IMU sensor(s) 1566 may be implemented as a miniature, high performance GPS-Aided Inertial Navigation System (“GPS/INS”) that combines micro-electro-mechanical systems (“MEMS”) inertial sensors, a high-sensitivity GPS receiver, and advanced Kalman filtering algorithms to provide estimates of position, velocity, and attitude. In at least one embodiment, IMU sensor(s) 1566 may enable vehicle 1500 to estimate its heading without requiring input from a magnetic sensor by directly observing and correlating changes in velocity from a GPS to IMU sensor(s) 1566. In at least one embodiment, IMU sensor(s) 1566 and GNSS sensor(s) 1558 may be combined in a single integrated unit.
In at least one embodiment, vehicle 1500 may include microphone(s) 1596 placed in and/or around vehicle 1500. In at least one embodiment, microphone(s) 1596 may be used for emergency vehicle detection and identification, among other things.
In at least one embodiment, vehicle 1500 may further include any number of camera types, including stereo camera(s) 1568, wide-view camera(s) 1570, infrared camera(s) 1572, surround camera(s) 1574, long-range camera(s) 1598, mid-range camera(s) 1576, and/or other camera types. In at least one embodiment, cameras may be used to capture image data around an entire periphery of vehicle 1500. In at least one embodiment, which types of cameras used depends on vehicle 1500. In at least one embodiment, any combination of camera types may be used to provide necessary coverage around vehicle 1500. In at least one embodiment, a number of cameras deployed may differ depending on embodiment. For example, in at least one embodiment, vehicle 1500 could include six cameras, seven cameras, ten cameras, twelve cameras, or another number of cameras. In at least one embodiment, cameras may support, as an example and without limitation, Gigabit Multimedia Serial Link (“GMSL”) and/or Gigabit Ethernet communications. In at least one embodiment, each camera might be as described with more detail previously herein with respect to
In at least one embodiment, vehicle 1500 may further include vibration sensor(s) 1542. In at least one embodiment, vibration sensor(s) 1542 may measure vibrations of components of vehicle 1500, such as axle(s). For example, in at least one embodiment, changes in vibrations may indicate a change in road surfaces. In at least one embodiment, when two or more vibration sensors 1542 are used, differences between vibrations may be used to determine friction or slippage of road surface (e.g., when a difference in vibration is between a power-driven axle and a freely rotating axle).
In at least one embodiment, vehicle 1500 may include ADAS system 1538. In at least one embodiment, ADAS system 1538 may include, without limitation, an SoC, in some examples. In at least one embodiment, ADAS system 1538 may include, without limitation, any number and combination of an autonomous/adaptive/automatic cruise control (“ACC”) system, a cooperative adaptive cruise control (“CACC”) system, a forward crash warning (“FCW”) system, an automatic emergency braking (“AEB”) system, a lane departure warning (“LDW)” system, a lane keep assist (“LKA”) system, a blind spot warning (“BSW”) system, a rear cross-traffic warning (“RCTW”) system, a collision warning (“CW”) system, a lane centering (“LC”) system, and/or other systems, features, and/or functionality.
In at least one embodiment, ACC system may use RADAR sensor(s) 1560, LIDAR sensor(s) 1564, and/or any number of camera(s). In at least one embodiment, ACC system may include a longitudinal ACC system and/or a lateral ACC system. In at least one embodiment, a longitudinal ACC system monitors and controls distance to another vehicle immediately ahead of vehicle 1500 and automatically adjusts speed of vehicle 1500 to maintain a safe distance from vehicles ahead. In at least one embodiment, a lateral ACC system performs distance keeping, and advises vehicle 1500 to change lanes when necessary. In at least one embodiment, a lateral ACC is related to other ADAS applications, such as LC and CW.
In at least one embodiment, a CACC system uses information from other vehicles that may be received via network interface 1524 and/or wireless antenna(s) 1526 from other vehicles via a wireless link, or indirectly, over a network connection (e.g., over Internet). In at least one embodiment, direct links may be provided by a vehicle-to-vehicle (“V2V”) communication link, while indirect links may be provided by an infrastructure-to-vehicle (“I2V”) communication link. In general, V2V communication provides information about immediately preceding vehicles (e.g., vehicles immediately ahead of and in same lane as vehicle 1500), while I2V communication provides information about traffic further ahead. In at least one embodiment, a CACC system may include either or both I2V and V2V information sources. In at least one embodiment, given information of vehicles ahead of vehicle 1500, a CACC system may be more reliable and it has potential to improve traffic flow smoothness and reduce congestion on road.
In at least one embodiment, an FCW system is designed to alert a driver to a hazard, so that such driver may take corrective action. In at least one embodiment, an FCW system uses a front-facing camera and/or RADAR sensor(s) 1560, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to provide driver feedback, such as a display, speaker, and/or vibrating component. In at least one embodiment, an FCW system may provide a warning, such as in form of a sound, visual warning, vibration and/or a quick brake pulse.
In at least one embodiment, an AEB system detects an impending forward collision with another vehicle or other object, and may automatically apply brakes if a driver does not take corrective action within a specified time or distance parameter. In at least one embodiment, AEB system may use front-facing camera(s) and/or RADAR sensor(s) 1560, coupled to a dedicated processor, DSP, FPGA, and/or ASIC. In at least one embodiment, when an AEB system detects a hazard, it will typically first alert a driver to take corrective action to avoid collision and, if that driver does not take corrective action, that AEB system may automatically apply brakes in an effort to prevent, or at least mitigate, an impact of a predicted collision. In at least one embodiment, an AEB system may include techniques such as dynamic brake support and/or crash imminent braking.
In at least one embodiment, an LDW system provides visual, audible, and/or tactile warnings, such as steering wheel or seat vibrations, to alert driver when vehicle 1500 crosses lane markings. In at least one embodiment, an LDW system does not activate when a driver indicates an intentional lane departure, such as by activating a turn signal. In at least one embodiment, an LDW system may use front-side facing cameras, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to provide driver feedback, such as a display, speaker, and/or vibrating component. In at least one embodiment, an LKA system is a variation of an LDW system. In at least one embodiment, an LKA system provides steering input or braking to correct vehicle 1500 if vehicle 1500 starts to exit its lane.
In at least one embodiment, a BSW system detects and warns a driver of vehicles in an automobile's blind spot. In at least one embodiment, a BSW system may provide a visual, audible, and/or tactile alert to indicate that merging or changing lanes is unsafe. In at least one embodiment, a BSW system may provide an additional warning when a driver uses a turn signal. In at least one embodiment, a BSW system may use rear-side facing camera(s) and/or RADAR sensor(s) 1560, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to driver feedback, such as a display, speaker, and/or vibrating component.
In at least one embodiment, an RCTW system may provide visual, audible, and/or tactile notification when an object is detected outside a rear-camera range when vehicle 1500 is backing up. In at least one embodiment, an RCTW system includes an AEB system to ensure that vehicle brakes are applied to avoid a crash. In at least one embodiment, an RCTW system may use one or more rear-facing RADAR sensor(s) 1560, coupled to a dedicated processor, DSP, FPGA, and/or ASIC, that is electrically coupled to provide driver feedback, such as a display, speaker, and/or vibrating component.
In at least one embodiment, conventional ADAS systems may be prone to false positive results which may be annoying and distracting to a driver, but typically are not catastrophic, because conventional ADAS systems alert a driver and allow that driver to decide whether a safety condition truly exists and act accordingly. In at least one embodiment, vehicle 1500 itself decides, in case of conflicting results, whether to heed result from a primary computer or a secondary computer (e.g., a first controller or a second controller of controllers 1536). For example, in at least one embodiment, ADAS system 1538 may be a backup and/or secondary computer for providing perception information to a backup computer rationality module. In at least one embodiment, a backup computer rationality monitor may run redundant diverse software on hardware components to detect faults in perception and dynamic driving tasks. In at least one embodiment, outputs from ADAS system 1538 may be provided to a supervisory MCU. In at least one embodiment, if outputs from a primary computer and outputs from a secondary computer conflict, a supervisory MCU determines how to reconcile conflict to ensure safe operation.
In at least one embodiment, a primary computer may be configured to provide a supervisory MCU with a confidence score, indicating that primary computer's confidence in a chosen result. In at least one embodiment, if that confidence score exceeds a threshold, that supervisory MCU may follow that primary computer's direction, regardless of whether that secondary computer provides a conflicting or inconsistent result. In at least one embodiment, where a confidence score does not meet a threshold, and where primary and secondary computers indicate different results (e.g., a conflict), a supervisory MCU may arbitrate between computers to determine an appropriate outcome.
In at least one embodiment, a supervisory MCU may be configured to run a neural network(s) that is trained and configured to determine, based at least in part on outputs from a primary computer and outputs from a secondary computer, conditions under which that secondary computer provides false alarms. In at least one embodiment, neural network(s) in a supervisory MCU may learn when a secondary computer's output may be trusted, and when it cannot. For example, in at least one embodiment, when that secondary computer is a RADAR-based FCW system, a neural network(s) in that supervisory MCU may learn when an FCW system is identifying metallic objects that are not, in fact, hazards, such as a drainage grate or manhole cover that triggers an alarm. In at least one embodiment, when a secondary computer is a camera-based LDW system, a neural network in a supervisory MCU may learn to override LDW when bicyclists or pedestrians are present and a lane departure is, in fact, a safest maneuver. In at least one embodiment, a supervisory MCU may include at least one of a DLA or a GPU suitable for running neural network(s) with associated memory. In at least one embodiment, a supervisory MCU may comprise and/or be included as a component of SoC(s) 1504.
In at least one embodiment, ADAS system 1538 may include a secondary computer that performs ADAS functionality using traditional rules of computer vision. In at least one embodiment, that secondary computer may use classic computer vision rules (if-then), and presence of a neural network(s) in a supervisory MCU may improve reliability, safety and performance. For example, in at least one embodiment, diverse implementation and intentional non-identity makes an overall system more fault-tolerant, especially to faults caused by software (or software-hardware interface) functionality. For example, in at least one embodiment, if there is a software bug or error in software running on a primary computer, and non-identical software code running on a secondary computer provides a consistent overall result, then a supervisory MCU may have greater confidence that an overall result is correct, and a bug in software or hardware on that primary computer is not causing a material error.
In at least one embodiment, an output of ADAS system 1538 may be fed into a primary computer's perception block and/or a primary computer's dynamic driving task block. For example, in at least one embodiment, if ADAS system 1538 indicates a forward crash warning due to an object immediately ahead, a perception block may use this information when identifying objects. In at least one embodiment, a secondary computer may have its own neural network that is trained and thus reduces a risk of false positives, as described herein.
In at least one embodiment, vehicle 1500 may further include infotainment SoC 1530 (e.g., an in-vehicle infotainment system (IVI)). Although illustrated and described as an SoC, infotainment system SoC 1530, in at least one embodiment, may not be an SoC, and may include, without limitation, two or more discrete components. In at least one embodiment, infotainment SoC 1530 may include, without limitation, a combination of hardware and software that may be used to provide audio (e.g., music, a personal digital assistant, navigational instructions, news, radio, etc.), video (e.g., TV, movies, streaming, etc.), phone (e.g., hands-free calling), network connectivity (e.g., LTE, WiFi, etc.), and/or information services (e.g., navigation systems, rear-parking assistance, a radio data system, vehicle related information such as fuel level, total distance covered, brake fuel level, oil level, door open/close, air filter information, etc.) to vehicle 1500. For example, infotainment SoC 1530 could include radios, disk players, navigation systems, video players, USB and Bluetooth connectivity, carputers, in-car entertainment, WiFi, steering wheel audio controls, hands free voice control, a heads-up display (“HUD”), HMI display 1534, a telematics device, a control panel (e.g., for controlling and/or interacting with various components, features, and/or systems), and/or other components. In at least one embodiment, infotainment SoC 1530 may further be used to provide information (e.g., visual and/or audible) to user(s) of vehicle 1500, such as information from ADAS system 1538, autonomous driving information such as planned vehicle maneuvers, trajectories, surrounding environment information (e.g., intersection information, vehicle information, road information, etc.), and/or other information.
In at least one embodiment, infotainment SoC 1530 may include any amount and type of GPU functionality. In at least one embodiment, infotainment SoC 1530 may communicate over bus 1502 with other devices, systems, and/or components of vehicle 1500. In at least one embodiment, infotainment SoC 1530 may be coupled to a supervisory MCU such that a GPU of an infotainment system may perform some self-driving functions in event that primary controller(s) 1536 (e.g., primary and/or backup computers of vehicle 1500) fail. In at least one embodiment, infotainment SoC 1530 may put vehicle 1500 into a chauffeur to safe stop mode, as described herein.
In at least one embodiment, vehicle 1500 may further include instrument cluster 1532 (e.g., a digital dash, an electronic instrument cluster, a digital instrument panel, etc.). In at least one embodiment, instrument cluster 1532 may include, without limitation, a controller and/or supercomputer (e.g., a discrete controller or supercomputer). In at least one embodiment, instrument cluster 1532 may include, without limitation, any number and combination of a set of instrumentation such as a speedometer, fuel level, oil pressure, tachometer, odometer, turn indicators, gearshift position indicator, seat belt warning light(s), parking-brake warning light(s), engine-malfunction light(s), supplemental restraint system (e.g., airbag) information, lighting controls, safety system controls, navigation information, etc. In some examples, information may be displayed and/or shared among infotainment SoC 1530 and instrument cluster 1532. In at least one embodiment, instrument cluster 1532 may be included as part of infotainment SoC 1530, or vice versa.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, server(s) 1578 may receive, over network(s) 1590 and from vehicles, image data representative of images showing unexpected or changed road conditions, such as recently commenced road-work. In at least one embodiment, server(s) 1578 may transmit, over network(s) 1590 and to vehicles, neural networks 1592, updated or otherwise, and/or map information 1594, including, without limitation, information regarding traffic and road conditions. In at least one embodiment, updates to map information 1594 may include, without limitation, updates for HD map 1522, such as information regarding construction sites, potholes, detours, flooding, and/or other obstructions. In at least one embodiment, neural networks 1592, and/or map information 1594 may have resulted from new training and/or experiences represented in data received from any number of vehicles in an environment, and/or based at least in part on training performed at a data center (e.g., using server(s) 1578 and/or other servers).
In at least one embodiment, server(s) 1578 may be used to train machine learning models (e.g., neural networks) based at least in part on training data. In at least one embodiment, training data may be generated by vehicles, and/or may be generated in a simulation (e.g., using a game engine). In at least one embodiment, any amount of training data is tagged (e.g., where associated neural network benefits from supervised learning) and/or undergoes other pre-processing. In at least one embodiment, any amount of training data is not tagged and/or pre-processed (e.g., where associated neural network does not require supervised learning). In at least one embodiment, once machine learning models are trained, machine learning models may be used by vehicles (e.g., transmitted to vehicles over network(s) 1590), and/or machine learning models may be used by server(s) 1578 to remotely monitor vehicles.
In at least one embodiment, server(s) 1578 may receive data from vehicles and apply data to up-to-date real-time neural networks for real-time intelligent inferencing. In at least one embodiment, server(s) 1578 may include deep-learning supercomputers and/or dedicated AI computers powered by GPU(s) 1584, such as a DGX and DGX Station machines developed by NVIDIA. However, in at least one embodiment, server(s) 1578 may include deep learning infrastructure that uses CPU-powered data centers.
In at least one embodiment, deep-learning infrastructure of server(s) 1578 may be capable of fast, real-time inferencing, and may use that capability to evaluate and verify health of processors, software, and/or associated hardware in vehicle 1500. For example, in at least one embodiment, deep-learning infrastructure may receive periodic updates from vehicle 1500, such as a sequence of images and/or objects that vehicle 1500 has located in that sequence of images (e.g., via computer vision and/or other machine learning object classification techniques). In at least one embodiment, deep-learning infrastructure may run its own neural network to identify objects and compare them with objects identified by vehicle 1500 and, if results do not match and deep-learning infrastructure concludes that AI in vehicle 1500 is malfunctioning, then server(s) 1578 may transmit a signal to vehicle 1500 instructing a fail-safe computer of vehicle 1500 to assume control, notify passengers, and complete a safe parking maneuver.
In at least one embodiment, server(s) 1578 may include GPU(s) 1584 and one or more programmable inference accelerators (e.g., NVIDIA's TensorRT 3 devices). In at least one embodiment, a combination of GPU-powered servers and inference acceleration may make real-time responsiveness possible. In at least one embodiment, such as where performance is less critical, servers powered by CPUs, FPGAs, and other processors may be used for inferencing. In at least one embodiment, hardware structure(s) 1215 are used to perform one or more embodiments. Details regarding hardware structure(x) 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
Embodiments may be used in other devices such as handheld devices and embedded applications. Some examples of handheld devices include cellular phones, Internet Protocol devices, digital cameras, personal digital assistants (“PDAs”), and handheld PCs. In at least one embodiment, embedded applications may include a microcontroller, a digital signal processor (“DSP”), system on a chip, network computers (“NetPCs”), set-top boxes, network hubs, wide area network (“WAN”) switches, or any other system that may perform one or more instructions in accordance with at least one embodiment.
In at least one embodiment, computer system 1600 may include, without limitation, processor 1602 that may include, without limitation, one or more execution units 1608 to perform machine learning model training and/or inferencing according to techniques described herein. In at least one embodiment, computer system 1600 is a single processor desktop or server system, but in another embodiment, computer system 1600 may be a multiprocessor system. In at least one embodiment, processor 1602 may include, without limitation, a complex instruction set computer (“CISC”) microprocessor, a reduced instruction set computing (“RISC”) microprocessor, a very long instruction word (“VLIW”) microprocessor, a processor implementing a combination of instruction sets, or any other processor device, such as a digital signal processor, for example. In at least one embodiment, processor 1602 may be coupled to a processor bus 1610 that may transmit data signals between processor 1602 and other components in computer system 1600.
In at least one embodiment, processor 1602 may include, without limitation, a Level 1 (“L1”) internal cache memory (“cache”) 1604. In at least one embodiment, processor 1602 may have a single internal cache or multiple levels of internal cache. In at least one embodiment, cache memory may reside external to processor 1602. Other embodiments may also include a combination of both internal and external caches depending on particular implementation and needs. In at least one embodiment, a register file 1606 may store different types of data in various registers including, without limitation, integer registers, floating point registers, status registers, and an instruction pointer register.
In at least one embodiment, execution unit 1608, including, without limitation, logic to perform integer and floating point operations, also resides in processor 1602. In at least one embodiment, processor 1602 may also include a microcode (“ucode”) read only memory (“ROM”) that stores microcode for certain macro instructions. In at least one embodiment, execution unit 1608 may include logic to handle a packed instruction set 1609. In at least one embodiment, by including packed instruction set 1609 in an instruction set of a general-purpose processor, along with associated circuitry to execute instructions, operations used by many multimedia applications may be performed using packed data in processor 1602. In at least one embodiment, many multimedia applications may be accelerated and executed more efficiently by using a full width of a processor's data bus for performing operations on packed data, which may eliminate a need to transfer smaller units of data across that processor's data bus to perform one or more operations one data element at a time.
In at least one embodiment, execution unit 1608 may also be used in microcontrollers, embedded processors, graphics devices, DSPs, and other types of logic circuits. In at least one embodiment, computer system 1600 may include, without limitation, a memory 1620. In at least one embodiment, memory 1620 may be a Dynamic Random Access Memory (“DRAM”) device, a Static Random Access Memory (“SRAM”) device, a flash memory device, or another memory device. In at least one embodiment, memory 1620 may store instruction(s) 1619 and/or data 1621 represented by data signals that may be executed by processor 1602.
In at least one embodiment, a system logic chip may be coupled to processor bus 1610 and memory 1620. In at least one embodiment, a system logic chip may include, without limitation, a memory controller hub (“MCH”) 1616, and processor 1602 may communicate with MCH 1616 via processor bus 1610. In at least one embodiment, MCH 1616 may provide a high bandwidth memory path 1618 to memory 1620 for instruction and data storage and for storage of graphics commands, data and textures. In at least one embodiment, MCH 1616 may direct data signals between processor 1602, memory 1620, and other components in computer system 1600 and to bridge data signals between processor bus 1610, memory 1620, and a system I/O interface 1622. In at least one embodiment, a system logic chip may provide a graphics port for coupling to a graphics controller. In at least one embodiment, MCH 1616 may be coupled to memory 1620 through high bandwidth memory path 1618 and a graphics/video card 1612 may be coupled to MCH 1616 through an Accelerated Graphics Port (“AGP”) interconnect 1614.
In at least one embodiment, computer system 1600 may use system I/O interface 1622 as a proprietary hub interface bus to couple MCH 1616 to an I/O controller hub (“ICH”) 1630. In at least one embodiment, ICH 1630 may provide direct connections to some I/O devices via a local I/O bus. In at least one embodiment, a local I/O bus may include, without limitation, a high-speed I/O bus for connecting peripherals to memory 1620, a chipset, and processor 1602. Examples may include, without limitation, an audio controller 1629, a firmware hub (“flash BIOS”) 1628, a wireless transceiver 1626, a data storage 1624, a legacy I/O controller 1623 containing user input and keyboard interfaces 1625, a serial expansion port 1627, such as a Universal Serial Bus (“USB”) port, and a network controller 1634. In at least one embodiment, data storage 1624 may comprise a hard disk drive, a floppy disk drive, a CD-ROM device, a flash memory device, or other mass storage device.
In at least one embodiment,
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, electronic device 1700 may include, without limitation, processor 1710 communicatively coupled to any suitable number or kind of components, peripherals, modules, or devices. In at least one embodiment, processor 1710 is coupled using a bus or interface, such as a I2C bus, a System Management Bus (“SMBus”), a Low Pin Count (LPC) bus, a Serial Peripheral Interface (“SPI”), a High Definition Audio (“HDA”) bus, a Serial Advance Technology Attachment (“SATA”) bus, a Universal Serial Bus (“USB”) (versions 1, 2, 3, etc.), or a Universal Asynchronous Receiver/Transmitter (“UART”) bus. In at least one embodiment,
In at least one embodiment,
In at least one embodiment, other components may be communicatively coupled to processor 1710 through components described herein. In at least one embodiment, an accelerometer 1741, an ambient light sensor (“ALS”) 1742, a compass 1743, and a gyroscope 1744 may be communicatively coupled to sensor hub 1740. In at least one embodiment, a thermal sensor 1739, a fan 1737, a keyboard 1736, and touch pad 1730 may be communicatively coupled to EC 1735. In at least one embodiment, speakers 1763, headphones 1764, and a microphone (“mic”) 1765 may be communicatively coupled to an audio unit (“audio codec and class D amp”) 1762, which may in turn be communicatively coupled to DSP 1760. In at least one embodiment, audio unit 1762 may include, for example and without limitation, an audio coder/decoder (“codec”) and a class D amplifier. In at least one embodiment, a SIM card (“SIM”) 1757 may be communicatively coupled to WWAN unit 1756. In at least one embodiment, components such as WLAN unit 1750 and Bluetooth unit 1752, as well as WWAN unit 1756 may be implemented in a Next Generation Form Factor (“NGFF”).
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, computer system 1800 comprises, without limitation, at least one central processing unit (“CPU”) 1802 that is connected to a communication bus 1810 implemented using any suitable protocol, such as PCI (“Peripheral Component Interconnect”), peripheral component interconnect express (“PCI-Express”), AGP (“Accelerated Graphics Port”), HyperTransport, or any other bus or point-to-point communication protocol(s). In at least one embodiment, computer system 1800 includes, without limitation, a main memory 1804 and control logic (e.g., implemented as hardware, software, or a combination thereof) and data are stored in main memory 1804, which may take form of random access memory (“RAM”). In at least one embodiment, a network interface subsystem (“network interface”) 1822 provides an interface to other computing devices and networks for receiving data from and transmitting data to other systems with computer system 1800.
In at least one embodiment, computer system 1800, in at least one embodiment, includes, without limitation, input devices 1808, a parallel processing system 1812, and display devices 1806 that can be implemented using a conventional cathode ray tube (“CRT”), a liquid crystal display (“LCD”), a light emitting diode (“LED”) display, a plasma display, or other suitable display technologies. In at least one embodiment, user input is received from input devices 1808 such as keyboard, mouse, touchpad, microphone, etc. In at least one embodiment, each module described herein can be situated on a single semiconductor platform to form a processing system.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, USB stick 1920 includes, without limitation, a processing unit 1930, a USB interface 1940, and USB interface logic 1950. In at least one embodiment, processing unit 1930 may be any instruction execution system, apparatus, or device capable of executing instructions. In at least one embodiment, processing unit 1930 may include, without limitation, any number and type of processing cores (not shown). In at least one embodiment, processing unit 1930 comprises an application specific integrated circuit (“ASIC”) that is optimized to perform any amount and type of operations associated with machine learning. For instance, in at least one embodiment, processing unit 1930 is a tensor processing unit (“TPC”) that is optimized to perform machine learning inference operations. In at least one embodiment, processing unit 1930 is a vision processing unit (“VPU”) that is optimized to perform machine vision and machine learning inference operations.
In at least one embodiment, USB interface 1940 may be any type of USB connector or USB socket. For instance, in at least one embodiment, USB interface 1940 is a USB 3.0 Type-C socket for data and power. In at least one embodiment, USB interface 1940 is a USB 3.0 Type-A connector. In at least one embodiment, USB interface logic 1950 may include any amount and type of logic that enables processing unit 1930 to interface with devices (e.g., computer 1910) via USB connector 1940.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In addition, and in at least one embodiment, two or more of GPUs 2010 are interconnected over high-speed links 2029(1)-2029(2), which may be implemented using similar or different protocols/links than those used for high-speed links 2040(1)-2040(N). Similarly, two or more of multi-core processors 2005 may be connected over a high-speed link 2028 which may be symmetric multi-processor (SMP) buses operating at 20 GB/s, 30 GB/s, 120 GB/s or higher. Alternatively, all communication between various system components shown in
In at least one embodiment, each multi-core processor 2005 is communicatively coupled to a processor memory 2001(1)-2001(M), via memory interconnects 2026(1)-2026(M), respectively, and each GPU 2010(1)-2010(N) is communicatively coupled to GPU memory 2020(1)-2020(N) over GPU memory interconnects 2050(1)-2050(N), respectively. In at least one embodiment, memory interconnects 2026 and 2050 may utilize similar or different memory access technologies. By way of example, and not limitation, processor memories 2001(1)-2001(M) and GPU memories 2020 may be volatile memories such as dynamic random access memories (DRAMs) (including stacked DRAMs), Graphics DDR SDRAM (GDDR) (e.g., GDDR5, GDDR6), or High Bandwidth Memory (HBM) and/or may be non-volatile memories such as 3D)(Point or Nano-Ram. In at least one embodiment, some portion of processor memories 2001 may be volatile memory and another portion may be non-volatile memory (e.g., using a two-level memory (2LM) hierarchy).
As described herein, although various multi-core processors 2005 and GPUs 2010 may be physically coupled to a particular memory 2001, 2020, respectively, and/or a unified memory architecture may be implemented in which a virtual system address space (also referred to as “effective address” space) is distributed among various physical memories. For example, processor memories 2001(1)-2001(M) may each comprise 64 GB of system memory address space and GPU memories 2020(1)-2020(N) may each comprise 32 GB of system memory address space resulting in a total of 256 GB addressable memory when M=2 and N=4. Other values for N and M are possible.
In at least one embodiment, processor 2007 includes a plurality of cores 2060A-2060D, each with a translation lookaside buffer (“TLB”) 2061A-2061D and one or more caches 2062A-2062D. In at least one embodiment, cores 2060A-2060D may include various other components for executing instructions and processing data that are not illustrated. In at least one embodiment, caches 2062A-2062D may comprise Level 1 (L1) and Level 2 (L2) caches. In addition, one or more shared caches 2056 may be included in caches 2062A-2062D and shared by sets of cores 2060A-2060D. For example, one embodiment of processor 2007 includes 24 cores, each with its own L1 cache, twelve shared L2 caches, and twelve shared L3 caches. In this embodiment, one or more L2 and L3 caches are shared by two adjacent cores. In at least one embodiment, processor 2007 and graphics acceleration module 2046 connect with system memory 2014, which may include processor memories 2001(1)-2001(M) of
In at least one embodiment, coherency is maintained for data and instructions stored in various caches 2062A-2062D, 2056 and system memory 2014 via inter-core communication over a coherence bus 2064. In at least one embodiment, for example, each cache may have cache coherency logic/circuitry associated therewith to communicate to over coherence bus 2064 in response to detected reads or writes to particular cache lines. In at least one embodiment, a cache snooping protocol is implemented over coherence bus 2064 to snoop cache accesses.
In at least one embodiment, a proxy circuit 2025 communicatively couples graphics acceleration module 2046 to coherence bus 2064, allowing graphics acceleration module 2046 to participate in a cache coherence protocol as a peer of cores 2060A-2060D. In particular, in at least one embodiment, an interface 2035 provides connectivity to proxy circuit 2025 over high-speed link 2040 and an interface 2037 connects graphics acceleration module 2046 to high-speed link 2040.
In at least one embodiment, an accelerator integration circuit 2036 provides cache management, memory access, context management, and interrupt management services on behalf of a plurality of graphics processing engines 2031(1)-2031(N) of graphics acceleration module 2046. In at least one embodiment, graphics processing engines 2031(1)-2031(N) may each comprise a separate graphics processing unit (GPU). In at least one embodiment, graphics processing engines 2031(1)-2031(N) alternatively may comprise different types of graphics processing engines within a GPU, such as graphics execution units, media processing engines (e.g., video encoders/decoders), samplers, and blit engines. In at least one embodiment, graphics acceleration module 2046 may be a GPU with a plurality of graphics processing engines 2031(1)-2031(N) or graphics processing engines 2031(1)-2031(N) may be individual GPUs integrated on a common package, line card, or chip.
In at least one embodiment, accelerator integration circuit 2036 includes a memory management unit (MMU) 2039 for performing various memory management functions such as virtual-to-physical memory translations (also referred to as effective-to-real memory translations) and memory access protocols for accessing system memory 2014. In at least one embodiment, MMU 2039 may also include a translation lookaside buffer (TLB) (not shown) for caching virtual/effective to physical/real address translations. In at least one embodiment, a cache 2038 can store commands and data for efficient access by graphics processing engines 2031(1)-2031(N). In at least one embodiment, data stored in cache 2038 and graphics memories 2033(1)-2033(M) is kept coherent with core caches 2062A-2062D, 2056 and system memory 2014, possibly using a fetch unit 2044. As mentioned, this may be accomplished via proxy circuit 2025 on behalf of cache 2038 and memories 2033(1)-2033(M) (e.g., sending updates to cache 2038 related to modifications/accesses of cache lines on processor caches 2062A-2062D, 2056 and receiving updates from cache 2038).
In at least one embodiment, a set of registers 2045 store context data for threads executed by graphics processing engines 2031(1)-2031(N) and a context management circuit 2048 manages thread contexts. For example, context management circuit 2048 may perform save and restore operations to save and restore contexts of various threads during contexts switches (e.g., where a first thread is saved and a second thread is stored so that a second thread can be execute by a graphics processing engine). For example, on a context switch, context management circuit 2048 may store current register values to a designated region in memory (e.g., identified by a context pointer). It may then restore register values when returning to a context. In at least one embodiment, an interrupt management circuit 2047 receives and processes interrupts received from system devices.
In at least one embodiment, virtual/effective addresses from a graphics processing engine 2031 are translated to real/physical addresses in system memory 2014 by MMU 2039. In at least one embodiment, accelerator integration circuit 2036 supports multiple (e.g., 4, 8, 16) graphics accelerator modules 2046 and/or other accelerator devices. In at least one embodiment, graphics accelerator module 2046 may be dedicated to a single application executed on processor 2007 or may be shared between multiple applications. In at least one embodiment, a virtualized graphics execution environment is presented in which resources of graphics processing engines 2031(1)-2031(N) are shared with multiple applications or virtual machines (VMs). In at least one embodiment, resources may be subdivided into “slices” which are allocated to different VMs and/or applications based on processing requirements and priorities associated with VMs and/or applications.
In at least one embodiment, accelerator integration circuit 2036 performs as a bridge to a system for graphics acceleration module 2046 and provides address translation and system memory cache services. In addition, in at least one embodiment, accelerator integration circuit 2036 may provide virtualization facilities for a host processor to manage virtualization of graphics processing engines 2031(1)-2031(N), interrupts, and memory management.
In at least one embodiment, because hardware resources of graphics processing engines 2031(1)-2031(N) are mapped explicitly to a real address space seen by host processor 2007, any host processor can address these resources directly using an effective address value. In at least one embodiment, one function of accelerator integration circuit 2036 is physical separation of graphics processing engines 2031(1)-2031(N) so that they appear to a system as independent units.
In at least one embodiment, one or more graphics memories 2033(1)-2033(M) are coupled to each of graphics processing engines 2031(1)-2031(N), respectively and N=M. In at least one embodiment, graphics memories 2033(1)-2033(M) store instructions and data being processed by each of graphics processing engines 2031(1)-2031(N). In at least one embodiment, graphics memories 2033(1)-2033(M) may be volatile memories such as DRAMs (including stacked DRAMs), GDDR memory (e.g., GDDR5, GDDR6), or HBM, and/or may be non-volatile memories such as 3D XPoint or Nano-Ram.
In at least one embodiment, to reduce data traffic over high-speed link 2040, biasing techniques can be used to ensure that data stored in graphics memories 2033(1)-2033(M) is data that will be used most frequently by graphics processing engines 2031(1)-2031(N) and preferably not used by cores 2060A-2060D (at least not frequently). Similarly, in at least one embodiment, a biasing mechanism attempts to keep data needed by cores (and preferably not graphics processing engines 2031(1)-2031(N)) within caches 2062A-2062D, 2056 and system memory 2014.
In at least one embodiment, graphics processing engines 2031(1)-2031(N) are dedicated to a single application or process under a single operating system. In at least one embodiment, a single application can funnel other application requests to graphics processing engines 2031(1)-2031(N), providing virtualization within a VM/partition.
In at least one embodiment, graphics processing engines 2031(1)-2031(N), may be shared by multiple VM/application partitions. In at least one embodiment, shared models may use a system hypervisor to virtualize graphics processing engines 2031(1)-2031(N) to allow access by each operating system. In at least one embodiment, for single-partition systems without a hypervisor, graphics processing engines 2031(1)-2031(N) are owned by an operating system. In at least one embodiment, an operating system can virtualize graphics processing engines 2031(1)-2031(N) to provide access to each process or application.
In at least one embodiment, graphics acceleration module 2046 or an individual graphics processing engine 2031(1)-2031(N) selects a process element using a process handle. In at least one embodiment, process elements are stored in system memory 2014 and are addressable using an effective address to real address translation technique described herein. In at least one embodiment, a process handle may be an implementation-specific value provided to a host process when registering its context with graphics processing engine 2031(1)-2031(N) (that is, calling system software to add a process element to a process element linked list). In at least one embodiment, a lower 16-bits of a process handle may be an offset of a process element within a process element linked list.
In at least one embodiment, graphics acceleration module 2046 and/or individual graphics processing engines 2031(1)-2031(N) can be shared by all or a subset of processes in a system. In at least one embodiment, an infrastructure for setting up process states and sending a WD 2084 to a graphics acceleration module 2046 to start a job in a virtualized environment may be included.
In at least one embodiment, a dedicated-process programming model is implementation-specific. In at least one embodiment, in this model, a single process owns graphics acceleration module 2046 or an individual graphics processing engine 2031. In at least one embodiment, when graphics acceleration module 2046 is owned by a single process, a hypervisor initializes accelerator integration circuit 2036 for an owning partition and an operating system initializes accelerator integration circuit 2036 for an owning process when graphics acceleration module 2046 is assigned.
In at least one embodiment, in operation, a WD fetch unit 2091 in accelerator integration slice 2090 fetches next WD 2084, which includes an indication of work to be done by one or more graphics processing engines of graphics acceleration module 2046. In at least one embodiment, data from WD 2084 may be stored in registers 2045 and used by MMU 2039, interrupt management circuit 2047 and/or context management circuit 2048 as illustrated. For example, one embodiment of MMU 2039 includes segment/page walk circuitry for accessing segment/page tables 2086 within an OS virtual address space 2085. In at least one embodiment, interrupt management circuit 2047 may process interrupt events 2092 received from graphics acceleration module 2046. In at least one embodiment, when performing graphics operations, an effective address 2093 generated by a graphics processing engine 2031(1)-2031(N) is translated to a real address by MMU 2039.
In at least one embodiment, registers 2045 are duplicated for each graphics processing engine 2031(1)-2031(N) and/or graphics acceleration module 2046 and may be initialized by a hypervisor or an operating system. In at least one embodiment, each of these duplicated registers may be included in an accelerator integration slice 2090. Exemplary registers that may be initialized by a hypervisor are shown in Table 1.
Exemplary registers that may be initialized by an operating system are shown in Table 2.
In at least one embodiment, each WD 2084 is specific to a particular graphics acceleration module 2046 and/or graphics processing engines 2031(1)-2031(N). In at least one embodiment, it contains all information required by a graphics processing engine 2031(1)-2031(N) to do work, or it can be a pointer to a memory location where an application has set up a command queue of work to be completed.
In at least one embodiment, shared programming models allow for all or a subset of processes from all or a subset of partitions in a system to use a graphics acceleration module 2046. In at least one embodiment, there are two programming models where graphics acceleration module 2046 is shared by multiple processes and partitions, namely time-sliced shared and graphics directed shared.
In at least one embodiment, in this model, system hypervisor 2096 owns graphics acceleration module 2046 and makes its function available to all operating systems 2095. In at least one embodiment, for a graphics acceleration module 2046 to support virtualization by system hypervisor 2096, graphics acceleration module 2046 may adhere to certain requirements, such as (1) an application's job request must be autonomous (that is, state does not need to be maintained between jobs), or graphics acceleration module 2046 must provide a context save and restore mechanism, (2) an application's job request is guaranteed by graphics acceleration module 2046 to complete in a specified amount of time, including any translation faults, or graphics acceleration module 2046 provides an ability to preempt processing of a job, and (3) graphics acceleration module 2046 must be guaranteed fairness between processes when operating in a directed shared programming model.
In at least one embodiment, application 2080 is required to make an operating system 2095 system call with a graphics acceleration module type, a work descriptor (WD), an authority mask register (AMR) value, and a context save/restore area pointer (CSRP). In at least one embodiment, graphics acceleration module type describes a targeted acceleration function for a system call. In at least one embodiment, graphics acceleration module type may be a system-specific value. In at least one embodiment, WD is formatted specifically for graphics acceleration module 2046 and can be in a form of a graphics acceleration module 2046 command, an effective address pointer to a user-defined structure, an effective address pointer to a queue of commands, or any other data structure to describe work to be done by graphics acceleration module 2046.
In at least one embodiment, an AMR value is an AMR state to use for a current process. In at least one embodiment, a value passed to an operating system is similar to an application setting an AMR. In at least one embodiment, if accelerator integration circuit 2036 (not shown) and graphics acceleration module 2046 implementations do not support a User Authority Mask Override Register (UAMOR), an operating system may apply a current UAMOR value to an AMR value before passing an AMR in a hypervisor call. In at least one embodiment, hypervisor 2096 may optionally apply a current Authority Mask Override Register (AMOR) value before placing an AMR into process element 2083. In at least one embodiment, CSRP is one of registers 2045 containing an effective address of an area in an application's effective address space 2082 for graphics acceleration module 2046 to save and restore context state. In at least one embodiment, this pointer is optional if no state is required to be saved between jobs or when a job is preempted. In at least one embodiment, context save/restore area may be pinned system memory.
Upon receiving a system call, operating system 2095 may verify that application 2080 has registered and been given authority to use graphics acceleration module 2046. In at least one embodiment, operating system 2095 then calls hypervisor 2096 with information shown in Table 3.
In at least one embodiment, upon receiving a hypervisor call, hypervisor 2096 verifies that operating system 2095 has registered and been given authority to use graphics acceleration module 2046. In at least one embodiment, hypervisor 2096 then puts process element 2083 into a process element linked list for a corresponding graphics acceleration module 2046 type. In at least one embodiment, a process element may include information shown in Table 4.
In at least one embodiment, hypervisor initializes a plurality of accelerator integration slice 2090 registers 2045.
As illustrated in
In at least one embodiment, bias/coherence management circuitry 2094A-2094E within one or more of MMUs 2039A-2039E ensures cache coherence between caches of one or more host processors (e.g., 2005) and GPUs 2010 and implements biasing techniques indicating physical memories in which certain types of data should be stored. In at least one embodiment, while multiple instances of bias/coherence management circuitry 2094A-2094E are illustrated in
One embodiment allows GPU memories 2020 to be mapped as part of system memory, and accessed using shared virtual memory (SVM) technology, but without suffering performance drawbacks associated with full system cache coherence. In at least one embodiment, an ability for GPU memories 2020 to be accessed as system memory without onerous cache coherence overhead provides a beneficial operating environment for GPU offload. In at least one embodiment, this arrangement allows software of host processor 2005 to setup operands and access computation results, without overhead of tradition I/O DMA data copies. In at least one embodiment, such traditional copies involve driver calls, interrupts and memory mapped I/O (MMIO) accesses that are all inefficient relative to simple memory accesses. In at least one embodiment, an ability to access GPU memories 2020 without cache coherence overheads can be critical to execution time of an offloaded computation. In at least one embodiment, in cases with substantial streaming write memory traffic, for example, cache coherence overhead can significantly reduce an effective write bandwidth seen by a GPU 2010. In at least one embodiment, efficiency of operand setup, efficiency of results access, and efficiency of GPU computation may play a role in determining effectiveness of a GPU offload.
In at least one embodiment, selection of GPU bias and host processor bias is driven by a bias tracker data structure. In at least one embodiment, a bias table may be used, for example, which may be a page-granular structure (e.g., controlled at a granularity of a memory page) that includes 1 or 2 bits per GPU-attached memory page. In at least one embodiment, a bias table may be implemented in a stolen memory range of one or more GPU memories 2020, with or without a bias cache in a GPU 2010 (e.g., to cache frequently/recently used entries of a bias table). Alternatively, in at least one embodiment, an entire bias table may be maintained within a GPU.
In at least one embodiment, a bias table entry associated with each access to a GPU attached memory 2020 is accessed prior to actual access to a GPU memory, causing following operations. In at least one embodiment, local requests from a GPU 2010 that find their page in GPU bias are forwarded directly to a corresponding GPU memory 2020. In at least one embodiment, local requests from a GPU that find their page in host bias are forwarded to processor 2005 (e.g., over a high-speed link as described herein). In at least one embodiment, requests from processor 2005 that find a requested page in host processor bias complete a request like a normal memory read. Alternatively, requests directed to a GPU-biased page may be forwarded to a GPU 2010. In at least one embodiment, a GPU may then transition a page to a host processor bias if it is not currently using a page. In at least one embodiment, a bias state of a page can be changed either by a software-based mechanism, a hardware-assisted software-based mechanism, or, for a limited set of cases, a purely hardware-based mechanism.
In at least one embodiment, one mechanism for changing bias state employs an API call (e.g., OpenCL), which, in turn, calls a GPU's device driver which, in turn, sends a message (or enqueues a command descriptor) to a GPU directing it to change a bias state and, for some transitions, perform a cache flushing operation in a host. In at least one embodiment, a cache flushing operation is used for a transition from host processor 2005 bias to GPU bias, but is not for an opposite transition.
In at least one embodiment, cache coherency is maintained by temporarily rendering GPU-biased pages uncacheable by host processor 2005. In at least one embodiment, to access these pages, processor 2005 may request access from GPU 2010, which may or may not grant access right away. In at least one embodiment, thus, to reduce communication between processor 2005 and GPU 2010 it is beneficial to ensure that GPU-biased pages are those which are required by a GPU but not host processor 2005 and vice versa.
Hardware structure(s) 1215 are used to perform one or more embodiments. Details regarding a hardware structure(s) 1215 may be provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, graphics processor 2210 includes a vertex processor 2205 and one or more fragment processor(s) 2215A-2215N (e.g., 2215A, 2215B, 2215C, 2215D, through 2215N-1, and 2215N). In at least one embodiment, graphics processor 2210 can execute different shader programs via separate logic, such that vertex processor 2205 is optimized to execute operations for vertex shader programs, while one or more fragment processor(s) 2215A-2215N execute fragment (e.g., pixel) shading operations for fragment or pixel shader programs. In at least one embodiment, vertex processor 2205 performs a vertex processing stage of a 3D graphics pipeline and generates primitives and vertex data. In at least one embodiment, fragment processor(s) 2215A-2215N use primitive and vertex data generated by vertex processor 2205 to produce a framebuffer that is displayed on a display device. In at least one embodiment, fragment processor(s) 2215A-2215N are optimized to execute fragment shader programs as provided for in an OpenGL API, which may be used to perform similar operations as a pixel shader program as provided for in a Direct 3D API.
In at least one embodiment, graphics processor 2210 additionally includes one or more memory management units (MMUs) 2220A-2220B, cache(s) 2225A-2225B, and circuit interconnect(s) 2230A-2230B. In at least one embodiment, one or more MMU(s) 2220A-2220B provide for virtual to physical address mapping for graphics processor 2210, including for vertex processor 2205 and/or fragment processor(s) 2215A-2215N, which may reference vertex or image/texture data stored in memory, in addition to vertex or image/texture data stored in one or more cache(s) 2225A-2225B. In at least one embodiment, one or more MMU(s) 2220A-2220B may be synchronized with other MMUs within a system, including one or more MMUs associated with one or more application processor(s) 2105, image processors 2115, and/or video processors 2120 of
In at least one embodiment, graphics processor 2240 includes one or more shader core(s) 2255A-2255N (e.g., 2255A, 2255B, 2255C, 2255D, 2255E, 2255F, through 2255N-1, and 2255N) as shown in
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, graphics core 2300 includes a shared instruction cache 2302, a texture unit 2318, and a cache/shared memory 2320 that are common to execution resources within graphics core 2300. In at least one embodiment, graphics core 2300 can include multiple slices 2301A-2301N or a partition for each core, and a graphics processor can include multiple instances of graphics core 2300. In at least one embodiment, slices 2301A-2301N can include support logic including a local instruction cache 2304A-2304N, a thread scheduler 2306A-2306N, a thread dispatcher 2308A-2308N, and a set of registers 2310A-2310N. In at least one embodiment, slices 2301A-2301N can include a set of additional function units (AFUs 2312A-2312N), floating-point units (FPUs 2314A-2314N), integer arithmetic logic units (ALUs 2316A-2316N), address computational units (ACUs 2313A-2313N), double-precision floating-point units (DPFPUs 2315A-2315N), and matrix processing units (MPUs 2317A-2317N).
In at least one embodiment, FPUs 2314A-2314N can perform single-precision (32-bit) and half-precision (16-bit) floating point operations, while DPFPUs 2315A-2315N perform double precision (64-bit) floating point operations. In at least one embodiment, ALUs 2316A-2316N can perform variable precision integer operations at 8-bit, 16-bit, and 32-bit precision, and can be configured for mixed precision operations. In at least one embodiment, MPUs 2317A-2317N can also be configured for mixed precision matrix operations, including half-precision floating point and 8-bit integer operations. In at least one embodiment, MPUs 2317-2317N can perform a variety of matrix operations to accelerate machine learning application frameworks, including enabling support for accelerated general matrix to matrix multiplication (GEMM). In at least one embodiment, AFUs 2312A-2312N can perform additional logic operations not supported by floating-point or integer units, including trigonometric operations (e.g., sine, cosine, etc.).
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, GPGPU 2330 includes memory 2344A-2344B coupled with compute clusters 2336A-2336H via a set of memory controllers 2342A-2342B. In at least one embodiment, memory 2344A-2344B can include various types of memory devices including dynamic random access memory (DRAM) or graphics random access memory, such as synchronous graphics random access memory (SGRAM), including graphics double data rate (GDDR) memory.
In at least one embodiment, compute clusters 2336A-2336H each include a set of graphics cores, such as graphics core 2300 of
In at least one embodiment, multiple instances of GPGPU 2330 can be configured to operate as a compute cluster. In at least one embodiment, communication used by compute clusters 2336A-2336H for synchronization and data exchange varies across embodiments. In at least one embodiment, multiple instances of GPGPU 2330 communicate over host interface 2332. In at least one embodiment, GPGPU 2330 includes an I/O hub 2339 that couples GPGPU 2330 with a GPU link 2340 that enables a direct connection to other instances of GPGPU 2330. In at least one embodiment, GPU link 2340 is coupled to a dedicated GPU-to-GPU bridge that enables communication and synchronization between multiple instances of GPGPU 2330. In at least one embodiment, GPU link 2340 couples with a high-speed interconnect to transmit and receive data to other GPGPUs or parallel processors. In at least one embodiment, multiple instances of GPGPU 2330 are located in separate data processing systems and communicate via a network device that is accessible via host interface 2332. In at least one embodiment GPU link 2340 can be configured to enable a connection to a host processor in addition to or as an alternative to host interface 2332.
In at least one embodiment, GPGPU 2330 can be configured to train neural networks. In at least one embodiment, GPGPU 2330 can be used within an inferencing platform. In at least one embodiment, in which GPGPU 2330 is used for inferencing, GPGPU 2330 may include fewer compute clusters 2336A-2336H relative to when GPGPU 2330 is used for training a neural network. In at least one embodiment, memory technology associated with memory 2344A-2344B may differ between inferencing and training configurations, with higher bandwidth memory technologies devoted to training configurations. In at least one embodiment, an inferencing configuration of GPGPU 2330 can support inferencing specific instructions. For example, in at least one embodiment, an inferencing configuration can provide support for one or more 8-bit integer dot product instructions, which may be used during inferencing operations for deployed neural networks.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, processing subsystem 2401 includes one or more parallel processor(s) 2412 coupled to memory hub 2405 via a bus or other communication link 2413. In at least one embodiment, communication link 2413 may use one of any number of standards based communication link technologies or protocols, such as, but not limited to PCI Express, or may be a vendor-specific communications interface or communications fabric. In at least one embodiment, one or more parallel processor(s) 2412 form a computationally focused parallel or vector processing system that can include a large number of processing cores and/or processing clusters, such as a many-integrated core (MIC) processor. In at least one embodiment, some or all of parallel processor(s) 2412 form a graphics processing subsystem that can output pixels to one of one or more display device(s) 2410A coupled via I/O Hub 2407. In at least one embodiment, parallel processor(s) 2412 can also include a display controller and display interface (not shown) to enable a direct connection to one or more display device(s) 2410B.
In at least one embodiment, a system storage unit 2414 can connect to I/O hub 2407 to provide a storage mechanism for computing system 2400. In at least one embodiment, an I/O switch 2416 can be used to provide an interface mechanism to enable connections between I/O hub 2407 and other components, such as a network adapter 2418 and/or a wireless network adapter 2419 that may be integrated into platform, and various other devices that can be added via one or more add-in device(s) 2420. In at least one embodiment, network adapter 2418 can be an Ethernet adapter or another wired network adapter. In at least one embodiment, wireless network adapter 2419 can include one or more of a Wi-Fi, Bluetooth, near field communication (NFC), or other network device that includes one or more wireless radios.
In at least one embodiment, computing system 2400 can include other components not explicitly shown, including USB or other port connections, optical storage drives, video capture devices, and like, may also be connected to I/O hub 2407. In at least one embodiment, communication paths interconnecting various components in
In at least one embodiment, parallel processor(s) 2412 incorporate circuitry optimized for graphics and video processing, including, for example, video output circuitry, and constitutes a graphics processing unit (GPU). In at least one embodiment, parallel processor(s) 2412 incorporate circuitry optimized for general purpose processing. In at least embodiment, components of computing system 2400 may be integrated with one or more other system elements on a single integrated circuit. For example, in at least one embodiment, parallel processor(s) 2412, memory hub 2405, processor(s) 2402, and I/O hub 2407 can be integrated into a system on chip (SoC) integrated circuit. In at least one embodiment, components of computing system 2400 can be integrated into a single package to form a system in package (SIP) configuration. In at least one embodiment, at least a portion of components of computing system 2400 can be integrated into a multi-chip module (MCM), which can be interconnected with other multi-chip modules into a modular computing system.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, parallel processor 2500 includes a parallel processing unit 2502. In at least one embodiment, parallel processing unit 2502 includes an I/O unit 2504 that enables communication with other devices, including other instances of parallel processing unit 2502. In at least one embodiment, I/O unit 2504 may be directly connected to other devices. In at least one embodiment, I/O unit 2504 connects with other devices via use of a hub or switch interface, such as a memory hub 2505. In at least one embodiment, connections between memory hub 2505 and I/O unit 2504 form a communication link 2513. In at least one embodiment, I/O unit 2504 connects with a host interface 2506 and a memory crossbar 2516, where host interface 2506 receives commands directed to performing processing operations and memory crossbar 2516 receives commands directed to performing memory operations.
In at least one embodiment, when host interface 2506 receives a command buffer via I/O unit 2504, host interface 2506 can direct work operations to perform those commands to a front end 2508. In at least one embodiment, front end 2508 couples with a scheduler 2510, which is configured to distribute commands or other work items to a processing cluster array 2512. In at least one embodiment, scheduler 2510 ensures that processing cluster array 2512 is properly configured and in a valid state before tasks are distributed to a cluster of processing cluster array 2512. In at least one embodiment, scheduler 2510 is implemented via firmware logic executing on a microcontroller. In at least one embodiment, microcontroller implemented scheduler 2510 is configurable to perform complex scheduling and work distribution operations at coarse and fine granularity, enabling rapid preemption and context switching of threads executing on processing array 2512. In at least one embodiment, host software can prove workloads for scheduling on processing cluster array 2512 via one of multiple graphics processing paths. In at least one embodiment, workloads can then be automatically distributed across processing array cluster 2512 by scheduler 2510 logic within a microcontroller including scheduler 2510.
In at least one embodiment, processing cluster array 2512 can include up to “N” processing clusters (e.g., cluster 2514A, cluster 2514B, through cluster 2514N), where “N” represents a positive integer (which may be a different integer “N” than used in other figures). In at least one embodiment, each cluster 2514A-2514N of processing cluster array 2512 can execute a large number of concurrent threads. In at least one embodiment, scheduler 2510 can allocate work to clusters 2514A-2514N of processing cluster array 2512 using various scheduling and/or work distribution algorithms, which may vary depending on workload arising for each type of program or computation. In at least one embodiment, scheduling can be handled dynamically by scheduler 2510, or can be assisted in part by compiler logic during compilation of program logic configured for execution by processing cluster array 2512. In at least one embodiment, different clusters 2514A-2514N of processing cluster array 2512 can be allocated for processing different types of programs or for performing different types of computations.
In at least one embodiment, processing cluster array 2512 can be configured to perform various types of parallel processing operations. In at least one embodiment, processing cluster array 2512 is configured to perform general-purpose parallel compute operations. For example, in at least one embodiment, processing cluster array 2512 can include logic to execute processing tasks including filtering of video and/or audio data, performing modeling operations, including physics operations, and performing data transformations.
In at least one embodiment, processing cluster array 2512 is configured to perform parallel graphics processing operations. In at least one embodiment, processing cluster array 2512 can include additional logic to support execution of such graphics processing operations, including but not limited to, texture sampling logic to perform texture operations, as well as tessellation logic and other vertex processing logic. In at least one embodiment, processing cluster array 2512 can be configured to execute graphics processing related shader programs such as, but not limited to, vertex shaders, tessellation shaders, geometry shaders, and pixel shaders. In at least one embodiment, parallel processing unit 2502 can transfer data from system memory via I/O unit 2504 for processing. In at least one embodiment, during processing, transferred data can be stored to on-chip memory (e.g., parallel processor memory 2522) during processing, then written back to system memory.
In at least one embodiment, when parallel processing unit 2502 is used to perform graphics processing, scheduler 2510 can be configured to divide a processing workload into approximately equal sized tasks, to better enable distribution of graphics processing operations to multiple clusters 2514A-2514N of processing cluster array 2512. In at least one embodiment, portions of processing cluster array 2512 can be configured to perform different types of processing. For example, in at least one embodiment, a first portion may be configured to perform vertex shading and topology generation, a second portion may be configured to perform tessellation and geometry shading, and a third portion may be configured to perform pixel shading or other screen space operations, to produce a rendered image for display. In at least one embodiment, intermediate data produced by one or more of clusters 2514A-2514N may be stored in buffers to allow intermediate data to be transmitted between clusters 2514A-2514N for further processing.
In at least one embodiment, processing cluster array 2512 can receive processing tasks to be executed via scheduler 2510, which receives commands defining processing tasks from front end 2508. In at least one embodiment, processing tasks can include indices of data to be processed, e.g., surface (patch) data, primitive data, vertex data, and/or pixel data, as well as state parameters and commands defining how data is to be processed (e.g., what program is to be executed). In at least one embodiment, scheduler 2510 may be configured to fetch indices corresponding to tasks or may receive indices from front end 2508. In at least one embodiment, front end 2508 can be configured to ensure processing cluster array 2512 is configured to a valid state before a workload specified by incoming command buffers (e.g., batch-buffers, push buffers, etc.) is initiated.
In at least one embodiment, each of one or more instances of parallel processing unit 2502 can couple with a parallel processor memory 2522. In at least one embodiment, parallel processor memory 2522 can be accessed via memory crossbar 2516, which can receive memory requests from processing cluster array 2512 as well as I/O unit 2504. In at least one embodiment, memory crossbar 2516 can access parallel processor memory 2522 via a memory interface 2518. In at least one embodiment, memory interface 2518 can include multiple partition units (e.g., partition unit 2520A, partition unit 2520B, through partition unit 2520N) that can each couple to a portion (e.g., memory unit) of parallel processor memory 2522. In at least one embodiment, a number of partition units 2520A-2520N is configured to be equal to a number of memory units, such that a first partition unit 2520A has a corresponding first memory unit 2524A, a second partition unit 2520B has a corresponding memory unit 2524B, and an N-th partition unit 2520N has a corresponding N-th memory unit 2524N. In at least one embodiment, a number of partition units 2520A-2520N may not be equal to a number of memory units.
In at least one embodiment, memory units 2524A-2524N can include various types of memory devices, including dynamic random access memory (DRAM) or graphics random access memory, such as synchronous graphics random access memory (SGRAM), including graphics double data rate (GDDR) memory. In at least one embodiment, memory units 2524A-2524N may also include 3D stacked memory, including but not limited to high bandwidth memory (HBM). In at least one embodiment, render targets, such as frame buffers or texture maps may be stored across memory units 2524A-2524N, allowing partition units 2520A-2520N to write portions of each render target in parallel to efficiently use available bandwidth of parallel processor memory 2522. In at least one embodiment, a local instance of parallel processor memory 2522 may be excluded in favor of a unified memory design that utilizes system memory in conjunction with local cache memory.
In at least one embodiment, any one of clusters 2514A-2514N of processing cluster array 2512 can process data that will be written to any of memory units 2524A-2524N within parallel processor memory 2522. In at least one embodiment, memory crossbar 2516 can be configured to transfer an output of each cluster 2514A-2514N to any partition unit 2520A-2520N or to another cluster 2514A-2514N, which can perform additional processing operations on an output. In at least one embodiment, each cluster 2514A-2514N can communicate with memory interface 2518 through memory crossbar 2516 to read from or write to various external memory devices. In at least one embodiment, memory crossbar 2516 has a connection to memory interface 2518 to communicate with I/O unit 2504, as well as a connection to a local instance of parallel processor memory 2522, enabling processing units within different processing clusters 2514A-2514N to communicate with system memory or other memory that is not local to parallel processing unit 2502. In at least one embodiment, memory crossbar 2516 can use virtual channels to separate traffic streams between clusters 2514A-2514N and partition units 2520A-2520N.
In at least one embodiment, multiple instances of parallel processing unit 2502 can be provided on a single add-in card, or multiple add-in cards can be interconnected. In at least one embodiment, different instances of parallel processing unit 2502 can be configured to interoperate even if different instances have different numbers of processing cores, different amounts of local parallel processor memory, and/or other configuration differences. For example, in at least one embodiment, some instances of parallel processing unit 2502 can include higher precision floating point units relative to other instances. In at least one embodiment, systems incorporating one or more instances of parallel processing unit 2502 or parallel processor 2500 can be implemented in a variety of configurations and form factors, including but not limited to desktop, laptop, or handheld personal computers, servers, workstations, game consoles, and/or embedded systems.
In at least one embodiment, ROP 2526 is a processing unit that performs raster operations such as stencil, z test, blending, etc. In at least one embodiment, ROP 2526 then outputs processed graphics data that is stored in graphics memory. In at least one embodiment, ROP 2526 includes compression logic to compress depth or color data that is written to memory and decompress depth or color data that is read from memory. In at least one embodiment, compression logic can be lossless compression logic that makes use of one or more of multiple compression algorithms. In at least one embodiment, a type of compression that is performed by ROP 2526 can vary based on statistical characteristics of data to be compressed. For example, in at least one embodiment, delta color compression is performed on depth and color data on a per-tile basis.
In at least one embodiment, ROP 2526 is included within each processing cluster (e.g., cluster 2514A-2514N of
In at least one embodiment, operation of processing cluster 2514 can be controlled via a pipeline manager 2532 that distributes processing tasks to SIMT parallel processors. In at least one embodiment, pipeline manager 2532 receives instructions from scheduler 2510 of
In at least one embodiment, each graphics multiprocessor 2534 within processing cluster 2514 can include an identical set of functional execution logic (e.g., arithmetic logic units, load-store units, etc.). In at least one embodiment, functional execution logic can be configured in a pipelined manner in which new instructions can be issued before previous instructions are complete. In at least one embodiment, functional execution logic supports a variety of operations including integer and floating point arithmetic, comparison operations, Boolean operations, bit-shifting, and computation of various algebraic functions. In at least one embodiment, same functional-unit hardware can be leveraged to perform different operations and any combination of functional units may be present.
In at least one embodiment, instructions transmitted to processing cluster 2514 constitute a thread. In at least one embodiment, a set of threads executing across a set of parallel processing engines is a thread group. In at least one embodiment, a thread group executes a common program on different input data. In at least one embodiment, each thread within a thread group can be assigned to a different processing engine within a graphics multiprocessor 2534. In at least one embodiment, a thread group may include fewer threads than a number of processing engines within graphics multiprocessor 2534. In at least one embodiment, when a thread group includes fewer threads than a number of processing engines, one or more of processing engines may be idle during cycles in which that thread group is being processed. In at least one embodiment, a thread group may also include more threads than a number of processing engines within graphics multiprocessor 2534. In at least one embodiment, when a thread group includes more threads than number of processing engines within graphics multiprocessor 2534, processing can be performed over consecutive clock cycles. In at least one embodiment, multiple thread groups can be executed concurrently on a graphics multiprocessor 2534.
In at least one embodiment, graphics multiprocessor 2534 includes an internal cache memory to perform load and store operations. In at least one embodiment, graphics multiprocessor 2534 can forego an internal cache and use a cache memory (e.g., L1 cache 2548) within processing cluster 2514. In at least one embodiment, each graphics multiprocessor 2534 also has access to L2 caches within partition units (e.g., partition units 2520A-2520N of
In at least one embodiment, each processing cluster 2514 may include an MMU 2545 (memory management unit) that is configured to map virtual addresses into physical addresses. In at least one embodiment, one or more instances of MMU 2545 may reside within memory interface 2518 of
In at least one embodiment, a processing cluster 2514 may be configured such that each graphics multiprocessor 2534 is coupled to a texture unit 2536 for performing texture mapping operations, e.g., determining texture sample positions, reading texture data, and filtering texture data. In at least one embodiment, texture data is read from an internal texture L1 cache (not shown) or from an L1 cache within graphics multiprocessor 2534 and is fetched from an L2 cache, local parallel processor memory, or system memory, as needed. In at least one embodiment, each graphics multiprocessor 2534 outputs processed tasks to data crossbar 2540 to provide processed task to another processing cluster 2514 for further processing or to store processed task in an L2 cache, local parallel processor memory, or system memory via memory crossbar 2516. In at least one embodiment, a preROP 2542 (pre-raster operations unit) is configured to receive data from graphics multiprocessor 2534, and direct data to ROP units, which may be located with partition units as described herein (e.g., partition units 2520A-2520N of
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, instruction cache 2552 receives a stream of instructions to execute from pipeline manager 2532. In at least one embodiment, instructions are cached in instruction cache 2552 and dispatched for execution by an instruction unit 2554. In at least one embodiment, instruction unit 2554 can dispatch instructions as thread groups (e.g., warps), with each thread of thread group assigned to a different execution unit within GPGPU cores 2562. In at least one embodiment, an instruction can access any of a local, shared, or global address space by specifying an address within a unified address space. In at least one embodiment, address mapping unit 2556 can be used to translate addresses in a unified address space into a distinct memory address that can be accessed by load/store units 2566.
In at least one embodiment, register file 2558 provides a set of registers for functional units of graphics multiprocessor 2534. In at least one embodiment, register file 2558 provides temporary storage for operands connected to data paths of functional units (e.g., GPGPU cores 2562, load/store units 2566) of graphics multiprocessor 2534. In at least one embodiment, register file 2558 is divided between each of functional units such that each functional unit is allocated a dedicated portion of register file 2558. In at least one embodiment, register file 2558 is divided between different warps being executed by graphics multiprocessor 2534.
In at least one embodiment, GPGPU cores 2562 can each include floating point units (FPUs) and/or integer arithmetic logic units (ALUs) that are used to execute instructions of graphics multiprocessor 2534. In at least one embodiment, GPGPU cores 2562 can be similar in architecture or can differ in architecture. In at least one embodiment, a first portion of GPGPU cores 2562 include a single precision FPU and an integer ALU while a second portion of GPGPU cores include a double precision FPU. In at least one embodiment, FPUs can implement IEEE 754-2008 standard floating point arithmetic or enable variable precision floating point arithmetic. In at least one embodiment, graphics multiprocessor 2534 can additionally include one or more fixed function or special function units to perform specific functions such as copy rectangle or pixel blending operations. In at least one embodiment, one or more of GPGPU cores 2562 can also include fixed or special function logic.
In at least one embodiment, GPGPU cores 2562 include SIMD logic capable of performing a single instruction on multiple sets of data. In at least one embodiment, GPGPU cores 2562 can physically execute SIMD4, SIMD8, and SIMD16 instructions and logically execute SIMD1, SIMD2, and SIMD32 instructions. In at least one embodiment, SIMD instructions for GPGPU cores can be generated at compile time by a shader compiler or automatically generated when executing programs written and compiled for single program multiple data (SPMD) or SIMT architectures. In at least one embodiment, multiple threads of a program configured for an SIMT execution model can executed via a single SIMD instruction. For example, in at least one embodiment, eight SIMT threads that perform same or similar operations can be executed in parallel via a single SIMD8 logic unit.
In at least one embodiment, memory and cache interconnect 2568 is an interconnect network that connects each functional unit of graphics multiprocessor 2534 to register file 2558 and to shared memory 2570. In at least one embodiment, memory and cache interconnect 2568 is a crossbar interconnect that allows load/store unit 2566 to implement load and store operations between shared memory 2570 and register file 2558. In at least one embodiment, register file 2558 can operate at a same frequency as GPGPU cores 2562, thus data transfer between GPGPU cores 2562 and register file 2558 can have very low latency. In at least one embodiment, shared memory 2570 can be used to enable communication between threads that execute on functional units within graphics multiprocessor 2534. In at least one embodiment, cache memory 2572 can be used as a data cache for example, to cache texture data communicated between functional units and texture unit 2536. In at least one embodiment, shared memory 2570 can also be used as a program managed cache. In at least one embodiment, threads executing on GPGPU cores 2562 can programmatically store data within shared memory in addition to automatically cached data that is stored within cache memory 2572.
In at least one embodiment, a parallel processor or GPGPU as described herein is communicatively coupled to host/processor cores to accelerate graphics operations, machine-learning operations, pattern analysis operations, and various general purpose GPU (GPGPU) functions. In at least one embodiment, a GPU may be communicatively coupled to host processor/cores over a bus or other interconnect (e.g., a high-speed interconnect such as PCIe or NVLink). In at least one embodiment, a GPU may be integrated on a package or chip as cores and communicatively coupled to cores over an internal processor bus/interconnect internal to a package or chip. In at least one embodiment, regardless a manner in which a GPU is connected, processor cores may allocate work to such GPU in a form of sequences of commands/instructions contained in a work descriptor. In at least one embodiment, that GPU then uses dedicated circuitry/logic for efficiently processing these commands/instructions.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, graphics processor 2700 receives batches of commands via ring interconnect 2702. In at least one embodiment, incoming commands are interpreted by a command streamer 2703 in pipeline front-end 2704. In at least one embodiment, graphics processor 2700 includes scalable execution logic to perform 3D geometry processing and media processing via graphics core(s) 2780A-2780N. In at least one embodiment, for 3D geometry processing commands, command streamer 2703 supplies commands to geometry pipeline 2736. In at least one embodiment, for at least some media processing commands, command streamer 2703 supplies commands to a video front end 2734, which couples with media engine 2737. In at least one embodiment, media engine 2737 includes a Video Quality Engine (VQE) 2730 for video and image post-processing and a multi-format encode/decode (MFX) 2733 engine to provide hardware-accelerated media data encoding and decoding. In at least one embodiment, geometry pipeline 2736 and media engine 2737 each generate execution threads for thread execution resources provided by at least one graphics core 2780.
In at least one embodiment, graphics processor 2700 includes scalable thread execution resources featuring graphics cores 2780A-2780N (which can be modular and are sometimes referred to as core slices), each having multiple sub-cores 2750A-50N, 2760A-2760N (sometimes referred to as core sub-slices). In at least one embodiment, graphics processor 2700 can have any number of graphics cores 2780A. In at least one embodiment, graphics processor 2700 includes a graphics core 2780A having at least a first sub-core 2750A and a second sub-core 2760A. In at least one embodiment, graphics processor 2700 is a low power processor with a single sub-core (e.g., 2750A). In at least one embodiment, graphics processor 2700 includes multiple graphics cores 2780A-2780N, each including a set of first sub-cores 2750A-2750N and a set of second sub-cores 2760A-2760N. In at least one embodiment, each sub-core in first sub-cores 2750A-2750N includes at least a first set of execution units 2752A-2752N and media/texture samplers 2754A-2754N. In at least one embodiment, each sub-core in second sub-cores 2760A-2760N includes at least a second set of execution units 2762A-2762N and samplers 2764A-2764N. In at least one embodiment, each sub-core 2750A-2750N, 2760A-2760N shares a set of shared resources 2770A-2770N. In at least one embodiment, shared resources include shared cache memory and pixel operation logic.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, processor 2800 includes an in-order front end (“front end”) 2801 to fetch instructions to be executed and prepare instructions to be used later in a processor pipeline. In at least one embodiment, front end 2801 may include several units. In at least one embodiment, an instruction prefetcher 2826 fetches instructions from memory and feeds instructions to an instruction decoder 2828 which in turn decodes or interprets instructions. For example, in at least one embodiment, instruction decoder 2828 decodes a received instruction into one or more operations called “micro-instructions” or “micro-operations” (also called “micro ops” or “uops”) that a machine may execute. In at least one embodiment, instruction decoder 2828 parses an instruction into an opcode and corresponding data and control fields that may be used by micro-architecture to perform operations in accordance with at least one embodiment. In at least one embodiment, a trace cache 2830 may assemble decoded uops into program ordered sequences or traces in a uop queue 2834 for execution. In at least one embodiment, when trace cache 2830 encounters a complex instruction, a microcode ROM 2832 provides uops needed to complete an operation.
In at least one embodiment, some instructions may be converted into a single micro-op, whereas others need several micro-ops to complete full operation. In at least one embodiment, if more than four micro-ops are needed to complete an instruction, instruction decoder 2828 may access microcode ROM 2832 to perform that instruction. In at least one embodiment, an instruction may be decoded into a small number of micro-ops for processing at instruction decoder 2828. In at least one embodiment, an instruction may be stored within microcode ROM 2832 should a number of micro-ops be needed to accomplish such operation. In at least one embodiment, trace cache 2830 refers to an entry point programmable logic array (“PLA”) to determine a correct micro-instruction pointer for reading microcode sequences to complete one or more instructions from microcode ROM 2832 in accordance with at least one embodiment. In at least one embodiment, after microcode ROM 2832 finishes sequencing micro-ops for an instruction, front end 2801 of a machine may resume fetching micro-ops from trace cache 2830.
In at least one embodiment, out-of-order execution engine (“out of order engine”) 2803 may prepare instructions for execution. In at least one embodiment, out-of-order execution logic has a number of buffers to smooth out and re-order flow of instructions to optimize performance as they go down a pipeline and get scheduled for execution. In at least one embodiment, out-of-order execution engine 2803 includes, without limitation, an allocator/register renamer 2840, a memory uop queue 2842, an integer/floating point uop queue 2844, a memory scheduler 2846, a fast scheduler 2802, a slow/general floating point scheduler (“slow/general FP scheduler”) 2804, and a simple floating point scheduler (“simple FP scheduler”) 2806. In at least one embodiment, fast schedule 2802, slow/general floating point scheduler 2804, and simple floating point scheduler 2806 are also collectively referred to herein as “uop schedulers 2802, 2804, 2806.” In at least one embodiment, allocator/register renamer 2840 allocates machine buffers and resources that each uop needs in order to execute. In at least one embodiment, allocator/register renamer 2840 renames logic registers onto entries in a register file. In at least one embodiment, allocator/register renamer 2840 also allocates an entry for each uop in one of two uop queues, memory uop queue 2842 for memory operations and integer/floating point uop queue 2844 for non-memory operations, in front of memory scheduler 2846 and uop schedulers 2802, 2804, 2806. In at least one embodiment, uop schedulers 2802, 2804, 2806, determine when a uop is ready to execute based on readiness of their dependent input register operand sources and availability of execution resources uops need to complete their operation. In at least one embodiment, fast scheduler 2802 may schedule on each half of a main clock cycle while slow/general floating point scheduler 2804 and simple floating point scheduler 2806 may schedule once per main processor clock cycle. In at least one embodiment, uop schedulers 2802, 2804, 2806 arbitrate for dispatch ports to schedule uops for execution.
In at least one embodiment, execution block 2811 includes, without limitation, an integer register file/bypass network 2808, a floating point register file/bypass network (“FP register file/bypass network”) 2810, address generation units (“AGUs”) 2812 and 2814, fast Arithmetic Logic Units (ALUs) (“fast ALUs”) 2816 and 2818, a slow Arithmetic Logic Unit (“slow ALU”) 2820, a floating point ALU (“FP”) 2822, and a floating point move unit (“FP move”) 2824. In at least one embodiment, integer register file/bypass network 2808 and floating point register file/bypass network 2810 are also referred to herein as “register files 2808, 2810.” In at least one embodiment, AGUSs 2812 and 2814, fast ALUs 2816 and 2818, slow ALU 2820, floating point ALU 2822, and floating point move unit 2824 are also referred to herein as “execution units 2812, 2814, 2816, 2818, 2820, 2822, and 2824.” In at least one embodiment, execution block 2811 may include, without limitation, any number (including zero) and type of register files, bypass networks, address generation units, and execution units, in any combination.
In at least one embodiment, register networks 2808, 2810 may be arranged between uop schedulers 2802, 2804, 2806, and execution units 2812, 2814, 2816, 2818, 2820, 2822, and 2824. In at least one embodiment, integer register file/bypass network 2808 performs integer operations. In at least one embodiment, floating point register file/bypass network 2810 performs floating point operations. In at least one embodiment, each of register networks 2808, 2810 may include, without limitation, a bypass network that may bypass or forward just completed results that have not yet been written into a register file to new dependent uops. In at least one embodiment, register networks 2808, 2810 may communicate data with each other. In at least one embodiment, integer register file/bypass network 2808 may include, without limitation, two separate register files, one register file for a low-order thirty-two bits of data and a second register file for a high order thirty-two bits of data. In at least one embodiment, floating point register file/bypass network 2810 may include, without limitation, 128-bit wide entries because floating point instructions typically have operands from 64 to 128 bits in width.
In at least one embodiment, execution units 2812, 2814, 2816, 2818, 2820, 2822, 2824 may execute instructions. In at least one embodiment, register networks 2808, 2810 store integer and floating point data operand values that micro-instructions need to execute. In at least one embodiment, processor 2800 may include, without limitation, any number and combination of execution units 2812, 2814, 2816, 2818, 2820, 2822, 2824. In at least one embodiment, floating point ALU 2822 and floating point move unit 2824, may execute floating point, MMX, SIMD, AVX and SSE, or other operations, including specialized machine learning instructions. In at least one embodiment, floating point ALU 2822 may include, without limitation, a 64-bit by 64-bit floating point divider to execute divide, square root, and remainder micro ops. In at least one embodiment, instructions involving a floating point value may be handled with floating point hardware. In at least one embodiment, ALU operations may be passed to fast ALUs 2816, 2818. In at least one embodiment, fast ALUS 2816, 2818 may execute fast operations with an effective latency of half a clock cycle. In at least one embodiment, most complex integer operations go to slow ALU 2820 as slow ALU 2820 may include, without limitation, integer execution hardware for long-latency type of operations, such as a multiplier, shifts, flag logic, and branch processing. In at least one embodiment, memory load/store operations may be executed by AGUs 2812, 2814. In at least one embodiment, fast ALU 2816, fast ALU 2818, and slow ALU 2820 may perform integer operations on 64-bit data operands. In at least one embodiment, fast ALU 2816, fast ALU 2818, and slow ALU 2820 may be implemented to support a variety of data bit sizes including sixteen, thirty-two, 128, 256, etc. In at least one embodiment, floating point ALU 2822 and floating point move unit 2824 may be implemented to support a range of operands having bits of various widths, such as 128-bit wide packed data operands in conjunction with SIMD and multimedia instructions.
In at least one embodiment, uop schedulers 2802, 2804, 2806 dispatch dependent operations before a parent load has finished executing. In at least one embodiment, as uops may be speculatively scheduled and executed in processor 2800, processor 2800 may also include logic to handle memory misses. In at least one embodiment, if a data load misses in a data cache, there may be dependent operations in flight in a pipeline that have left a scheduler with temporarily incorrect data. In at least one embodiment, a replay mechanism tracks and re-executes instructions that use incorrect data. In at least one embodiment, dependent operations might need to be replayed and independent ones may be allowed to complete. In at least one embodiment, schedulers and a replay mechanism of at least one embodiment of a processor may also be designed to catch instruction sequences for text string comparison operations.
In at least one embodiment, “registers” may refer to on-board processor storage locations that may be used as part of instructions to identify operands. In at least one embodiment, registers may be those that may be usable from outside of a processor (from a programmer's perspective). In at least one embodiment, registers might not be limited to a particular type of circuit. Rather, in at least one embodiment, a register may store data, provide data, and perform functions described herein. In at least one embodiment, registers described herein may be implemented by circuitry within a processor using any number of different techniques, such as dedicated physical registers, dynamically allocated physical registers using register renaming, combinations of dedicated and dynamically allocated physical registers, etc. In at least one embodiment, integer registers store 32-bit integer data. A register file of at least one embodiment also contains eight multimedia SIMD registers for packed data.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, processing clusters 2910 may perform deep learning operations, including inference or prediction operations based on weight parameters calculated one or more training techniques, including those described herein. In at least one embodiment, each processing cluster 2910 may include, without limitation, any number and type of processors. In at least one embodiment, deep learning application processor 2900 may include any number and type of processing clusters 2900. In at least one embodiment, Inter-Chip Links 2920 are bi-directional. In at least one embodiment, Inter-Chip Links 2920 and Inter-Chip Controllers 2930 enable multiple deep learning application processors 2900 to exchange information, including activation information resulting from performing one or more machine learning algorithms embodied in one or more neural networks. In at least one embodiment, deep learning application processor 2900 may include any number (including zero) and type of ICLs 2920 and ICCs 2930.
In at least one embodiment, HBM2s 2940 provide a total of 32 Gigabytes (GB) of memory. In at least one embodiment, HBM2 2940(i) is associated with both memory controller 2942(i) and HBM PHY 2944(i) where “i” is an arbitrary integer. In at least one embodiment, any number of HBM2s 2940 may provide any type and total amount of high bandwidth memory and may be associated with any number (including zero) and type of memory controllers 2942 and HBM PHYs 2944. In at least one embodiment, SPI, I2C, GPIO 2960, PCIe Controller and DMA 2970, and/or PCIe 2980 may be replaced with any number and type of blocks that enable any number and type of communication standards in any technically feasible fashion.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, neurons 3002 and synapses 3008 may be interconnected such that neuromorphic processor 3000 operates to process or analyze information received by neuromorphic processor 3000. In at least one embodiment, neurons 3002 may transmit an output pulse (or “fire” or “spike”) when inputs received through neuron input 3004 exceed a threshold. In at least one embodiment, neurons 3002 may sum or integrate signals received at neuron inputs 3004. For example, in at least one embodiment, neurons 3002 may be implemented as leaky integrate-and-fire neurons, wherein if a sum (referred to as a “membrane potential”) exceeds a threshold value, neuron 3002 may generate an output (or “fire”) using a transfer function such as a sigmoid or threshold function. In at least one embodiment, a leaky integrate-and-fire neuron may sum signals received at neuron inputs 3004 into a membrane potential and may also apply a decay factor (or leak) to reduce a membrane potential. In at least one embodiment, a leaky integrate-and-fire neuron may fire if multiple input signals are received at neuron inputs 3004 rapidly enough to exceed a threshold value (e.g., before a membrane potential decays too low to fire). In at least one embodiment, neurons 3002 may be implemented using circuits or logic that receive inputs, integrate inputs into a membrane potential, and decay a membrane potential. In at least one embodiment, inputs may be averaged, or any other suitable transfer function may be used. Furthermore, in at least one embodiment, neurons 3002 may include, without limitation, comparator circuits or logic that generate an output spike at neuron output 3006 when result of applying a transfer function to neuron input 3004 exceeds a threshold. In at least one embodiment, once neuron 3002 fires, it may disregard previously received input information by, for example, resetting a membrane potential to 0 or another suitable default value. In at least one embodiment, once membrane potential is reset to 0, neuron 3002 may resume normal operation after a suitable period of time (or refractory period).
In at least one embodiment, neurons 3002 may be interconnected through synapses 3008. In at least one embodiment, synapses 3008 may operate to transmit signals from an output of a first neuron 3002 to an input of a second neuron 3002. In at least one embodiment, neurons 3002 may transmit information over more than one instance of synapse 3008. In at least one embodiment, one or more instances of neuron output 3006 may be connected, via an instance of synapse 3008, to an instance of neuron input 3004 in same neuron 3002. In at least one embodiment, an instance of neuron 3002 generating an output to be transmitted over an instance of synapse 3008 may be referred to as a “pre-synaptic neuron” with respect to that instance of synapse 3008. In at least one embodiment, an instance of neuron 3002 receiving an input transmitted over an instance of synapse 3008 may be referred to as a “post-synaptic neuron” with respect to that instance of synapse 3008. Because an instance of neuron 3002 may receive inputs from one or more instances of synapse 3008, and may also transmit outputs over one or more instances of synapse 3008, a single instance of neuron 3002 may therefore be both a “pre-synaptic neuron” and “post-synaptic neuron,” with respect to various instances of synapses 3008, in at least one embodiment.
In at least one embodiment, neurons 3002 may be organized into one or more layers. In at least one embodiment, each instance of neuron 3002 may have one neuron output 3006 that may fan out through one or more synapses 3008 to one or more neuron inputs 3004. In at least one embodiment, neuron outputs 3006 of neurons 3002 in a first layer 3010 may be connected to neuron inputs 3004 of neurons 3002 in a second layer 3012. In at least one embodiment, layer 3010 may be referred to as a “feed-forward layer.” In at least one embodiment, each instance of neuron 3002 in an instance of first layer 3010 may fan out to each instance of neuron 3002 in second layer 3012. In at least one embodiment, first layer 3010 may be referred to as a “fully connected feed-forward layer.” In at least one embodiment, each instance of neuron 3002 in an instance of second layer 3012 may fan out to fewer than all instances of neuron 3002 in a third layer 3014. In at least one embodiment, second layer 3012 may be referred to as a “sparsely connected feed-forward layer.” In at least one embodiment, neurons 3002 in second layer 3012 may fan out to neurons 3002 in multiple other layers, including to neurons 3002 also in second layer 3012. In at least one embodiment, second layer 3012 may be referred to as a “recurrent layer.” In at least one embodiment, neuromorphic processor 3000 may include, without limitation, any suitable combination of recurrent layers and feed-forward layers, including, without limitation, both sparsely connected feed-forward layers and fully connected feed-forward layers.
In at least one embodiment, neuromorphic processor 3000 may include, without limitation, a reconfigurable interconnect architecture or dedicated hard-wired interconnects to connect synapse 3008 to neurons 3002. In at least one embodiment, neuromorphic processor 3000 may include, without limitation, circuitry or logic that allows synapses to be allocated to different neurons 3002 as needed based on neural network topology and neuron fan-in/out. For example, in at least one embodiment, synapses 3008 may be connected to neurons 3002 using an interconnect fabric, such as network-on-chip, or with dedicated connections. In at least one embodiment, synapse interconnections and components thereof may be implemented using circuitry or logic.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, system 3100 can include, or be incorporated within a server-based gaming platform, a game console, including a game and media console, a mobile gaming console, a handheld game console, or an online game console. In at least one embodiment, system 3100 is a mobile phone, a smart phone, a tablet computing device or a mobile Internet device. In at least one embodiment, processing system 3100 can also include, couple with, or be integrated within a wearable device, such as a smart watch wearable device, a smart eyewear device, an augmented reality device, or a virtual reality device. In at least one embodiment, processing system 3100 is a television or set top box device having one or more processors 3102 and a graphical interface generated by one or more graphics processors 3108.
In at least one embodiment, one or more processors 3102 each include one or more processor cores 3107 to process instructions which, when executed, perform operations for system and user software. In at least one embodiment, each of one or more processor cores 3107 is configured to process a specific instruction sequence 3109. In at least one embodiment, instruction sequence 3109 may facilitate Complex Instruction Set Computing (CISC), Reduced Instruction Set Computing (RISC), or computing via a Very Long Instruction Word (VLIW). In at least one embodiment, processor cores 3107 may each process a different instruction sequence 3109, which may include instructions to facilitate emulation of other instruction sequences. In at least one embodiment, processor core 3107 may also include other processing devices, such a Digital Signal Processor (DSP).
In at least one embodiment, processor 3102 includes a cache memory 3104. In at least one embodiment, processor 3102 can have a single internal cache or multiple levels of internal cache. In at least one embodiment, cache memory is shared among various components of processor 3102. In at least one embodiment, processor 3102 also uses an external cache (e.g., a Level-3 (L3) cache or Last Level Cache (LLC)) (not shown), which may be shared among processor cores 3107 using known cache coherency techniques. In at least one embodiment, a register file 3106 is additionally included in processor 3102, which may include different types of registers for storing different types of data (e.g., integer registers, floating point registers, status registers, and an instruction pointer register). In at least one embodiment, register file 3106 may include general-purpose registers or other registers.
In at least one embodiment, one or more processor(s) 3102 are coupled with one or more interface bus(es) 3110 to transmit communication signals such as address, data, or control signals between processor 3102 and other components in system 3100. In at least one embodiment, interface bus 3110 can be a processor bus, such as a version of a Direct Media Interface (DMI) bus. In at least one embodiment, interface bus 3110 is not limited to a DMI bus, and may include one or more Peripheral Component Interconnect buses (e.g., PCI, PCI Express), memory busses, or other types of interface busses. In at least one embodiment processor(s) 3102 include an integrated memory controller 3116 and a platform controller hub 3130. In at least one embodiment, memory controller 3116 facilitates communication between a memory device and other components of system 3100, while platform controller hub (PCH) 3130 provides connections to I/O devices via a local I/O bus.
In at least one embodiment, a memory device 3120 can be a dynamic random access memory (DRAM) device, a static random access memory (SRAM) device, flash memory device, phase-change memory device, or some other memory device having suitable performance to serve as process memory. In at least one embodiment, memory device 3120 can operate as system memory for system 3100, to store data 3122 and instructions 3121 for use when one or more processors 3102 executes an application or process. In at least one embodiment, memory controller 3116 also couples with an optional external graphics processor 3112, which may communicate with one or more graphics processors 3108 in processors 3102 to perform graphics and media operations. In at least one embodiment, a display device 3111 can connect to processor(s) 3102. In at least one embodiment, display device 3111 can include one or more of an internal display device, as in a mobile electronic device or a laptop device, or an external display device attached via a display interface (e.g., DisplayPort, etc.). In at least one embodiment, display device 3111 can include a head mounted display (HMD) such as a stereoscopic display device for use in virtual reality (VR) applications or augmented reality (AR) applications.
In at least one embodiment, platform controller hub 3130 enables peripherals to connect to memory device 3120 and processor 3102 via a high-speed I/O bus. In at least one embodiment, I/O peripherals include, but are not limited to, an audio controller 3146, a network controller 3134, a firmware interface 3128, a wireless transceiver 3126, touch sensors 3125, a data storage device 3124 (e.g., hard disk drive, flash memory, etc.). In at least one embodiment, data storage device 3124 can connect via a storage interface (e.g., SATA) or via a peripheral bus, such as a Peripheral Component Interconnect bus (e.g., PCI, PCI Express). In at least one embodiment, touch sensors 3125 can include touch screen sensors, pressure sensors, or fingerprint sensors. In at least one embodiment, wireless transceiver 3126 can be a Wi-Fi transceiver, a Bluetooth transceiver, or a mobile network transceiver such as a 3G, 4G, or Long Term Evolution (LTE) transceiver. In at least one embodiment, firmware interface 3128 enables communication with system firmware, and can be, for example, a unified extensible firmware interface (UEFI). In at least one embodiment, network controller 3134 can enable a network connection to a wired network. In at least one embodiment, a high-performance network controller (not shown) couples with interface bus 3110. In at least one embodiment, audio controller 3146 is a multi-channel high definition audio controller. In at least one embodiment, system 3100 includes an optional legacy I/O controller 3140 for coupling legacy (e.g., Personal System 2 (PS/2)) devices to system 3100. In at least one embodiment, platform controller hub 3130 can also connect to one or more Universal Serial Bus (USB) controllers 3142 connect input devices, such as keyboard and mouse 3143 combinations, a camera 3144, or other USB input devices.
In at least one embodiment, an instance of memory controller 3116 and platform controller hub 3130 may be integrated into a discreet external graphics processor, such as external graphics processor 3112. In at least one embodiment, platform controller hub 3130 and/or memory controller 3116 may be external to one or more processor(s) 3102. For example, in at least one embodiment, system 3100 can include an external memory controller 3116 and platform controller hub 3130, which may be configured as a memory controller hub and peripheral controller hub within a system chipset that is in communication with processor(s) 3102.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, internal cache units 3204A-3204N and shared cache units 3206 represent a cache memory hierarchy within processor 3200. In at least one embodiment, cache memory units 3204A-3204N may include at least one level of instruction and data cache within each processor core and one or more levels of shared mid-level cache, such as a Level 2 (L2), Level 3 (L3), Level 4 (L4), or other levels of cache, where a highest level of cache before external memory is classified as an LLC. In at least one embodiment, cache coherency logic maintains coherency between various cache units 3206 and 3204A-3204N.
In at least one embodiment, processor 3200 may also include a set of one or more bus controller units 3216 and a system agent core 3210. In at least one embodiment, bus controller units 3216 manage a set of peripheral buses, such as one or more PCI or PCI express busses. In at least one embodiment, system agent core 3210 provides management functionality for various processor components. In at least one embodiment, system agent core 3210 includes one or more integrated memory controllers 3214 to manage access to various external memory devices (not shown).
In at least one embodiment, one or more of processor cores 3202A-3202N include support for simultaneous multi-threading. In at least one embodiment, system agent core 3210 includes components for coordinating and operating cores 3202A-3202N during multi-threaded processing. In at least one embodiment, system agent core 3210 may additionally include a power control unit (PCU), which includes logic and components to regulate one or more power states of processor cores 3202A-3202N and graphics processor 3208.
In at least one embodiment, processor 3200 additionally includes graphics processor 3208 to execute graphics processing operations. In at least one embodiment, graphics processor 3208 couples with shared cache units 3206, and system agent core 3210, including one or more integrated memory controllers 3214. In at least one embodiment, system agent core 3210 also includes a display controller 3211 to drive graphics processor output to one or more coupled displays. In at least one embodiment, display controller 3211 may also be a separate module coupled with graphics processor 3208 via at least one interconnect, or may be integrated within graphics processor 3208.
In at least one embodiment, a ring-based interconnect unit 3212 is used to couple internal components of processor 3200. In at least one embodiment, an alternative interconnect unit may be used, such as a point-to-point interconnect, a switched interconnect, or other techniques. In at least one embodiment, graphics processor 3208 couples with ring interconnect 3212 via an I/O link 3213.
In at least one embodiment, I/O link 3213 represents at least one of multiple varieties of I/O interconnects, including an on package I/O interconnect which facilitates communication between various processor components and a high-performance embedded memory module 3218, such as an eDRAM module. In at least one embodiment, each of processor cores 3202A-3202N and graphics processor 3208 use embedded memory module 3218 as a shared Last Level Cache.
In at least one embodiment, processor cores 3202A-3202N are homogeneous cores executing a common instruction set architecture. In at least one embodiment, processor cores 3202A-3202N are heterogeneous in terms of instruction set architecture (ISA), where one or more of processor cores 3202A-3202N execute a common instruction set, while one or more other cores of processor cores 3202A-3202N executes a subset of a common instruction set or a different instruction set. In at least one embodiment, processor cores 3202A-3202N are heterogeneous in terms of microarchitecture, where one or more cores having a relatively higher power consumption couple with one or more power cores having a lower power consumption. In at least one embodiment, processor 3200 can be implemented on one or more chips or as an SoC integrated circuit.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, graphics processor 3300 also includes a display controller 3302 to drive display output data to a display device 3320. In at least one embodiment, display controller 3302 includes hardware for one or more overlay planes for display device 3320 and composition of multiple layers of video or user interface elements. In at least one embodiment, display device 3320 can be an internal or external display device. In at least one embodiment, display device 3320 is a head mounted display device, such as a virtual reality (VR) display device or an augmented reality (AR) display device. In at least one embodiment, graphics processor 3300 includes a video codec engine 3306 to encode, decode, or transcode media to, from, or between one or more media encoding formats, including, but not limited to Moving Picture Experts Group (MPEG) formats such as MPEG-2, Advanced Video Coding (AVC) formats such as H.264/MPEG-4 AVC, as well as Society of Motion Picture & Television Engineers (SMPTE) 421M/VC-1, and Joint Photographic Experts Group (JPEG) formats such as JPEG, and Motion JPEG (MJPEG) formats.
In at least one embodiment, graphics processor 3300 includes a block image transfer (BLIT) engine 3304 to perform two-dimensional (2D) rasterizer operations including, for example, bit-boundary block transfers. However, in at least one embodiment, 2D graphics operations are performed using one or more components of a graphics processing engine (GPE) 3310. In at least one embodiment, GPE 3310 is a compute engine for performing graphics operations, including three-dimensional (3D) graphics operations and media operations.
In at least one embodiment, GPE 3310 includes a 3D pipeline 3312 for performing 3D operations, such as rendering three-dimensional images and scenes using processing functions that act upon 3D primitive shapes (e.g., rectangle, triangle, etc.). In at least one embodiment, 3D pipeline 3312 includes programmable and fixed function elements that perform various tasks and/or spawn execution threads to a 3D/Media sub-system 3315. While 3D pipeline 3312 can be used to perform media operations, in at least one embodiment, GPE 3310 also includes a media pipeline 3316 that is used to perform media operations, such as video post-processing and image enhancement.
In at least one embodiment, media pipeline 3316 includes fixed function or programmable logic units to perform one or more specialized media operations, such as video decode acceleration, video de-interlacing, and video encode acceleration in place of, or on behalf of, video codec engine 3306. In at least one embodiment, media pipeline 3316 additionally includes a thread spawning unit to spawn threads for execution on 3D/Media sub-system 3315. In at least one embodiment, spawned threads perform computations for media operations on one or more graphics execution units included in 3D/Media sub-system 3315.
In at least one embodiment, 3D/Media subsystem 3315 includes logic for executing threads spawned by 3D pipeline 3312 and media pipeline 3316. In at least one embodiment, 3D pipeline 3312 and media pipeline 3316 send thread execution requests to 3D/Media subsystem 3315, which includes thread dispatch logic for arbitrating and dispatching various requests to available thread execution resources. In at least one embodiment, execution resources include an array of graphics execution units to process 3D and media threads. In at least one embodiment, 3D/Media subsystem 3315 includes one or more internal caches for thread instructions and data. In at least one embodiment, subsystem 3315 also includes shared memory, including registers and addressable memory, to share data between threads and to store output data.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, GPE 3410 is coupled to or includes a command streamer 3403, which provides a command stream to a 3D pipeline 3412 and/or media pipeline 3416. In at least one embodiment, command streamer 3403 is coupled to memory, which can be system memory, or one or more of internal cache memory and shared cache memory. In at least one embodiment, command streamer 3403 receives commands from memory and sends commands to 3D pipeline 3412 and/or media pipeline 3416. In at least one embodiment, commands are instructions, primitives, or micro-operations fetched from a ring buffer, which stores commands for 3D pipeline 3412 and media pipeline 3416. In at least one embodiment, a ring buffer can additionally include batch command buffers storing batches of multiple commands. In at least one embodiment, commands for 3D pipeline 3412 can also include references to data stored in memory, such as, but not limited to, vertex and geometry data for 3D pipeline 3412 and/or image data and memory objects for media pipeline 3416. In at least one embodiment, 3D pipeline 3412 and media pipeline 3416 process commands and data by performing operations or by dispatching one or more execution threads to a graphics core array 3414. In at least one embodiment, graphics core array 3414 includes one or more blocks of graphics cores (e.g., graphics core(s) 3415A, graphics core(s) 3415B), each block including one or more graphics cores. In at least one embodiment, each graphics core includes a set of graphics execution resources that includes general-purpose and graphics specific execution logic to perform graphics and compute operations, as well as fixed function texture processing and/or machine learning and artificial intelligence acceleration logic, including inference and/or training logic 1215 in
In at least one embodiment, 3D pipeline 3412 includes fixed function and programmable logic to process one or more shader programs, such as vertex shaders, geometry shaders, pixel shaders, fragment shaders, compute shaders, or other shader programs, by processing instructions and dispatching execution threads to graphics core array 3414. In at least one embodiment, graphics core array 3414 provides a unified block of execution resources for use in processing shader programs. In at least one embodiment, a multi-purpose execution logic (e.g., execution units) within graphics core(s) 3415A-3415B of graphic core array 3414 includes support for various 3D API shader languages and can execute multiple simultaneous execution threads associated with multiple shaders.
In at least one embodiment, graphics core array 3414 also includes execution logic to perform media functions, such as video and/or image processing. In at least one embodiment, execution units additionally include general-purpose logic that is programmable to perform parallel general-purpose computational operations, in addition to graphics processing operations.
In at least one embodiment, output data generated by threads executing on graphics core array 3414 can output data to memory in a unified return buffer (URB) 3418. In at least one embodiment, URB 3418 can store data for multiple threads. In at least one embodiment, URB 3418 may be used to send data between different threads executing on graphics core array 3414. In at least one embodiment, URB 3418 may additionally be used for synchronization between threads on graphics core array 3414 and fixed function logic within shared function logic 3420.
In at least one embodiment, graphics core array 3414 is scalable, such that graphics core array 3414 includes a variable number of graphics cores, each having a variable number of execution units based on a target power and performance level of GPE 3410. In at least one embodiment, execution resources are dynamically scalable, such that execution resources may be enabled or disabled as needed.
In at least one embodiment, graphics core array 3414 is coupled to shared function logic 3420 that includes multiple resources that are shared between graphics cores in graphics core array 3414. In at least one embodiment, shared functions performed by shared function logic 3420 are embodied in hardware logic units that provide specialized supplemental functionality to graphics core array 3414. In at least one embodiment, shared function logic 3420 includes but is not limited to a sampler unit 3421, a math unit 3422, and inter-thread communication (ITC) logic 3423. In at least one embodiment, one or more cache(s) 3425 are included in, or coupled to, shared function logic 3420.
In at least one embodiment, a shared function is used if demand for a specialized function is insufficient for inclusion within graphics core array 3414. In at least one embodiment, a single instantiation of a specialized function is used in shared function logic 3420 and shared among other execution resources within graphics core array 3414. In at least one embodiment, specific shared functions within shared function logic 3420 that are used extensively by graphics core array 3414 may be included within shared function logic 3426 within graphics core array 3414. In at least one embodiment, shared function logic 3426 within graphics core array 3414 can include some or all logic within shared function logic 3420. In at least one embodiment, all logic elements within shared function logic 3420 may be duplicated within shared function logic 3426 of graphics core array 3414. In at least one embodiment, shared function logic 3420 is excluded in favor of shared function logic 3426 within graphics core array 3414.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, fixed function block 3530 includes a geometry and fixed function pipeline 3536 that can be shared by all sub-cores in graphics processor 3500, for example, in lower performance and/or lower power graphics processor implementations. In at least one embodiment, geometry and fixed function pipeline 3536 includes a 3D fixed function pipeline, a video front-end unit, a thread spawner and thread dispatcher, and a unified return buffer manager, which manages unified return buffers.
In at least one embodiment, fixed function block 3530 also includes a graphics SoC interface 3537, a graphics microcontroller 3538, and a media pipeline 3539. In at least one embodiment, graphics SoC interface 3537 provides an interface between graphics core 3500 and other processor cores within a system on a chip integrated circuit. In at least one embodiment, graphics microcontroller 3538 is a programmable sub-processor that is configurable to manage various functions of graphics processor 3500, including thread dispatch, scheduling, and pre-emption. In at least one embodiment, media pipeline 3539 includes logic to facilitate decoding, encoding, pre-processing, and/or post-processing of multimedia data, including image and video data. In at least one embodiment, media pipeline 3539 implements media operations via requests to compute or sampling logic within sub-cores 3501A-3501F.
In at least one embodiment, SoC interface 3537 enables graphics core 3500 to communicate with general-purpose application processor cores (e.g., CPUs) and/or other components within an SoC, including memory hierarchy elements such as a shared last level cache memory, system RAM, and/or embedded on-chip or on-package DRAM. In at least one embodiment, SoC interface 3537 can also enable communication with fixed function devices within an SoC, such as camera imaging pipelines, and enables use of and/or implements global memory atomics that may be shared between graphics core 3500 and CPUs within an SoC. In at least one embodiment, graphics SoC interface 3537 can also implement power management controls for graphics processor core 3500 and enable an interface between a clock domain of graphics processor core 3500 and other clock domains within an SoC. In at least one embodiment, SoC interface 3537 enables receipt of command buffers from a command streamer and global thread dispatcher that are configured to provide commands and instructions to each of one or more graphics cores within a graphics processor. In at least one embodiment, commands and instructions can be dispatched to media pipeline 3539, when media operations are to be performed, or a geometry and fixed function pipeline (e.g., geometry and fixed function pipeline 3536, and/or a geometry and fixed function pipeline 3514) when graphics processing operations are to be performed.
In at least one embodiment, graphics microcontroller 3538 can be configured to perform various scheduling and management tasks for graphics core 3500. In at least one embodiment, graphics microcontroller 3538 can perform graphics and/or compute workload scheduling on various graphics parallel engines within execution unit (EU) arrays 3502A-3502F, 3504A-3504F within sub-cores 3501A-3501F. In at least one embodiment, host software executing on a CPU core of an SoC including graphics core 3500 can submit workloads to one of multiple graphic processor paths, which invokes a scheduling operation on an appropriate graphics engine. In at least one embodiment, scheduling operations include determining which workload to run next, submitting a workload to a command streamer, pre-empting existing workloads running on an engine, monitoring progress of a workload, and notifying host software when a workload is complete. In at least one embodiment, graphics microcontroller 3538 can also facilitate low-power or idle states for graphics core 3500, providing graphics core 3500 with an ability to save and restore registers within graphics core 3500 across low-power state transitions independently from an operating system and/or graphics driver software on a system.
In at least one embodiment, graphics core 3500 may have greater than or fewer than illustrated sub-cores 3501A-3501F, up to N modular sub-cores. For each set of N sub-cores, in at least one embodiment, graphics core 3500 can also include shared function logic 3510, shared and/or cache memory 3512, geometry/fixed function pipeline 3514, as well as additional fixed function logic 3516 to accelerate various graphics and compute processing operations. In at least one embodiment, shared function logic 3510 can include logic units (e.g., sampler, math, and/or inter-thread communication logic) that can be shared by each N sub-cores within graphics core 3500. In at least one embodiment, shared and/or cache memory 3512 can be a last-level cache for N sub-cores 3501A-3501F within graphics core 3500 and can also serve as shared memory that is accessible by multiple sub-cores. In at least one embodiment, geometry/fixed function pipeline 3514 can be included instead of geometry/fixed function pipeline 3536 within fixed function block 3530 and can include similar logic units.
In at least one embodiment, graphics core 3500 includes additional fixed function logic 3516 that can include various fixed function acceleration logic for use by graphics core 3500. In at least one embodiment, additional fixed function logic 3516 includes an additional geometry pipeline for use in position-only shading. In position-only shading, at least two geometry pipelines exist, whereas in a full geometry pipeline within geometry and fixed function pipelines 3514, 3536, and a cull pipeline, which is an additional geometry pipeline that may be included within additional fixed function logic 3516. In at least one embodiment, a cull pipeline is a trimmed down version of a full geometry pipeline. In at least one embodiment, a full pipeline and a cull pipeline can execute different instances of an application, each instance having a separate context. In at least one embodiment, position only shading can hide long cull runs of discarded triangles, enabling shading to be completed earlier in some instances. For example, in at least one embodiment, cull pipeline logic within additional fixed function logic 3516 can execute position shaders in parallel with a main application and generally generates critical results faster than a full pipeline, as a cull pipeline fetches and shades position attributes of vertices, without performing rasterization and rendering of pixels to a frame buffer. In at least one embodiment, a cull pipeline can use generated critical results to compute visibility information for all triangles without regard to whether those triangles are culled. In at least one embodiment, a full pipeline (which in this instance may be referred to as a replay pipeline) can consume visibility information to skip culled triangles to shade only visible triangles that are finally passed to a rasterization phase.
In at least one embodiment, additional fixed function logic 3516 can also include machine-learning acceleration logic, such as fixed function matrix multiplication logic, for implementations including optimizations for machine learning training or inferencing.
In at least one embodiment, within each graphics sub-core 3501A-3501F includes a set of execution resources that may be used to perform graphics, media, and compute operations in response to requests by graphics pipeline, media pipeline, or shader programs. In at least one embodiment, graphics sub-cores 3501A-3501F include multiple EU arrays 3502A-3502F, 3504A-3504F, thread dispatch and inter-thread communication (TD/IC) logic 3503A-3503F, a 3D (e.g., texture) sampler 3505A-3505F, a media sampler 3506A-3506F, a shader processor 3507A-3507F, and shared local memory (SLM) 3508A-3508F. In at least one embodiment, EU arrays 3502A-3502F, 3504A-3504F each include multiple execution units, which are general-purpose graphics processing units capable of performing floating-point and integer/fixed-point logic operations in service of a graphics, media, or compute operation, including graphics, media, or compute shader programs. In at least one embodiment, TD/IC logic 3503A-3503F performs local thread dispatch and thread control operations for execution units within a sub-core and facilitates communication between threads executing on execution units of a sub-core. In at least one embodiment, 3D samplers 3505A-3505F can read texture or other 3D graphics related data into memory. In at least one embodiment, 3D samplers can read texture data differently based on a configured sample state and texture format associated with a given texture. In at least one embodiment, media samplers 3506A-3506F can perform similar read operations based on a type and format associated with media data. In at least one embodiment, each graphics sub-core 3501A-3501F can alternately include a unified 3D and media sampler. In at least one embodiment, threads executing on execution units within each of sub-cores 3501A-3501F can make use of shared local memory 3508A-3508F within each sub-core, to enable threads executing within a thread group to execute using a common pool of on-chip memory.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
As illustrated in
In at least one embodiment, execution units 3607 and/or 3608 are primarily used to execute shader programs. In at least one embodiment, shader processor 3602 can process various shader programs and dispatch execution threads associated with shader programs via a thread dispatcher 3604. In at least one embodiment, thread dispatcher 3604 includes logic to arbitrate thread initiation requests from graphics and media pipelines and instantiate requested threads on one or more execution units in execution units 3607 and/or 3608. For example, in at least one embodiment, a geometry pipeline can dispatch vertex, tessellation, or geometry shaders to thread execution logic for processing. In at least one embodiment, thread dispatcher 3604 can also process runtime thread spawning requests from executing shader programs.
In at least one embodiment, execution units 3607 and/or 3608 support an instruction set that includes native support for many standard 3D graphics shader instructions, such that shader programs from graphics libraries (e.g., Direct 3D and OpenGL) are executed with a minimal translation. In at least one embodiment, execution units support vertex and geometry processing (e.g., vertex programs, geometry programs, and/or vertex shaders), pixel processing (e.g., pixel shaders, fragment shaders) and general-purpose processing (e.g., compute and media shaders). In at least one embodiment, each of execution units 3607 and/or 3608, which include one or more arithmetic logic units (ALUs), is capable of multi-issue single instruction multiple data (SIMD) execution and multi-threaded operation enables an efficient execution environment despite higher latency memory accesses. In at least one embodiment, each hardware thread within each execution unit has a dedicated high-bandwidth register file and associated independent thread-state. In at least one embodiment, execution is multi-issue per clock to pipelines capable of integer, single and double precision floating point operations, SIMD branch capability, logical operations, transcendental operations, and other miscellaneous operations. In at least one embodiment, while waiting for data from memory or one of shared functions, dependency logic within execution units 3607 and/or 3608 causes a waiting thread to sleep until requested data has been returned. In at least one embodiment, while an awaiting thread is sleeping, hardware resources may be devoted to processing other threads. For example, in at least one embodiment, during a delay associated with a vertex shader operation, an execution unit can perform operations for a pixel shader, fragment shader, or another type of shader program, including a different vertex shader.
In at least one embodiment, each execution unit in execution units 3607 and/or 3608 operates on arrays of data elements. In at least one embodiment, a number of data elements is an “execution size,” or number of channels for an instruction. In at least one embodiment, an execution channel is a logical unit of execution for data element access, masking, and flow control within instructions. In at least one embodiment, a number of channels may be independent of a number of physical arithmetic logic units (ALUs) or floating point units (FPUs) for a particular graphics processor. In at least one embodiment, execution units 3607 and/or 3608 support integer and floating-point data types.
In at least one embodiment, an execution unit instruction set includes SIMD instructions. In at least one embodiment, various data elements can be stored as a packed data type in a register and execution unit will process various elements based on data size of elements. For example, in at least one embodiment, when operating on a 256-bit wide vector, 256 bits of a vector are stored in a register and an execution unit operates on a vector as four separate 64-bit packed data elements (Quad-Word (QW) size data elements), eight separate 32-bit packed data elements (Double Word (DW) size data elements), sixteen separate 16-bit packed data elements (Word (W) size data elements), or thirty-two separate 8-bit data elements (byte (B) size data elements). However, in at least one embodiment, different vector widths and register sizes are possible.
In at least one embodiment, one or more execution units can be combined into a fused execution unit 3609A-3609N having thread control logic (3611A-3611N) that is common to fused EUs such as execution unit 3607A fused with execution unit 3608A into fused execution unit 3609A. In at least one embodiment, multiple EUs can be fused into an EU group. In at least one embodiment, each EU in a fused EU group can be configured to execute a separate SIMD hardware thread, with a number of EUs in a fused EU group possibly varying according to various embodiments. In at least one embodiment, various SIMD widths can be performed per-EU, including but not limited to SIMD8, SIMD16, and SIMD32. In at least one embodiment, each fused graphics execution unit 3609A-3609N includes at least two execution units. For example, in at least one embodiment, fused execution unit 3609A includes a first EU 3607A, second EU 3608A, and thread control logic 3611A that is common to first EU 3607A and second EU 3608A. In at least one embodiment, thread control logic 3611A controls threads executed on fused graphics execution unit 3609A, allowing each EU within fused execution units 3609A-3609N to execute using a common instruction pointer register.
In at least one embodiment, one or more internal instruction caches (e.g., 3606) are included in thread execution logic 3600 to cache thread instructions for execution units. In at least one embodiment, one or more data caches (e.g., 3612) are included to cache thread data during thread execution. In at least one embodiment, sampler 3610 is included to provide texture sampling for 3D operations and media sampling for media operations. In at least one embodiment, sampler 3610 includes specialized texture or media sampling functionality to process texture or media data during sampling process before providing sampled data to an execution unit.
During execution, in at least one embodiment, graphics and media pipelines send thread initiation requests to thread execution logic 3600 via thread spawning and dispatch logic. In at least one embodiment, once a group of geometric objects has been processed and rasterized into pixel data, pixel processor logic (e.g., pixel shader logic, fragment shader logic, etc.) within shader processor 3602 is invoked to further compute output information and cause results to be written to output surfaces (e.g., color buffers, depth buffers, stencil buffers, etc.). In at least one embodiment, a pixel shader or a fragment shader calculates values of various vertex attributes that are to be interpolated across a rasterized object. In at least one embodiment, pixel processor logic within shader processor 3602 then executes an application programming interface (API)-supplied pixel or fragment shader program. In at least one embodiment, to execute a shader program, shader processor 3602 dispatches threads to an execution unit (e.g., 3608A) via thread dispatcher 3604. In at least one embodiment, shader processor 3602 uses texture sampling logic in sampler 3610 to access texture data in texture maps stored in memory. In at least one embodiment, arithmetic operations on texture data and input geometry data compute pixel color data for each geometric fragment, or discards one or more pixels from further processing.
In at least one embodiment, data port 3614 provides a memory access mechanism for thread execution logic 3600 to output processed data to memory for further processing on a graphics processor output pipeline. In at least one embodiment, data port 3614 includes or couples to one or more cache memories (e.g., data cache 3612) to cache data for memory access via a data port.
As illustrated in
In at least one embodiment, graphics execution unit 3608 has an architecture that is a combination of Simultaneous Multi-Threading (SMT) and fine-grained Interleaved Multi-Threading (IMT). In at least one embodiment, architecture has a modular configuration that can be fine-tuned at design time based on a target number of simultaneous threads and number of registers per execution unit, where execution unit resources are divided across logic used to execute multiple simultaneous threads.
In at least one embodiment, graphics execution unit 3608 can co-issue multiple instructions, which may each be different instructions. In at least one embodiment, thread arbiter 3622 of graphics execution unit thread 3608 can dispatch instructions to one of send unit 3630, branch unit 3632, or SIMD FPU(s) 3634 for execution. In at least one embodiment, each execution thread can access 128 general-purpose registers within GRF 3624, where each register can store 32 bytes, accessible as a SIMD 8-element vector of 32-bit data elements. In at least one embodiment, each execution unit thread has access to 4 kilobytes within GRF 3624, although embodiments are not so limited, and greater or fewer register resources may be provided in other embodiments. In at least one embodiment, up to seven threads can execute simultaneously, although a number of threads per execution unit can also vary according to embodiments. In at least one embodiment, in which seven threads may access 4 kilobytes, GRF 3624 can store a total of 28 kilobytes. In at least one embodiment, flexible addressing modes can permit registers to be addressed together to build effectively wider registers or to represent strided rectangular block data structures.
In at least one embodiment, memory operations, sampler operations, and other longer-latency system communications are dispatched via “send” instructions that are executed by message passing to send unit 3630. In at least one embodiment, branch instructions are dispatched to branch unit 3632 to facilitate SIMD divergence and eventual convergence.
In at least one embodiment, graphics execution unit 3608 includes one or more SIMD floating point units (FPU(s)) 3634 to perform floating-point operations. In at least one embodiment, FPU(s) 3634 also support integer computation. In at least one embodiment, FPU(s) 3634 can SIMD execute up to M number of 32-bit floating-point (or integer) operations, or SIMD execute up to 2M 16-bit integer or 16-bit floating-point operations. In at least one embodiment, at least one FPU provides extended math capability to support high-throughput transcendental math functions and double precision 64-bit floating-point. In at least one embodiment, a set of 8-bit integer SIMD ALUs 3635 are also present, and may be specifically optimized to perform operations associated with machine learning computations.
In at least one embodiment, arrays of multiple instances of graphics execution unit 3608 can be instantiated in a graphics sub-core grouping (e.g., a sub-slice). In at least one embodiment, execution unit 3608 can execute instructions across a plurality of execution channels. In at least one embodiment, each thread executed on graphics execution unit 3608 is executed on a different channel.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, one or more PPUs 3700 are configured to accelerate High Performance Computing (“HPC”), data center, and machine learning applications. In at least one embodiment, PPU 3700 is configured to accelerate deep learning systems and applications including following non-limiting examples: autonomous vehicle platforms, deep learning, high-accuracy speech, image, text recognition systems, intelligent video analytics, molecular simulations, drug discovery, disease diagnosis, weather forecasting, big data analytics, astronomy, molecular dynamics simulation, financial modeling, robotics, factory automation, real-time language translation, online search optimizations, and personalized user recommendations, and more.
In at least one embodiment, PPU 3700 includes, without limitation, an Input/Output (“I/O”) unit 3706, a front-end unit 3710, a scheduler unit 3712, a work distribution unit 3714, a hub 3716, a crossbar (“XBar”) 3720, one or more general processing clusters (“GPCs”) 3718, and one or more partition units (“memory partition units”) 3722. In at least one embodiment, PPU 3700 is connected to a host processor or other PPUs 3700 via one or more high-speed GPU interconnects (“GPU interconnects”) 3708. In at least one embodiment, PPU 3700 is connected to a host processor or other peripheral devices via a system bus 3702. In at least one embodiment, PPU 3700 is connected to a local memory comprising one or more memory devices (“memory”) 3704. In at least one embodiment, memory devices 3704 include, without limitation, one or more dynamic random access memory (“DRAM”) devices. In at least one embodiment, one or more DRAM devices are configured and/or configurable as high-bandwidth memory (“HBM”) subsystems, with multiple DRAM dies stacked within each device.
In at least one embodiment, high-speed GPU interconnect 3708 may refer to a wire-based multi-lane communications link that is used by systems to scale and include one or more PPUs 3700 combined with one or more central processing units (“CPUs”), supports cache coherence between PPUs 3700 and CPUs, and CPU mastering. In at least one embodiment, data and/or commands are transmitted by high-speed GPU interconnect 3708 through hub 3716 to/from other units of PPU 3700 such as one or more copy engines, video encoders, video decoders, power management units, and other components which may not be explicitly illustrated in
In at least one embodiment, I/O unit 3706 is configured to transmit and receive communications (e.g., commands, data) from a host processor (not illustrated in
In at least one embodiment, I/O unit 3706 decodes packets received via system bus 3702. In at least one embodiment, at least some packets represent commands configured to cause PPU 3700 to perform various operations. In at least one embodiment, I/O unit 3706 transmits decoded commands to various other units of PPU 3700 as specified by commands. In at least one embodiment, commands are transmitted to front-end unit 3710 and/or transmitted to hub 3716 or other units of PPU 3700 such as one or more copy engines, a video encoder, a video decoder, a power management unit, etc. (not explicitly illustrated in
In at least one embodiment, a program executed by host processor encodes a command stream in a buffer that provides workloads to PPU 3700 for processing. In at least one embodiment, a workload comprises instructions and data to be processed by those instructions. In at least one embodiment, a buffer is a region in a memory that is accessible (e.g., read/write) by both a host processor and PPU 3700—a host interface unit may be configured to access that buffer in a system memory connected to system bus 3702 via memory requests transmitted over system bus 3702 by I/O unit 3706. In at least one embodiment, a host processor writes a command stream to a buffer and then transmits a pointer to a start of a command stream to PPU 3700 such that front-end unit 3710 receives pointers to one or more command streams and manages one or more command streams, reading commands from command streams and forwarding commands to various units of PPU 3700.
In at least one embodiment, front-end unit 3710 is coupled to scheduler unit 3712 that configures various GPCs 3718 to process tasks defined by one or more command streams. In at least one embodiment, scheduler unit 3712 is configured to track state information related to various tasks managed by scheduler unit 3712 where state information may indicate which of GPCs 3718 a task is assigned to, whether task is active or inactive, a priority level associated with task, and so forth. In at least one embodiment, scheduler unit 3712 manages execution of a plurality of tasks on one or more of GPCs 3718.
In at least one embodiment, scheduler unit 3712 is coupled to work distribution unit 3714 that is configured to dispatch tasks for execution on GPCs 3718. In at least one embodiment, work distribution unit 3714 tracks a number of scheduled tasks received from scheduler unit 3712 and work distribution unit 3714 manages a pending task pool and an active task pool for each of GPCs 3718. In at least one embodiment, pending task pool comprises a number of slots (e.g., 32 slots) that contain tasks assigned to be processed by a particular GPC 3718; an active task pool may comprise a number of slots (e.g., 4 slots) for tasks that are actively being processed by GPCs 3718 such that as one of GPCs 3718 completes execution of a task, that task is evicted from that active task pool for GPC 3718 and another task from a pending task pool is selected and scheduled for execution on GPC 3718. In at least one embodiment, if an active task is idle on GPC 3718, such as while waiting for a data dependency to be resolved, then that active task is evicted from GPC 3718 and returned to that pending task pool while another task in that pending task pool is selected and scheduled for execution on GPC 3718.
In at least one embodiment, work distribution unit 3714 communicates with one or more GPCs 3718 via XBar 3720. In at least one embodiment, XBar 3720 is an interconnect network that couples many of units of PPU 3700 to other units of PPU 3700 and can be configured to couple work distribution unit 3714 to a particular GPC 3718. In at least one embodiment, one or more other units of PPU 3700 may also be connected to XBar 3720 via hub 3716.
In at least one embodiment, tasks are managed by scheduler unit 3712 and dispatched to one of GPCs 3718 by work distribution unit 3714. In at least one embodiment, GPC 3718 is configured to process task and generate results. In at least one embodiment, results may be consumed by other tasks within GPC 3718, routed to a different GPC 3718 via XBar 3720, or stored in memory 3704. In at least one embodiment, results can be written to memory 3704 via partition units 3722, which implement a memory interface for reading and writing data to/from memory 3704. In at least one embodiment, results can be transmitted to another PPU or CPU via high-speed GPU interconnect 3708. In at least one embodiment, PPU 3700 includes, without limitation, a number U of partition units 3722 that is equal to a number of separate and distinct memory devices 3704 coupled to PPU 3700, as described in more detail herein in conjunction with
In at least one embodiment, a host processor executes a driver kernel that implements an application programming interface (“API”) that enables one or more applications executing on a host processor to schedule operations for execution on PPU 3700. In at least one embodiment, multiple compute applications are simultaneously executed by PPU 3700 and PPU 3700 provides isolation, quality of service (“QoS”), and independent address spaces for multiple compute applications. In at least one embodiment, an application generates instructions (e.g., in form of API calls) that cause a driver kernel to generate one or more tasks for execution by PPU 3700 and that driver kernel outputs tasks to one or more streams being processed by PPU 3700. In at least one embodiment, each task comprises one or more groups of related threads, which may be referred to as a warp. In at least one embodiment, a warp comprises a plurality of related threads (e.g., 32 threads) that can be executed in parallel. In at least one embodiment, cooperating threads can refer to a plurality of threads including instructions to perform task and that exchange data through shared memory. In at least one embodiment, threads and cooperating threads are described in more detail in conjunction with
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, operation of GPC 3800 is controlled by pipeline manager 3802. In at least one embodiment, pipeline manager 3802 manages configuration of one or more DPCs 3806 for processing tasks allocated to GPC 3800. In at least one embodiment, pipeline manager 3802 configures at least one of one or more DPCs 3806 to implement at least a portion of a graphics rendering pipeline. In at least one embodiment, DPC 3806 is configured to execute a vertex shader program on a programmable streaming multi-processor (“SM”) 3814. In at least one embodiment, pipeline manager 3802 is configured to route packets received from a work distribution unit to appropriate logical units within GPC 3800, in at least one embodiment, and some packets may be routed to fixed function hardware units in preROP 3804 and/or raster engine 3808 while other packets may be routed to DPCs 3806 for processing by a primitive engine 3812 or SM 3814. In at least one embodiment, pipeline manager 3802 configures at least one of DPCs 3806 to implement a neural network model and/or a computing pipeline.
In at least one embodiment, preROP unit 3804 is configured, in at least one embodiment, to route data generated by raster engine 3808 and DPCs 3806 to a Raster Operations (“ROP”) unit in partition unit 3722, described in more detail above in conjunction with
In at least one embodiment, each DPC 3806 included in GPC 3800 comprises, without limitation, an M-Pipe Controller (“MPC”) 3810; primitive engine 3812; one or more SMs 3814; and any suitable combination thereof. In at least one embodiment, MPC 3810 controls operation of DPC 3806, routing packets received from pipeline manager 3802 to appropriate units in DPC 3806. In at least one embodiment, packets associated with a vertex are routed to primitive engine 3812, which is configured to fetch vertex attributes associated with a vertex from memory; in contrast, packets associated with a shader program may be transmitted to SM 3814.
In at least one embodiment, SM 3814 comprises, without limitation, a programmable streaming processor that is configured to process tasks represented by a number of threads. In at least one embodiment, SM 3814 is multi-threaded and configured to execute a plurality of threads (e.g., 32 threads) from a particular group of threads concurrently and implements a Single-Instruction, Multiple-Data (“SIMD”) architecture where each thread in a group of threads (e.g., a warp) is configured to process a different set of data based on same set of instructions. In at least one embodiment, all threads in group of threads execute a common set of instructions. In at least one embodiment, SM 3814 implements a Single-Instruction, Multiple Thread (“SIMT”) architecture wherein each thread in a group of threads is configured to process a different set of data based on that common set of instructions, but where individual threads in a group of threads are allowed to diverge during execution. In at least one embodiment, a program counter, call stack, and execution state is maintained for each warp, enabling concurrency between warps and serial execution within warps when threads within a warp diverge. In another embodiment, a program counter, call stack, and execution state is maintained for each individual thread, enabling equal concurrency between all threads, within and between warps. In at least one embodiment, execution state is maintained for each individual thread and threads executing common instructions may be converged and executed in parallel for better efficiency. At least one embodiment of SM 3814 is described in more detail herein.
In at least one embodiment, MMU 3818 provides an interface between GPC 3800 and a memory partition unit (e.g., partition unit 3722 of
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
In at least one embodiment, memory interface 3906 implements a high bandwidth memory second generation (“HBM2”) memory interface and Y equals half of U. In at least one embodiment, HBM2 memory stacks are located on a physical package with a PPU, providing substantial power and area savings compared with conventional GDDR5 SDRAM systems. In at least one embodiment, each HBM2 stack includes, without limitation, four memory dies with Y=4, with each HBM2 stack including two 128-bit channels per die for a total of 8 channels and a data bus width of 1024 bits. In at least one embodiment, that memory supports Single-Error Correcting Double-Error Detecting (“SECDED”) Error Correction Code (“ECC”) to protect data. In at least one embodiment, ECC can provide higher reliability for compute applications that are sensitive to data corruption.
In at least one embodiment, PPU implements a multi-level memory hierarchy. In at least one embodiment, memory partition unit 3900 supports a unified memory to provide a single unified virtual address space for central processing unit (“CPU”) and PPU memory, enabling data sharing between virtual memory systems. In at least one embodiment frequency of accesses by a PPU to a memory located on other processors is traced to ensure that memory pages are moved to physical memory of PPU that is accessing pages more frequently. In at least one embodiment, high-speed GPU interconnect 3708 supports address translation services allowing PPU to directly access a CPU's page tables and providing full access to CPU memory by a PPU.
In at least one embodiment, copy engines transfer data between multiple PPUs or between PPUs and CPUs. In at least one embodiment, copy engines can generate page faults for addresses that are not mapped into page tables and memory partition unit 3900 then services page faults, mapping addresses into page table, after which copy engine performs a transfer. In at least one embodiment, memory is pinned (e.g., non-pageable) for multiple copy engine operations between multiple processors, substantially reducing available memory. In at least one embodiment, with hardware page faulting, addresses can be passed to copy engines without regard as to whether memory pages are resident, and a copy process is transparent.
Data from memory 3704 of
ROP unit 3902 performs graphics raster operations related to pixel color, such as color compression, pixel blending, and more, in at least one embodiment. ROP unit 3902, in at least one embodiment, implements depth testing in conjunction with raster engine 3808, receiving a depth for a sample location associated with a pixel fragment from a culling engine of raster engine 3808. In at least one embodiment, depth is tested against a corresponding depth in a depth buffer for a sample location associated with a fragment. In at least one embodiment, if that fragment passes that depth test for that sample location, then ROP unit 3902 updates depth buffer and transmits a result of that depth test to raster engine 3808. It will be appreciated that a number of partition units 3900 may be different than a number of GPCs and, therefore, each ROP unit 3902 can, in at least one embodiment, be coupled to each GPC. In at least one embodiment, ROP unit 3902 tracks packets received from different GPCs and determines whether a result generated by ROP unit 3902 is to be routed to through XBar 3720.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, a work distribution unit dispatches tasks for execution on general processing clusters (“GPCs”) of parallel processing units (“PPUs”) and each task is allocated to a particular Data Processing Cluster (“DPC”) within a GPC and, if a task is associated with a shader program, that task is allocated to one of SMs 4000. In at least one embodiment, scheduler unit 4004 receives tasks from a work distribution unit and manages instruction scheduling for one or more thread blocks assigned to SM 4000. In at least one embodiment, scheduler unit 4004 schedules thread blocks for execution as warps of parallel threads, wherein each thread block is allocated at least one warp. In at least one embodiment, each warp executes threads. In at least one embodiment, scheduler unit 4004 manages a plurality of different thread blocks, allocating warps to different thread blocks and then dispatching instructions from plurality of different cooperative groups to various functional units (e.g., processing cores 4010, SFUs 4012, and LSUs 4014) during each clock cycle.
In at least one embodiment, Cooperative Groups may refer to a programming model for organizing groups of communicating threads that allows developers to express granularity at which threads are communicating, enabling expression of richer, more efficient parallel decompositions. In at least one embodiment, cooperative launch APIs support synchronization amongst thread blocks for execution of parallel algorithms. In at least one embodiment, applications of conventional programming models provide a single, simple construct for synchronizing cooperating threads: a barrier across all threads of a thread block (e.g., syncthreads( ) function). However, in at least one embodiment, programmers may define groups of threads at smaller than thread block granularities and synchronize within defined groups to enable greater performance, design flexibility, and software reuse in form of collective group-wide function interfaces. In at least one embodiment, Cooperative Groups enables programmers to define groups of threads explicitly at sub-block (e.g., as small as a single thread) and multi-block granularities, and to perform collective operations such as synchronization on threads in a cooperative group. In at least one embodiment, that programming model supports clean composition across software boundaries, so that libraries and utility functions can synchronize safely within their local context without having to make assumptions about convergence. In at least one embodiment, Cooperative Groups primitives enable new patterns of cooperative parallelism, including, without limitation, producer-consumer parallelism, opportunistic parallelism, and global synchronization across an entire grid of thread blocks.
In at least one embodiment, a dispatch unit 4006 is configured to transmit instructions to one or more functional units and scheduler unit 4004 and includes, without limitation, two dispatch units 4006 that enable two different instructions from a common warp to be dispatched during each clock cycle. In at least one embodiment, each scheduler unit 4004 includes a single dispatch unit 4006 or additional dispatch units 4006.
In at least one embodiment, each SM 4000, in at least one embodiment, includes, without limitation, register file 4008 that provides a set of registers for functional units of SM 4000. In at least one embodiment, register file 4008 is divided between each functional unit such that each functional unit is allocated a dedicated portion of register file 4008. In at least one embodiment, register file 4008 is divided between different warps being executed by SM 4000 and register file 4008 provides temporary storage for operands connected to data paths of functional units. In at least one embodiment, each SM 4000 comprises, without limitation, a plurality of L processing cores 4010, where L is a positive integer. In at least one embodiment, SM 4000 includes, without limitation, a large number (e.g., 128 or more) of distinct processing cores 4010. In at least one embodiment, each processing core 4010 includes, without limitation, a fully-pipelined, single-precision, double-precision, and/or mixed precision processing unit that includes, without limitation, a floating point arithmetic logic unit and an integer arithmetic logic unit. In at least one embodiment, floating point arithmetic logic units implement IEEE 754-2008 standard for floating point arithmetic. In at least one embodiment, processing cores 4010 include, without limitation, 64 single-precision (32-bit) floating point cores, 64 integer cores, 32 double-precision (64-bit) floating point cores, and 8 tensor cores.
Tensor cores are configured to perform matrix operations in accordance with at least one embodiment. In at least one embodiment, one or more tensor cores are included in processing cores 4010. In at least one embodiment, tensor cores are configured to perform deep learning matrix arithmetic, such as convolution operations for neural network training and inferencing. In at least one embodiment, each tensor core operates on a 4×4 matrix and performs a matrix multiply and accumulate operation, D=A×B+C, where A, B, C, and D are 4×4 matrices.
In at least one embodiment, matrix multiply inputs A and B are 16-bit floating point matrices and accumulation matrices C and D are 16-bit floating point or 32-bit floating point matrices. In at least one embodiment, tensor cores operate on 16-bit floating point input data with 32-bit floating point accumulation. In at least one embodiment, 16-bit floating point multiply uses 64 operations and results in a full precision product that is then accumulated using 32-bit floating point addition with other intermediate products for a 4×4×4 matrix multiply. Tensor cores are used to perform much larger two-dimensional or higher dimensional matrix operations, built up from these smaller elements, in at least one embodiment. In at least one embodiment, an API, such as a CUDA 9 C++ API, exposes specialized matrix load, matrix multiply and accumulate, and matrix store operations to efficiently use tensor cores from a CUDA-C++ program. In at least one embodiment, at a CUDA level, a warp-level interface assumes 16×16 size matrices spanning all 32 threads of warp.
In at least one embodiment, each SM 4000 comprises, without limitation, M SFUs 4012 that perform special functions (e.g., attribute evaluation, reciprocal square root, and like). In at least one embodiment, SFUs 4012 include, without limitation, a tree traversal unit configured to traverse a hierarchical tree data structure. In at least one embodiment, SFUs 4012 include, without limitation, a texture unit configured to perform texture map filtering operations. In at least one embodiment, texture units are configured to load texture maps (e.g., a 2D array of texels) from memory and sample texture maps to produce sampled texture values for use in shader programs executed by SM 4000. In at least one embodiment, texture maps are stored in shared memory/L1 cache 4018. In at least one embodiment, texture units implement texture operations such as filtering operations using mip-maps (e.g., texture maps of varying levels of detail), in accordance with at least one embodiment. In at least one embodiment, each SM 4000 includes, without limitation, two texture units.
Each SM 4000 comprises, without limitation, N LSUs 4014 that implement load and store operations between shared memory/L1 cache 4018 and register file 4008, in at least one embodiment. Interconnect network 4016 connects each functional unit to register file 4008 and LSU 4014 to register file 4008 and shared memory/L1 cache 4018 in at least one embodiment. In at least one embodiment, interconnect network 4016 is a crossbar that can be configured to connect any functional units to any registers in register file 4008 and connect LSUs 4014 to register file 4008 and memory locations in shared memory/L1 cache 4018.
In at least one embodiment, shared memory/L1 cache 4018 is an array of on-chip memory that allows for data storage and communication between SM 4000 and primitive engine and between threads in SM 4000, in at least one embodiment. In at least one embodiment, shared memory/L1 cache 4018 comprises, without limitation, 128 KB of storage capacity and is in a path from SM 4000 to a partition unit. In at least one embodiment, shared memory/L1 cache 4018, in at least one embodiment, is used to cache reads and writes. In at least one embodiment, one or more of shared memory/L1 cache 4018, L2 cache, and memory are backing stores.
Combining data cache and shared memory functionality into a single memory block provides improved performance for both types of memory accesses, in at least one embodiment. In at least one embodiment, capacity is used or is usable as a cache by programs that do not use shared memory, such as if shared memory is configured to use half of a capacity, and texture and load/store operations can use remaining capacity. Integration within shared memory/L1 cache 4018 enables shared memory/L1 cache 4018 to function as a high-throughput conduit for streaming data while simultaneously providing high-bandwidth and low-latency access to frequently reused data, in accordance with at least one embodiment. In at least one embodiment, when configured for general purpose parallel computation, a simpler configuration can be used compared with graphics processing. In at least one embodiment, fixed function graphics processing units are bypassed, creating a much simpler programming model. In a general purpose parallel computation configuration, a work distribution unit assigns and distributes blocks of threads directly to DPCs, in at least one embodiment. In at least one embodiment, threads in a block execute a common program, using a unique thread ID in calculation to ensure each thread generates unique results, using SM 4000 to execute program and perform calculations, shared memory/L1 cache 4018 to communicate between threads, and LSU 4014 to read and write global memory through shared memory/L1 cache 4018 and memory partition unit. In at least one embodiment, when configured for general purpose parallel computation, SM 4000 writes commands that scheduler unit 4004 can use to launch new work on DPCs.
In at least one embodiment, a PPU is included in or coupled to a desktop computer, a laptop computer, a tablet computer, servers, supercomputers, a smart-phone (e.g., a wireless, hand-held device), personal digital assistant (“PDA”), a digital camera, a vehicle, a head mounted display, a hand-held electronic device, and more. In at least one embodiment, a PPU is embodied on a single semiconductor substrate. In at least one embodiment, a PPU is included in a system-on-a-chip (“SoC”) along with one or more other devices such as additional PPUs, memory, a reduced instruction set computer (“RISC”) CPU, a memory management unit (“MMU”), a digital-to-analog converter (“DAC”), and like.
In at least one embodiment, a PPU may be included on a graphics card that includes one or more memory devices. In at least one embodiment, that graphics card may be configured to interface with a PCIe slot on a motherboard of a desktop computer. In at least one embodiment, that PPU may be an integrated graphics processing unit (“iGPU”) included in chipset of a motherboard.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
Embodiments are disclosed related a virtualized computing platform for advanced computing, such as image inferencing and image processing in medical applications. Without limitation, embodiments may include radiography, magnetic resonance imaging (MM), nuclear medicine, ultrasound, sonography, elastography, photoacoustic imaging, tomography, echocardiography, functional near-infrared spectroscopy, and magnetic particle imaging, or a combination thereof. In at least one embodiment, a virtualized computing platform and associated processes described herein may additionally or alternatively be used, without limitation, in forensic science analysis, sub-surface detection and imaging (e.g., oil exploration, archaeology, paleontology, etc.), topography, oceanography, geology, osteology, meteorology, intelligent area or object tracking and monitoring, sensor data processing (e.g., RADAR, SONAR, LIDAR, etc.), and/or genomics and gene sequencing.
With reference to
In at least one embodiment, process 4100 may be executed within a training system 4104 and/or a deployment system 4106. In at least one embodiment, training system 4104 may be used to perform training, deployment, and implementation of machine learning models (e.g., neural networks, object detection algorithms, computer vision algorithms, etc.) for use in deployment system 4106. In at least one embodiment, deployment system 4106 may be configured to offload processing and compute resources among a distributed computing environment to reduce infrastructure requirements at facility 4102. In at least one embodiment, deployment system 4106 may provide a streamlined platform for selecting, customizing, and implementing virtual instruments for use with imaging devices (e.g., Mill, Conn. Scan, X-Ray, Ultrasound, etc.) or sequencing devices at facility 4102. In at least one embodiment, virtual instruments may include software-defined applications for performing one or more processing operations with respect to imaging data generated by imaging devices, sequencing devices, radiology devices, and/or other device types. In at least one embodiment, one or more applications in a pipeline may use or call upon services (e.g., inference, visualization, compute, AI, etc.) of deployment system 4106 during execution of applications.
In at least one embodiment, some of applications used in advanced processing and inferencing pipelines may use machine learning models or other AI to perform one or more processing steps. In at least one embodiment, machine learning models may be trained at facility 4102 using data 4108 (such as imaging data) generated at facility 4102 (and stored on one or more picture archiving and communication system (PACS) servers at facility 4102), may be trained using imaging or sequencing data 4108 from another facility or facilities (e.g., a different hospital, lab, clinic, etc.), or a combination thereof. In at least one embodiment, training system 4104 may be used to provide applications, services, and/or other resources for generating working, deployable machine learning models for deployment system 4106.
In at least one embodiment, a model registry 4124 may be backed by object storage that may support versioning and object metadata. In at least one embodiment, object storage may be accessible through, for example, a cloud storage (e.g., a cloud 4226 of
In at least one embodiment, a training pipeline 4204 (
In at least one embodiment, training pipeline 4204 (
In at least one embodiment, training pipeline 4204 (
In at least one embodiment, deployment system 4106 may include software 4118, services 4120, hardware 4122, and/or other components, features, and functionality. In at least one embodiment, deployment system 4106 may include a software “stack,” such that software 4118 may be built on top of services 4120 and may use services 4120 to perform some or all of processing tasks, and services 4120 and software 4118 may be built on top of hardware 4122 and use hardware 4122 to execute processing, storage, and/or other compute tasks of deployment system 4106.
In at least one embodiment, software 4118 may include any number of different containers, where each container may execute an instantiation of an application. In at least one embodiment, each application may perform one or more processing tasks in an advanced processing and inferencing pipeline (e.g., inferencing, object detection, feature detection, segmentation, image enhancement, calibration, etc.). In at least one embodiment, for each type of imaging device (e.g., CT, MM, X-Ray, ultrasound, sonography, echocardiography, etc.), sequencing device, radiology device, genomics device, etc., there may be any number of containers that may perform a data processing task with respect to imaging data 4108 (or other data types, such as those described herein) generated by a device. In at least one embodiment, an advanced processing and inferencing pipeline may be defined based on selections of different containers that are desired or required for processing imaging data 4108, in addition to containers that receive and configure imaging data for use by each container and/or for use by facility 4102 after processing through a pipeline (e.g., to convert outputs back to a usable data type, such as digital imaging and communications in medicine (DICOM) data, radiology information system (RIS) data, clinical information system (CIS) data, remote procedure call (RPC) data, data substantially compliant with a representation state transfer (REST) interface, data substantially compliant with a file-based interface, and/or raw data, for storage and display at facility 4102). In at least one embodiment, a combination of containers within software 4118 (e.g., that make up a pipeline) may be referred to as a virtual instrument (as described in more detail herein), and a virtual instrument may leverage services 4120 and hardware 4122 to execute some or all processing tasks of applications instantiated in containers.
In at least one embodiment, a data processing pipeline may receive input data (e.g., imaging data 4108) in a DICOM, RIS, CIS, REST compliant, RPC, raw, and/or other format in response to an inference request (e.g., a request from a user of deployment system 4106, such as a clinician, a doctor, a radiologist, etc.). In at least one embodiment, input data may be representative of one or more images, video, and/or other data representations generated by one or more imaging devices, sequencing devices, radiology devices, genomics devices, and/or other device types. In at least one embodiment, data may undergo pre-processing as part of data processing pipeline to prepare data for processing by one or more applications. In at least one embodiment, post-processing may be performed on an output of one or more inferencing tasks or other processing tasks of a pipeline to prepare an output data for a next application and/or to prepare output data for transmission and/or use by a user (e.g., as a response to an inference request). In at least one embodiment, inferencing tasks may be performed by one or more machine learning models, such as trained or deployed neural networks, which may include output models 4116 of training system 4104.
In at least one embodiment, tasks of data processing pipeline may be encapsulated in a container(s) that each represent a discrete, fully functional instantiation of an application and virtualized computing environment that is able to reference machine learning models. In at least one embodiment, containers or applications may be published into a private (e.g., limited access) area of a container registry (described in more detail herein), and trained or deployed models may be stored in model registry 4124 and associated with one or more applications. In at least one embodiment, images of applications (e.g., container images) may be available in a container registry, and once selected by a user from a container registry for deployment in a pipeline, an image may be used to generate a container for an instantiation of an application for use by a user's system.
In at least one embodiment, developers (e.g., software developers, clinicians, doctors, etc.) may develop, publish, and store applications (e.g., as containers) for performing image processing and/or inferencing on supplied data. In at least one embodiment, development, publishing, and/or storing may be performed using a software development kit (SDK) associated with a system (e.g., to ensure that an application and/or container developed is compliant with or compatible with a system). In at least one embodiment, an application that is developed may be tested locally (e.g., at a first facility, on data from a first facility) with an SDK which may support at least some of services 4120 as a system (e.g., system 4200 of
In at least one embodiment, developers may then share applications or containers through a network for access and use by users of a system (e.g., system 4200 of
In at least one embodiment, to aid in processing or execution of applications or containers in pipelines, services 4120 may be leveraged. In at least one embodiment, services 4120 may include compute services, artificial intelligence (AI) services, visualization services, and/or other service types. In at least one embodiment, services 4120 may provide functionality that is common to one or more applications in software 4118, so functionality may be abstracted to a service that may be called upon or leveraged by applications. In at least one embodiment, functionality provided by services 4120 may run dynamically and more efficiently, while also scaling well by allowing applications to process data in parallel (e.g., using a parallel computing platform 4230 (
In at least one embodiment, where a service 4120 includes an AI service (e.g., an inference service), one or more machine learning models associated with an application for anomaly detection (e.g., tumors, growth abnormalities, scarring, etc.) may be executed by calling upon (e.g., as an API call) an inference service (e.g., an inference server) to execute machine learning model(s), or processing thereof, as part of application execution. In at least one embodiment, where another application includes one or more machine learning models for segmentation tasks, an application may call upon an inference service to execute machine learning models for performing one or more of processing operations associated with segmentation tasks. In at least one embodiment, software 4118 implementing advanced processing and inferencing pipeline that includes segmentation application and anomaly detection application may be streamlined because each application may call upon a same inference service to perform one or more inferencing tasks.
In at least one embodiment, hardware 4122 may include GPUs, CPUs, graphics cards, an AI/deep learning system (e.g., an AI supercomputer, such as NVIDIA's DGX supercomputer system), a cloud platform, or a combination thereof. In at least one embodiment, different types of hardware 4122 may be used to provide efficient, purpose-built support for software 4118 and services 4120 in deployment system 4106. In at least one embodiment, use of GPU processing may be implemented for processing locally (e.g., at facility 4102), within an AI/deep learning system, in a cloud system, and/or in other processing components of deployment system 4106 to improve efficiency, accuracy, and efficacy of image processing, image reconstruction, segmentation, Mill exams, stroke or heart attack detection (e.g., in real-time), image quality in rendering, etc. In at least one embodiment, a facility may include imaging devices, genomics devices, sequencing devices, and/or other device types on-premises that may leverage GPUs to generate imaging data representative of a subject's anatomy.
In at least one embodiment, software 4118 and/or services 4120 may be optimized for GPU processing with respect to deep learning, machine learning, and/or high-performance computing, as non-limiting examples. In at least one embodiment, at least some of computing environment of deployment system 4106 and/or training system 4104 may be executed in a datacenter one or more supercomputers or high performance computing systems, with GPU optimized software (e.g., hardware and software combination of NVIDIA's DGX system). In at least one embodiment, datacenters may be compliant with provisions of HIPAA, such that receipt, processing, and transmission of imaging data and/or other patient data is securely handled with respect to privacy of patient data. In at least one embodiment, hardware 4122 may include any number of GPUs that may be called upon to perform processing of data in parallel, as described herein. In at least one embodiment, cloud platform may further include GPU processing for GPU-optimized execution of deep learning tasks, machine learning tasks, or other computing tasks. In at least one embodiment, cloud platform (e.g., NVIDIA's NGC) may be executed using an AI/deep learning supercomputer(s) and/or GPU-optimized software (e.g., as provided on NVIDIA's DGX systems) as a hardware abstraction and scaling platform. In at least one embodiment, cloud platform may integrate an application container clustering system or orchestration system (e.g., KUBERNETES) on multiple GPUs to enable seamless scaling and load balancing.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, system 4200 (e.g., training system 4104 and/or deployment system 4106) may implemented in a cloud computing environment (e.g., using cloud 4226). In at least one embodiment, system 4200 may be implemented locally with respect to a healthcare services facility, or as a combination of both cloud and local computing resources. In at least one embodiment, in embodiments where cloud computing is implemented, patient data may be separated from, or unprocessed by, by one or more components of system 4200 that would render processing non-compliant with HIPAA and/or other data handling and privacy regulations or laws. In at least one embodiment, access to APIs in cloud 4226 may be restricted to authorized users through enacted security measures or protocols. In at least one embodiment, a security protocol may include web tokens that may be signed by an authentication (e.g., AuthN, AuthZ, Gluecon, etc.) service and may carry appropriate authorization. In at least one embodiment, APIs of virtual instruments (described herein), or other instantiations of system 4200, may be restricted to a set of public IPs that have been vetted or authorized for interaction.
In at least one embodiment, various components of system 4200 may communicate between and among one another using any of a variety of different network types, including but not limited to local area networks (LANs) and/or wide area networks (WANs) via wired and/or wireless communication protocols. In at least one embodiment, communication between facilities and components of system 4200 (e.g., for transmitting inference requests, for receiving results of inference requests, etc.) may be communicated over a data bus or data busses, wireless data protocols (Wi-Fi), wired data protocols (e.g., Ethernet), etc.
In at least one embodiment, training system 4104 may execute training pipelines 4204, similar to those described herein with respect to
In at least one embodiment, output model(s) 4116 and/or pre-trained model(s) 4206 may include any types of machine learning models depending on implementation or embodiment. In at least one embodiment, and without limitation, machine learning models used by system 4200 may include machine learning model(s) using linear regression, logistic regression, decision trees, support vector machines (SVM), Naïve Bayes, k-nearest neighbor (Knn), K means clustering, random forest, dimensionality reduction algorithms, gradient boosting algorithms, neural networks (e.g., auto-encoders, convolutional, recurrent, perceptrons, Long/Short Term Memory (LSTM), Hopfield, Boltzmann, deep belief, deconvolutional, generative adversarial, liquid state machine, etc.), and/or other types of machine learning models.
In at least one embodiment, training pipelines 4204 may include AI-assisted annotation, as described in more detail herein with respect to at least
In at least one embodiment, a software layer may be implemented as a secure, encrypted, and/or authenticated API through which applications or containers may be invoked (e.g., called) from an external environment(s) (e.g., facility 4102). In at least one embodiment, applications may then call or execute one or more services 4120 for performing compute, AI, or visualization tasks associated with respective applications, and software 4118 and/or services 4120 may leverage hardware 4122 to perform processing tasks in an effective and efficient manner.
In at least one embodiment, deployment system 4106 may execute deployment pipelines 4210. In at least one embodiment, deployment pipelines 4210 may include any number of applications that may be sequentially, non-sequentially, or otherwise applied to imaging data (and/or other data types) generated by imaging devices, sequencing devices, genomics devices, etc.—including AI-assisted annotation, as described above. In at least one embodiment, as described herein, a deployment pipeline 4210 for an individual device may be referred to as a virtual instrument for a device (e.g., a virtual ultrasound instrument, a virtual CT scan instrument, a virtual sequencing instrument, etc.). In at least one embodiment, for a single device, there may be more than one deployment pipeline 4210 depending on information desired from data generated by a device. In at least one embodiment, where detections of anomalies are desired from an MM machine, there may be a first deployment pipeline 4210, and where image enhancement is desired from output of an Mill machine, there may be a second deployment pipeline 4210.
In at least one embodiment, applications available for deployment pipelines 4210 may include any application that may be used for performing processing tasks on imaging data or other data from devices. In at least one embodiment, different applications may be responsible for image enhancement, segmentation, reconstruction, anomaly detection, object detection, feature detection, treatment planning, dosimetry, beam planning (or other radiation treatment procedures), and/or other analysis, image processing, or inferencing tasks. In at least one embodiment, deployment system 4106 may define constructs for each of applications, such that users of deployment system 4106 (e.g., medical facilities, labs, clinics, etc.) may understand constructs and adapt applications for implementation within their respective facility. In at least one embodiment, an application for image reconstruction may be selected for inclusion in deployment pipeline 4210, but data type generated by an imaging device may be different from a data type used within an application. In at least one embodiment, DICOM adapter 4202B (and/or a DICOM reader) or another data type adapter or reader (e.g., RIS, CIS, REST compliant, RPC, raw, etc.) may be used within deployment pipeline 4210 to convert data to a form useable by an application within deployment system 4106. In at least one embodiment, access to DICOM, RIS, CIS, REST compliant, RPC, raw, and/or other data type libraries may be accumulated and pre-processed, including decoding, extracting, and/or performing any convolutions, color corrections, sharpness, gamma, and/or other augmentations to data. In at least one embodiment, DICOM, RIS, CIS, REST compliant, RPC, and/or raw data may be unordered and a pre-pass may be executed to organize or sort collected data. In at least one embodiment, because various applications may share common image operations, in some embodiments, a data augmentation library (e.g., as one of services 4120) may be used to accelerate these operations. In at least one embodiment, to avoid bottlenecks of conventional processing approaches that rely on CPU processing, parallel computing platform 4230 may be used for GPU acceleration of these processing tasks.
In at least one embodiment, an image reconstruction application may include a processing task that includes use of a machine learning model. In at least one embodiment, a user may desire to use their own machine learning model, or to select a machine learning model from model registry 4124. In at least one embodiment, a user may implement their own machine learning model or select a machine learning model for inclusion in an application for performing a processing task. In at least one embodiment, applications may be selectable and customizable, and by defining constructs of applications, deployment and implementation of applications for a particular user are presented as a more seamless user experience. In at least one embodiment, by leveraging other features of system 4200—such as services 4120 and hardware 4122—deployment pipelines 4210 may be even more user friendly, provide for easier integration, and produce more accurate, efficient, and timely results.
In at least one embodiment, deployment system 4106 may include a user interface 4214 (e.g., a graphical user interface, a web interface, etc.) that may be used to select applications for inclusion in deployment pipeline(s) 4210, arrange applications, modify or change applications or parameters or constructs thereof, use and interact with deployment pipeline(s) 4210 during set-up and/or deployment, and/or to otherwise interact with deployment system 4106. In at least one embodiment, although not illustrated with respect to training system 4104, user interface 4214 (or a different user interface) may be used for selecting models for use in deployment system 4106, for selecting models for training, or retraining, in training system 4104, and/or for otherwise interacting with training system 4104.
In at least one embodiment, pipeline manager 4212 may be used, in addition to an application orchestration system 4228, to manage interaction between applications or containers of deployment pipeline(s) 4210 and services 4120 and/or hardware 4122. In at least one embodiment, pipeline manager 4212 may be configured to facilitate interactions from application to application, from application to service 4120, and/or from application or service to hardware 4122. In at least one embodiment, although illustrated as included in software 4118, this is not intended to be limiting, and in some examples (e.g., as illustrated in
In at least one embodiment, each application and/or container (or image thereof) may be individually developed, modified, and deployed (e.g., a first user or developer may develop, modify, and deploy a first application and a second user or developer may develop, modify, and deploy a second application separate from a first user or developer), which may allow for focus on, and attention to, a task of a single application and/or container(s) without being hindered by tasks of another application(s) or container(s). In at least one embodiment, communication, and cooperation between different containers or applications may be aided by pipeline manager 4212 and application orchestration system 4228. In at least one embodiment, so long as an expected input and/or output of each container or application is known by a system (e.g., based on constructs of applications or containers), application orchestration system 4228 and/or pipeline manager 4212 may facilitate communication among and between, and sharing of resources among and between, each of applications or containers. In at least one embodiment, because one or more of applications or containers in deployment pipeline(s) 4210 may share same services and resources, application orchestration system 4228 may orchestrate, load balance, and determine sharing of services or resources between and among various applications or containers. In at least one embodiment, a scheduler may be used to track resource requirements of applications or containers, current usage or planned usage of these resources, and resource availability. In at least one embodiment, a scheduler may thus allocate resources to different applications and distribute resources between and among applications in view of requirements and availability of a system. In some examples, a scheduler (and/or other component of application orchestration system 4228) may determine resource availability and distribution based on constraints imposed on a system (e.g., user constraints), such as quality of service (QoS), urgency of need for data outputs (e.g., to determine whether to execute real-time processing or delayed processing), etc.
In at least one embodiment, services 4120 leveraged by and shared by applications or containers in deployment system 4106 may include compute services 4216, AI services 4218, visualization services 4220, and/or other service types. In at least one embodiment, applications may call (e.g., execute) one or more of services 4120 to perform processing operations for an application. In at least one embodiment, compute services 4216 may be leveraged by applications to perform super-computing or other high-performance computing (HPC) tasks. In at least one embodiment, compute service(s) 4216 may be leveraged to perform parallel processing (e.g., using a parallel computing platform 4230) for processing data through one or more of applications and/or one or more tasks of a single application, substantially simultaneously. In at least one embodiment, parallel computing platform 4230 (e.g., NVIDIA's CUDA) may enable general purpose computing on GPUs (GPGPU) (e.g., GPUs 4222). In at least one embodiment, a software layer of parallel computing platform 4230 may provide access to virtual instruction sets and parallel computational elements of GPUs, for execution of compute kernels. In at least one embodiment, parallel computing platform 4230 may include memory and, in some embodiments, a memory may be shared between and among multiple containers, and/or between and among different processing tasks within a single container. In at least one embodiment, inter-process communication (IPC) calls may be generated for multiple containers and/or for multiple processes within a container to use same data from a shared segment of memory of parallel computing platform 4230 (e.g., where multiple different stages of an application or multiple applications are processing same information). In at least one embodiment, rather than making a copy of data and moving data to different locations in memory (e.g., a read/write operation), same data in same location of a memory may be used for any number of processing tasks (e.g., at a same time, at different times, etc.). In at least one embodiment, as data is used to generate new data as a result of processing, this information of a new location of data may be stored and shared between various applications. In at least one embodiment, location of data and a location of updated or modified data may be part of a definition of how a payload is understood within containers.
In at least one embodiment, AI services 4218 may be leveraged to perform inferencing services for executing machine learning model(s) associated with applications (e.g., tasked with performing one or more processing tasks of an application). In at least one embodiment, AI services 4218 may leverage AI system 4224 to execute machine learning model(s) (e.g., neural networks, such as CNNs) for segmentation, reconstruction, object detection, feature detection, classification, and/or other inferencing tasks. In at least one embodiment, applications of deployment pipeline(s) 4210 may use one or more of output models 4116 from training system 4104 and/or other models of applications to perform inference on imaging data (e.g., DICOM data, RIS data, CIS data, REST compliant data, RPC data, raw data, etc.). In at least one embodiment, two or more examples of inferencing using application orchestration system 4228 (e.g., a scheduler) may be available. In at least one embodiment, a first category may include a high priority/low latency path that may achieve higher service level agreements, such as for performing inference on urgent requests during an emergency, or for a radiologist during diagnosis. In at least one embodiment, a second category may include a standard priority path that may be used for requests that may be non-urgent or where analysis may be performed at a later time. In at least one embodiment, application orchestration system 4228 may distribute resources (e.g., services 4120 and/or hardware 4122) based on priority paths for different inferencing tasks of AI services 4218.
In at least one embodiment, shared storage may be mounted to AI services 4218 within system 4200. In at least one embodiment, shared storage may operate as a cache (or other storage device type) and may be used to process inference requests from applications. In at least one embodiment, when an inference request is submitted, a request may be received by a set of API instances of deployment system 4106, and one or more instances may be selected (e.g., for best fit, for load balancing, etc.) to process a request. In at least one embodiment, to process a request, a request may be entered into a database, a machine learning model may be located from model registry 4124 if not already in a cache, a validation step may ensure appropriate machine learning model is loaded into a cache (e.g., shared storage), and/or a copy of a model may be saved to a cache. In at least one embodiment, a scheduler (e.g., of pipeline manager 4212) may be used to launch an application that is referenced in a request if an application is not already running or if there are not enough instances of an application. In at least one embodiment, if an inference server is not already launched to execute a model, an inference server may be launched. In at least one embodiment, any number of inference servers may be launched per model. In at least one embodiment, in a pull model, in which inference servers are clustered, models may be cached whenever load balancing is advantageous. In at least one embodiment, inference servers may be statically loaded in corresponding, distributed servers.
In at least one embodiment, inferencing may be performed using an inference server that runs in a container. In at least one embodiment, an instance of an inference server may be associated with a model (and optionally a plurality of versions of a model). In at least one embodiment, if an instance of an inference server does not exist when a request to perform inference on a model is received, a new instance may be loaded. In at least one embodiment, when starting an inference server, a model may be passed to an inference server such that a same container may be used to serve different models so long as inference server is running as a different instance.
In at least one embodiment, during application execution, an inference request for a given application may be received, and a container (e.g., hosting an instance of an inference server) may be loaded (if not already), and a start procedure may be called. In at least one embodiment, pre-processing logic in a container may load, decode, and/or perform any additional pre-processing on incoming data (e.g., using a CPU(s) and/or GPU(s)). In at least one embodiment, once data is prepared for inference, a container may perform inference as necessary on data. In at least one embodiment, this may include a single inference call on one image (e.g., a hand X-ray), or may require inference on hundreds of images (e.g., a chest CT). In at least one embodiment, an application may summarize results before completing, which may include, without limitation, a single confidence score, pixel level-segmentation, voxel-level segmentation, generating a visualization, or generating text to summarize findings. In at least one embodiment, different models or applications may be assigned different priorities. For example, some models may have a real-time (TAT less than one minute) priority while others may have lower priority (e.g., TAT less than 10 minutes). In at least one embodiment, model execution times may be measured from requesting institution or entity and may include partner network traversal time, as well as execution on an inference service.
In at least one embodiment, transfer of requests between services 4120 and inference applications may be hidden behind a software development kit (SDK), and robust transport may be provided through a queue. In at least one embodiment, a request will be placed in a queue via an API for an individual application/tenant ID combination and an SDK will pull a request from a queue and give a request to an application. In at least one embodiment, a name of a queue may be provided in an environment from where an SDK will pick it up. In at least one embodiment, asynchronous communication through a queue may be useful as it may allow any instance of an application to pick up work as it becomes available. In at least one embodiment, results may be transferred back through a queue, to ensure no data is lost. In at least one embodiment, queues may also provide an ability to segment work, as highest priority work may go to a queue with most instances of an application connected to it, while lowest priority work may go to a queue with a single instance connected to it that processes tasks in an order received. In at least one embodiment, an application may run on a GPU-accelerated instance generated in cloud 4226, and an inference service may perform inferencing on a GPU.
In at least one embodiment, visualization services 4220 may be leveraged to generate visualizations for viewing outputs of applications and/or deployment pipeline(s) 4210. In at least one embodiment, GPUs 4222 may be leveraged by visualization services 4220 to generate visualizations. In at least one embodiment, rendering effects, such as ray-tracing, may be implemented by visualization services 4220 to generate higher quality visualizations. In at least one embodiment, visualizations may include, without limitation, 2D image renderings, 3D volume renderings, 3D volume reconstruction, 2D tomographic slices, virtual reality displays, augmented reality displays, etc. In at least one embodiment, virtualized environments may be used to generate a virtual interactive display or environment (e.g., a virtual environment) for interaction by users of a system (e.g., doctors, nurses, radiologists, etc.). In at least one embodiment, visualization services 4220 may include an internal visualizer, cinematics, and/or other rendering or image processing capabilities or functionality (e.g., ray tracing, rasterization, internal optics, etc.).
In at least one embodiment, hardware 4122 may include GPUs 4222, AI system 4224, cloud 4226, and/or any other hardware used for executing training system 4104 and/or deployment system 4106. In at least one embodiment, GPUs 4222 (e.g., NVIDIA's TESLA and/or QUADRO GPUs) may include any number of GPUs that may be used for executing processing tasks of compute services 4216, AI services 4218, visualization services 4220, other services, and/or any of features or functionality of software 4118. For example, with respect to AI services 4218, GPUs 4222 may be used to perform pre-processing on imaging data (or other data types used by machine learning models), post-processing on outputs of machine learning models, and/or to perform inferencing (e.g., to execute machine learning models). In at least one embodiment, cloud 4226, AI system 4224, and/or other components of system 4200 may use GPUs 4222. In at least one embodiment, cloud 4226 may include a GPU-optimized platform for deep learning tasks. In at least one embodiment, AI system 4224 may use GPUs, and cloud 4226—or at least a portion tasked with deep learning or inferencing—may be executed using one or more AI systems 4224. As such, although hardware 4122 is illustrated as discrete components, this is not intended to be limiting, and any components of hardware 4122 may be combined with, or leveraged by, any other components of hardware 4122.
In at least one embodiment, AI system 4224 may include a purpose-built computing system (e.g., a super-computer or an HPC) configured for inferencing, deep learning, machine learning, and/or other artificial intelligence tasks. In at least one embodiment, AI system 4224 (e.g., NVIDIA's DGX) may include GPU-optimized software (e.g., a software stack) that may be executed using a plurality of GPUs 4222, in addition to CPUs, RAM, storage, and/or other components, features, or functionality. In at least one embodiment, one or more AI systems 4224 may be implemented in cloud 4226 (e.g., in a data center) for performing some or all of AI-based processing tasks of system 4200.
In at least one embodiment, cloud 4226 may include a GPU-accelerated infrastructure (e.g., NVIDIA's NGC) that may provide a GPU-optimized platform for executing processing tasks of system 4200. In at least one embodiment, cloud 4226 may include an AI system(s) 4224 for performing one or more of AI-based tasks of system 4200 (e.g., as a hardware abstraction and scaling platform). In at least one embodiment, cloud 4226 may integrate with application orchestration system 4228 leveraging multiple GPUs to enable seamless scaling and load balancing between and among applications and services 4120. In at least one embodiment, cloud 4226 may tasked with executing at least some of services 4120 of system 4200, including compute services 4216, AI services 4218, and/or visualization services 4220, as described herein. In at least one embodiment, cloud 4226 may perform small and large batch inference (e.g., executing NVIDIA's TENSOR RT), provide an accelerated parallel computing API and platform 4230 (e.g., NVIDIA's CUDA), execute application orchestration system 4228 (e.g., KUBERNETES), provide a graphics rendering API and platform (e.g., for ray-tracing, 2D graphics, 3D graphics, and/or other rendering techniques to produce higher quality cinematics), and/or may provide other functionality for system 4200.
In at least one embodiment, in an effort to preserve patient confidentiality (e.g., where patient data or records are to be used off-premises), cloud 4226 may include a registry—such as a deep learning container registry. In at least one embodiment, a registry may store containers for instantiations of applications that may perform pre-processing, post-processing, or other processing tasks on patient data. In at least one embodiment, cloud 4226 may receive data that includes patient data as well as sensor data in containers, perform requested processing for just sensor data in those containers, and then forward a resultant output and/or visualizations to appropriate parties and/or devices (e.g., on-premises medical devices used for visualization or diagnoses), all without having to extract, store, or otherwise access patient data. In at least one embodiment, confidentiality of patient data is preserved in compliance with HIPAA and/or other data regulations.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, deployment pipeline 4210A of
In at least one embodiment, CT reconstruction 4308 application and/or container may be executed once data (e.g., raw sinogram data) is available for processing by CT reconstruction 4308 application. In at least one embodiment, CT reconstruction 4308 may read raw sinogram data from a cache, reconstruct an image file out of raw sinogram data (e.g., as illustrated in visualization 4316B), and store resulting image file in a cache. In at least one embodiment, at completion of reconstruction, pipeline manager 4212 may be signaled that reconstruction task is complete. In at least one embodiment, once reconstruction is complete, and a reconstructed image file may be stored in a cache (or other storage device), organ segmentation 4310 application and/or container may be triggered by pipeline manager 4212. In at least one embodiment, organ segmentation 4310 application and/or container may read an image file from a cache, normalize or convert an image file to format suitable for inference (e.g., convert an image file to an input resolution of a machine learning model), and run inference against a normalized image. In at least one embodiment, to run inference on a normalized image, organ segmentation 4310 application and/or container may rely on services 4120, and pipeline manager 4212 and/or application orchestration system 4228 may facilitate use of services 4120 by organ segmentation 4310 application and/or container. In at least one embodiment, for example, organ segmentation 4310 application and/or container may leverage AI services 4218 to perform inference on a normalized image, and AI services 4218 may leverage hardware 4122 (e.g., AI system 4224) to execute AI services 4218. In at least one embodiment, a result of an inference may be a mask file (e.g., as illustrated in visualization 4316C) that may be stored in a cache (or other storage device).
In at least one embodiment, once applications that process DICOM data and/or data extracted from DICOM data have completed processing, a signal may be generated for pipeline manager 4212. In at least one embodiment, pipeline manager 4212 may then execute DICOM writer 4312 to read results from a cache (or other storage device), package results into a DICOM format (e.g., as DICOM output 4314) for use by users at a facility who generated a request. In at least one embodiment, DICOM output 4314 may then be transmitted to DICOM adapter 4202B to prepare DICOM output 4314 for storage on PACS server(s) 4304 (e.g., for viewing by a DICOM viewer at a facility). In at least one embodiment, in response to a request for reconstruction and segmentation, visualizations 4316B and 4316C may be generated and available to a user for diagnoses, research, and/or for other purposes.
Although illustrated as consecutive application in deployment pipeline 4210A, CT reconstruction 4308 and organ segmentation 4310 applications may be processed in parallel in at least one embodiment. In at least one embodiment, where applications do not have dependencies on one another, and data is available for each application (e.g., after DICOM reader 4306 extracts data), applications may be executed at a same time, substantially at a same time, or with some overlap. In at least one embodiment, where two or more applications require similar services 4120, a scheduler of system 4200 may be used to load balance and distribute compute or processing resources between and among various applications. In at least one embodiment, in some embodiments, parallel computing platform 4230 may be used to perform parallel processing for applications to decrease run-time of deployment pipeline 4210A to provide real-time results.
In at least one embodiment, and with reference to
In at least one embodiment, system 4200 may be instantiated or executed as one or more virtual instruments on-premise at a facility in, for example, a computing system deployed next to or otherwise in communication with a radiology machine, an imaging device, and/or another device type at a facility. In at least one embodiment, however, an on-premise installation may be instantiated or executed within a computing system of a device itself (e.g., a computing system integral to an imaging device), in a local datacenter (e.g., a datacenter on-premise), and/or in a cloud-environment (e.g., in cloud 4226). In at least one embodiment, deployment system 4106, operating as a virtual instrument, may be instantiated by a supercomputer or other HPC system in some examples. In at least one embodiment, on-premise installation may allow for high-bandwidth uses (via, for example, higher throughput local communication interfaces, such as RF over Ethernet) for real-time processing. In at least one embodiment, real-time or near real-time processing may be particularly useful where a virtual instrument supports an ultrasound device or other imaging modality where immediate visualizations are expected or required for accurate diagnoses and analyses. In at least one embodiment, a cloud-computing architecture may be capable of dynamic bursting to a cloud computing service provider, or other compute cluster, when local demand exceeds on-premise capacity or capability. In at least one embodiment, a cloud architecture, when implemented, may be tuned for training neural networks or other machine learning models, as described herein with respect to training system 4104. In at least one embodiment, with training pipelines in place, machine learning models may be continuously learn and improve as they process additional data from devices they support. In at least one embodiment, virtual instruments may be continually improved using additional data, new data, existing machine learning models, and/or new or updated machine learning models.
In at least one embodiment, a computing system may include some or all of hardware 4122 described herein, and hardware 4122 may be distributed in any of a number of ways including within a device, as part of a computing device coupled to and located proximate a device, in a local datacenter at a facility, and/or in cloud 4226. In at least one embodiment, because deployment system 4106 and associated applications or containers are created in software (e.g., as discrete containerized instantiations of applications), behavior, operation, and configuration of virtual instruments, as well as outputs generated by virtual instruments, may be modified or customized as desired, without having to change or alter raw output of a device that a virtual instrument supports.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, process 4400 may include receipt of imaging data from an ultrasound device 4402. In at least one embodiment, imaging data may be stored on PACS server(s) in a DICOM format (or other format, such as RIS, CIS, REST compliant, RPC, raw, etc.), and may be received by system 4200 for processing through deployment pipeline 4210 selected or customized as a virtual instrument (e.g., a virtual ultrasound) for ultrasound device 4402. In at least one embodiment, imaging data may be received directly from an imaging device (e.g., ultrasound device 4402) and processed by a virtual instrument. In at least one embodiment, a transducer or other signal converter communicatively coupled between an imaging device and a virtual instrument may convert signal data generated by an imaging device to image data that may be processed by a virtual instrument. In at least one embodiment, raw data and/or image data may be applied to DICOM reader 4306 to extract data for use by applications or containers of deployment pipeline 4210B. In at least one embodiment, DICOM reader 4306 may leverage data augmentation library 4414 (e.g., NVIDIA's DALI) as a service 4120 (e.g., as one of compute service(s) 4216) for extracting, resizing, rescaling, and/or otherwise preparing data for use by applications or containers.
In at least one embodiment, once data is prepared, a reconstruction 4406 application and/or container may be executed to reconstruct data from ultrasound device 4402 into an image file. In at least one embodiment, after reconstruction 4406, or at a same time as reconstruction 4406, a detection 4408 application and/or container may be executed for anomaly detection, object detection, feature detection, and/or other detection tasks related to data. In at least one embodiment, an image file generated during reconstruction 4406 may be used during detection 4408 to identify anomalies, objects, features, etc. In at least one embodiment, detection 4408 application may leverage an inference engine 4416 (e.g., as one of AI service(s) 4218) to perform inference on data to generate detections. In at least one embodiment, one or more machine learning models (e.g., from training system 4104) may be executed or called by detection 4408 application.
In at least one embodiment, once reconstruction 4406 and/or detection 4408 is/are complete, data output from these application and/or containers may be used to generate visualizations 4410, such as visualization 4412 (e.g., a grayscale output) displayed on a workstation or display terminal. In at least one embodiment, visualization may allow a technician or other user to visualize results of deployment pipeline 4210B with respect to ultrasound device 4402. In at least one embodiment, visualization 4410 may be executed by leveraging a render component 4418 of system 4200 (e.g., one of visualization service(s) 4220). In at least one embodiment, render component 4418 may execute a 2D, OpenGL, or ray-tracing service to generate visualization 4412.
In at least one embodiment, process 4420 may include CT scanner 4422 generating raw data that may be received by DICOM reader 4306 (e.g., directly, via a PACS server 4304, after processing, etc.). In at least one embodiment, a Virtual CT (instantiated by deployment pipeline 4210C) may include a first, real-time pipeline for monitoring a patient (e.g., patient movement detection AI 4426) and/or for adjusting or optimizing exposure of CT scanner 4422 (e.g., using exposure control AI 4424). In at least one embodiment, one or more of applications (e.g., 4424 and 4426) may leverage a service 4120, such as AI service(s) 4218. In at least one embodiment, outputs of exposure control AI 4424 application (or container) and/or patient movement detection AI 4426 application (or container) may be used as feedback to CT scanner 4422 and/or a technician for adjusting exposure (or other settings of CT scanner 4422) and/or informing a patient to move less.
In at least one embodiment, deployment pipeline 4210C may include a non-real-time pipeline for analyzing data generated by CT scanner 4422. In at least one embodiment, a second pipeline may include CT reconstruction 4308 application and/or container, a coarse detection AI 4428 application and/or container, a fine detection AI 4432 application and/or container (e.g., where certain results are detected by coarse detection AI 4428), a visualization 4430 application and/or container, and a DICOM writer 4312 (and/or other data type writer, such as RIS, CIS, REST compliant, RPC, raw, etc.) application and/or container. In at least one embodiment, raw data generated by CT scanner 4422 may be passed through pipelines of deployment pipeline 4210C (instantiated as a virtual CT instrument) to generate results. In at least one embodiment, results from DICOM writer 4312 may be transmitted for display and/or may be stored on PACS server(s) 4304 for later retrieval, analysis, or display by a technician, practitioner, or other user.
In at least one embodiment, one or more systems depicted in
In at least one embodiment, model training 4114 may include retraining or updating an initial model 4504 (e.g., a pre-trained model) using new training data (e.g., new input data, such as customer dataset 4506, and/or new ground truth data associated with input data). In at least one embodiment, to retrain, or update, initial model 4504, output or loss layer(s) of initial model 4504 may be reset, or deleted, and/or replaced with an updated or new output or loss layer(s). In at least one embodiment, initial model 4504 may have previously fine-tuned parameters (e.g., weights and/or biases) that remain from prior training, so training or retraining 4114 may not take as long or require as much processing as training a model from scratch. In at least one embodiment, during model training 4114, by having reset or replaced output or loss layer(s) of initial model 4504, parameters may be updated and re-tuned for a new data set based on loss calculations associated with accuracy of output or loss layer(s) at generating predictions on new, customer dataset 4506 (e.g., image data 4108 of
In at least one embodiment, pre-trained models 4206 may be stored in a data store, or registry (e.g., model registry 4124 of
In at least one embodiment, when selecting applications for use in deployment pipelines 4210, a user may also select machine learning models to be used for specific applications. In at least one embodiment, a user may not have a model for use, so a user may select a pre-trained model 4206 to use with an application. In at least one embodiment, pre-trained model 4206 may not be optimized for generating accurate results on customer dataset 4506 of a facility of a user (e.g., based on patient diversity, demographics, types of medical imaging devices used, etc.). In at least one embodiment, prior to deploying pre-trained model 4206 into deployment pipeline 4210 for use with an application(s), pre-trained model 4206 may be updated, retrained, and/or fine-tuned for use at a respective facility.
In at least one embodiment, a user may select pre-trained model 4206 that is to be updated, retrained, and/or fine-tuned, and pre-trained model 4206 may be referred to as initial model 4504 for training system 4104 within process 4500. In at least one embodiment, customer dataset 4506 (e.g., imaging data, genomics data, sequencing data, or other data types generated by devices at a facility) may be used to perform model training 4114 (which may include, without limitation, transfer learning) on initial model 4504 to generate refined model 4512. In at least one embodiment, ground truth data corresponding to customer dataset 4506 may be generated by training system 4104. In at least one embodiment, ground truth data may be generated, at least in part, by clinicians, scientists, doctors, practitioners, at a facility (e.g., as labeled clinic data 4112 of
In at least one embodiment, AI-assisted annotation 4110 may be used in some examples to generate ground truth data. In at least one embodiment, AI-assisted annotation 4110 (e.g., implemented using an AI-assisted annotation SDK) may leverage machine learning models (e.g., neural networks) to generate suggested or predicted ground truth data for a customer dataset. In at least one embodiment, user 4510 may use annotation tools within a user interface (a graphical user interface (GUI)) on computing device 4508.
In at least one embodiment, user 4510 may interact with a GUI via computing device 4508 to edit or fine-tune annotations or auto-annotations. In at least one embodiment, a polygon editing feature may be used to move vertices of a polygon to more accurate or fine-tuned locations.
In at least one embodiment, once customer dataset 4506 has associated ground truth data, ground truth data (e.g., from AI-assisted annotation, manual labeling, etc.) may be used by during model training 4114 to generate refined model 4512. In at least one embodiment, customer dataset 4506 may be applied to initial model 4504 any number of times, and ground truth data may be used to update parameters of initial model 4504 until an acceptable level of accuracy is attained for refined model 4512. In at least one embodiment, once refined model 4512 is generated, refined model 4512 may be deployed within one or more deployment pipelines 4210 at a facility for performing one or more processing tasks with respect to medical imaging data.
In at least one embodiment, refined model 4512 may be uploaded to pre-trained models 4206 in model registry 4124 to be selected by another facility. In at least one embodiment, his process may be completed at any number of facilities such that refined model 4512 may be further refined on new datasets any number of times to generate a more universal model.
Inference and/or training logic 1215 are used to perform inferencing and/or training operations associated with one or more embodiments. Details regarding inference and/or training logic 1215 are provided herein in conjunction with
In at least one embodiment, one or more systems depicted in
At least one embodiment of the disclosure can be described in view of the following clauses:
Clause 1. A processor, comprising:
one or more circuits to use one or more neural networks to calculate a plurality of paths, through which an autonomous device is to traverse.
Clause 2. The processor of clause 1, wherein the one or more circuits are to use the one or more neural networks to calculate the plurality of paths by at least:
obtaining a first location, a set of locations, and a final location;
causing the one or more neural networks to calculate a set of distances based at least in part on the set of locations and the final location; and
calculating the plurality of paths based at least in part on the set of distances, wherein the plurality of paths form a path from the first location to the final location.
Clause 3. The processor of any of clauses 1-2, wherein:
the first location is a location of the autonomous device; and
a subset of locations of the set of locations are accessible to the autonomous device from the first location.
Clause 4. The processor of any of clauses 1-3, wherein the one or more circuits are to calculate a first path of the plurality of paths by at least:
obtaining a subset of distances of the set of distances corresponding to the subset of locations;
selecting a second location of the subset of locations based at least in part on the subset of distances; and
calculating the first path comprising a path from the first location to the second location.
Clause 5. The processor of any of clauses 1-4, wherein the second location corresponds to a minimum distance of the subset of distances.
Clause 6. The processor of any of clauses 1-5, wherein the one or more neural networks calculate the set of distances in a single forward pass.
Clause 7. The processor of any of clauses 1-6, wherein a distance of the set of distances corresponds to a distance along a path from a location of the set of locations to the final location.
Clause 8. A machine-readable medium having stored thereon a set of instructions, which if performed by one or more processors, cause the one or more processors to use one or more neural networks to calculate a plurality of paths, through which an autonomous device is to traverse.
Clause 9. The machine-readable medium of clause 8, wherein the set of instructions further comprise instructions, which if performed by the one or more processors, cause the one or more processors to:
obtain features of an environment;
select a location in the environment; and
input at least the features and the location to the one or more neural networks to obtain a plurality of distances corresponding to a plurality of locations in the environment.
Clause 10. The machine-readable medium of any of clauses 8-9, wherein the set of instructions further comprise instructions, which if performed by the one or more processors, cause the one or more processors to:
select a set of locations accessible to the autonomous device;
obtain a set of distances of the plurality of distances corresponding to the set of locations; and
select a first location of the set of locations based at least in part on the set of distances, wherein a first path of the plurality of paths indicates a path from the autonomous device to the first location.
Clause 11. The machine-readable medium of any of clauses 8-10, wherein the set of instructions further comprise instructions, which if performed by the one or more processors, cause the one or more processors to cause the autonomous device to navigate to the first location using the first path.
Clause 12. The machine-readable medium of any of clauses 8-11, wherein the autonomous device is an autonomous car.
Clause 13. The machine-readable medium of any of clauses 8-12, wherein the features are generated by one or more encoders based on a representation of the environment.
Clause 14. The machine-readable medium of any of clauses 8-13, wherein the representation of the environment is an image or a point cloud.
Clause 15. A system, comprising:
one or more computers having one or more processors to use one or more neural networks to calculate a plurality of paths, through which an autonomous device is to traverse.
Clause 16. The system of clause 15, wherein the one or more processors are further to:
capture a representation of an environment; and
use the one or more neural networks to calculate the plurality of paths from a first location of the autonomous device to a second location in the environment.
Clause 17. The system of any of clauses 15-16, wherein the one or more processors are further to use the one or more neural networks to calculate one or more distance values for one or more locations in the environment based at least in part on the representation of the environment.
Clause 18. The system of any of clauses 15-17, wherein the one or more processors are further to:
calculate a size for a step of the autonomous device;
select a set of locations accessible through the step from the first location of the autonomous device; and
select a third location of the set of locations based at least in part on the one or more distance values.
Clause 19. The system of any of clauses 15-18, wherein the representation of the environment is captured through one or more depth cameras.
Clause 20. The system of any of clauses 15-19, wherein the representation of the environment is a 2D or 3D representation.
Clause 21. The system of any of clauses 15-20, wherein the autonomous device is an autonomous robot.
Clause 22. A machine-readable medium having stored thereon a set of instructions, which if performed by one or more processors, cause the one or more processors to at least:
train one or more neural networks to calculate a plurality of paths, through which an autonomous device is to traverse.
Clause 23. The machine-readable medium of clause 22, wherein the set of instructions further include instructions, which if performed by the one or more processors, cause the one or more processors to:
obtain an environment and a location;
cause one or more algorithms to determine one or more reaching distance values for one or more locations in the environment to the location; and
train the one or more neural networks at least using the one or more reaching distance values.
Clause 24. The machine-readable medium of any of clauses 22-23, wherein the set of instructions further include instructions, which if performed by the one or more processors, cause the one or more processors to:
cause the one or more neural networks to process at least the one or more locations to calculate one or more predicted reaching distance values; and
update the one or more neural networks based at least in part on differences between the one or more predicted reaching distance values and the one or more reaching distance values.
Clause 25. The machine-readable medium of any of clauses 22-24, wherein the one or more algorithms include one or more fast marching method (FMM) algorithms.
Clause 26. The machine-readable medium of any of clauses 22-25, wherein a first reaching distance value of the one or more reaching distance values corresponds to a first location of the one or more locations and indicates a distance along a path from the first location to the location.
Clause 27. The machine-readable medium of any of clauses 22-26, wherein the path is a geometrically feasible path.
Clause 28. A processor comprising:
one or more circuits to train one or more neural networks to calculate a plurality of paths, through which an autonomous device is to traverse.
Clause 29. The processor of clause 28, wherein the one or more circuits are further to:
cause one or more algorithms to process at least an environment, a set of positions, and a goal position;
obtain a set of distance values based at least in part on results of the one or more algorithms; and
train the one or more neural networks using the set of distance values.
Clause 30. The processor of any of clauses 28-29, wherein the one or more circuits are further to train the one or more neural networks to process at least the environment, the set of positions, and the goal position to calculate the set of distance values.
Clause 31. The processor of any of clauses 28-30, wherein a first distance value of the set of distance values indicates a distance along a semantically feasible path from a first position of the set of positions to the goal position.
Clause 32. The processor of any of clauses 28-31, wherein the one or more algorithms include one or more path planning algorithms.
Clause 33. The processor of any of clauses 28-32, wherein the one or more neural networks include one or more implicit environment functions.
In at least one embodiment, a single semiconductor platform may refer to a sole unitary semiconductor-based integrated circuit or chip. In at least one embodiment, multi-chip modules may be used with increased connectivity which simulate on-chip operation, and make substantial improvements over utilizing a conventional central processing unit (“CPU”) and bus implementation. In at least one embodiment, various modules may also be situated separately or in various combinations of semiconductor platforms per desires of user.
In at least one embodiment, referring back to
In at least one embodiment, architecture and/or functionality of various previous figures are implemented in context of a general computer system, a circuit board system, a game console system dedicated for entertainment purposes, an application-specific system, and more. In at least one embodiment, computer system 1800 may take form of a desktop computer, a laptop computer, a tablet computer, servers, supercomputers, a smart-phone (e.g., a wireless, hand-held device), personal digital assistant (“PDA”), a digital camera, a vehicle, a head mounted display, a hand-held electronic device, a mobile phone device, a television, workstation, game consoles, embedded system, and/or any other type of logic.
In at least one embodiment, parallel processing system 1812 includes, without limitation, a plurality of parallel processing units (“PPUs”) 1814 and associated memories 1816. In at least one embodiment, PPUs 1814 are connected to a host processor or other peripheral devices via an interconnect 1818 and a switch 1820 or multiplexer. In at least one embodiment, parallel processing system 1812 distributes computational tasks across PPUs 1814 which can be parallelizable—for example, as part of distribution of computational tasks across multiple graphics processing unit (“GPU”) thread blocks. In at least one embodiment, memory is shared and accessible (e.g., for read and/or write access) across some or all of PPUs 1814, although such shared memory may incur performance penalties relative to use of local memory and registers resident to a PPU 1814. In at least one embodiment, operation of PPUs 1814 is synchronized through use of a command such as _syncthreads( ), wherein all threads in a block (e.g., executed across multiple PPUs 1814) to reach a certain point of execution of code before proceeding.
Other variations are within spirit of present disclosure. Thus, while disclosed techniques are susceptible to various modifications and alternative constructions, certain illustrated embodiments thereof are shown in drawings and have been described above in detail. It should be understood, however, that there is no intention to limit disclosure to specific form or forms disclosed, but on contrary, intention is to cover all modifications, alternative constructions, and equivalents falling within spirit and scope of disclosure, as defined in appended claims.
Use of terms “a” and “an” and “the” and similar referents in context of describing disclosed embodiments (especially in context of following claims) are to be construed to cover both singular and plural, unless otherwise indicated herein or clearly contradicted by context, and not as a definition of a term. Terms “comprising,” “having,” “including,” and “containing” are to be construed as open-ended terms (meaning “including, but not limited to,”) unless otherwise noted. “Connected,” when unmodified and referring to physical connections, is to be construed as partly or wholly contained within, attached to, or joined together, even if there is something intervening. Recitation of ranges of values herein are merely intended to serve as a shorthand method of referring individually to each separate value falling within range, unless otherwise indicated herein and each separate value is incorporated into specification as if it were individually recited herein. In at least one embodiment, use of term “set” (e.g., “a set of items”) or “subset” unless otherwise noted or contradicted by context, is to be construed as a nonempty collection comprising one or more members. Further, unless otherwise noted or contradicted by context, term “subset” of a corresponding set does not necessarily denote a proper subset of corresponding set, but subset and corresponding set may be equal.
Conjunctive language, such as phrases of form “at least one of A, B, and C,” or “at least one of A, B and C,” unless specifically stated otherwise or otherwise clearly contradicted by context, is otherwise understood with context as used in general to present that an item, term, etc., may be either A or B or C, or any nonempty subset of set of A and B and C. For instance, in illustrative example of a set having three members, conjunctive phrases “at least one of A, B, and C” and “at least one of A, B and C” refer to any of following sets: {A}, {B}, {C}, {A, B}, {A, C}, {B, C}, {A, B, C}. Thus, such conjunctive language is not generally intended to imply that certain embodiments require at least one of A, at least one of B and at least one of C each to be present. In addition, unless otherwise noted or contradicted by context, term “plurality” indicates a state of being plural (e.g., “a plurality of items” indicates multiple items). In at least one embodiment, number of items in a plurality is at least two, but can be more when so indicated either explicitly or by context. Further, unless stated otherwise or otherwise clear from context, phrase “based on” means “based at least in part on” and not “based solely on.”
Operations of processes described herein can be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. In at least one embodiment, a process such as those processes described herein (or variations and/or combinations thereof) is performed under control of one or more computer systems configured with executable instructions and is implemented as code (e.g., executable instructions, one or more computer programs or one or more applications) executing collectively on one or more processors, by hardware or combinations thereof. In at least one embodiment, code is stored on a computer-readable storage medium, for example, in form of a computer program comprising a plurality of instructions executable by one or more processors. In at least one embodiment, a computer-readable storage medium is a non-transitory computer-readable storage medium that excludes transitory signals (e.g., a propagating transient electric or electromagnetic transmission) but includes non-transitory data storage circuitry (e.g., buffers, cache, and queues) within transceivers of transitory signals. In at least one embodiment, code (e.g., executable code or source code) is stored on a set of one or more non-transitory computer-readable storage media having stored thereon executable instructions (or other memory to store executable instructions) that, when executed (e.g., as a result of being executed) by one or more processors of a computer system, cause computer system to perform operations described herein. In at least one embodiment, set of non-transitory computer-readable storage media comprises multiple non-transitory computer-readable storage media and one or more of individual non-transitory storage media of multiple non-transitory computer-readable storage media lack all of code while multiple non-transitory computer-readable storage media collectively store all of code. In at least one embodiment, executable instructions are executed such that different instructions are executed by different processors—for example, a non-transitory computer-readable storage medium store instructions and a main central processing unit (“CPU”) executes some of instructions while a graphics processing unit (“GPU”) executes other instructions. In at least one embodiment, different components of a computer system have separate processors and different processors execute different subsets of instructions.
In at least one embodiment, an arithmetic logic unit is a set of combinational logic circuitry that takes one or more inputs to produce a result. In at least one embodiment, an arithmetic logic unit is used by a processor to implement mathematical operation such as addition, subtraction, or multiplication. In at least one embodiment, an arithmetic logic unit is used to implement logical operations such as logical AND/OR or XOR. In at least one embodiment, an arithmetic logic unit is stateless, and made from physical switching components such as semiconductor transistors arranged to form logical gates. In at least one embodiment, an arithmetic logic unit may operate internally as a stateful logic circuit with an associated clock. In at least one embodiment, an arithmetic logic unit may be constructed as an asynchronous logic circuit with an internal state not maintained in an associated register set. In at least one embodiment, an arithmetic logic unit is used by a processor to combine operands stored in one or more registers of the processor and produce an output that can be stored by the processor in another register or a memory location.
In at least one embodiment, as a result of processing an instruction retrieved by the processor, the processor presents one or more inputs or operands to an arithmetic logic unit, causing the arithmetic logic unit to produce a result based at least in part on an instruction code provided to inputs of the arithmetic logic unit. In at least one embodiment, the instruction codes provided by the processor to the ALU are based at least in part on the instruction executed by the processor. In at least one embodiment combinational logic in the ALU processes the inputs and produces an output which is placed on a bus within the processor. In at least one embodiment, the processor selects a destination register, memory location, output device, or output storage location on the output bus so that clocking the processor causes the results produced by the ALU to be sent to the desired location.
In the scope of this application, the term arithmetic logic unit, or ALU, is used to refer to any computational logic circuit that processes operands to produce a result. For example, in the present document, the term ALU can refer to a floating point unit, a DSP, a tensor core, a shader core, a coprocessor, or a CPU.
Accordingly, in at least one embodiment, computer systems are configured to implement one or more services that singly or collectively perform operations of processes described herein and such computer systems are configured with applicable hardware and/or software that enable performance of operations. Further, a computer system that implements at least one embodiment of present disclosure is a single device and, in another embodiment, is a distributed computer system comprising multiple devices that operate differently such that distributed computer system performs operations described herein and such that a single device does not perform all operations.
Use of any and all examples, or exemplary language (e.g., “such as”) provided herein, is intended merely to better illuminate embodiments of disclosure and does not pose a limitation on scope of disclosure unless otherwise claimed. No language in specification should be construed as indicating any non-claimed element as essential to practice of disclosure.
All references, including publications, patent applications, and patents, cited herein are hereby incorporated by reference to same extent as if each reference were individually and specifically indicated to be incorporated by reference and were set forth in its entirety herein.
In description and claims, terms “coupled” and “connected,” along with their derivatives, may be used. It should be understood that these terms may be not intended as synonyms for each other. Rather, in particular examples, “connected” or “coupled” may be used to indicate that two or more elements are in direct or indirect physical or electrical contact with each other. “Coupled” may also mean that two or more elements are not in direct contact with each other, but yet still co-operate or interact with each other.
Unless specifically stated otherwise, it may be appreciated that throughout specification terms such as “processing,” “computing,” “calculating,” “determining,” or like, refer to action and/or processes of a computer or computing system, or similar electronic computing device, that manipulate and/or transform data represented as physical, such as electronic, quantities within computing system's registers and/or memories into other data similarly represented as physical quantities within computing system's memories, registers or other such information storage, transmission or display devices.
In a similar manner, term “processor” may refer to any device or portion of a device that processes electronic data from registers and/or memory and transform that electronic data into other electronic data that may be stored in registers and/or memory. As non-limiting examples, “processor” may be a CPU or a GPU. A “computing platform” may comprise one or more processors. As used herein, “software” processes may include, for example, software and/or hardware entities that perform work over time, such as tasks, threads, and intelligent agents. Also, each process may refer to multiple processes, for carrying out instructions in sequence or in parallel, continuously or intermittently. In at least one embodiment, terms “system” and “method” are used herein interchangeably insofar as system may embody one or more methods and methods may be considered a system.
In present document, references may be made to obtaining, acquiring, receiving, or inputting analog or digital data into a subsystem, computer system, or computer-implemented machine. In at least one embodiment, process of obtaining, acquiring, receiving, or inputting analog and digital data can be accomplished in a variety of ways such as by receiving data as a parameter of a function call or a call to an application programming interface. In at least one embodiment, processes of obtaining, acquiring, receiving, or inputting analog or digital data can be accomplished by transferring data via a serial or parallel interface. In at least one embodiment, processes of obtaining, acquiring, receiving, or inputting analog or digital data can be accomplished by transferring data via a computer network from providing entity to acquiring entity. In at least one embodiment, references may also be made to providing, outputting, transmitting, sending, or presenting analog or digital data. In various examples, processes of providing, outputting, transmitting, sending, or presenting analog or digital data can be accomplished by transferring data as an input or output parameter of a function call, a parameter of an application programming interface or interprocess communication mechanism.
Although descriptions herein set forth example implementations of described techniques, other architectures may be used to implement described functionality, and are intended to be within scope of this disclosure. Furthermore, although specific distributions of responsibilities may be defined above for purposes of description, various functions and responsibilities might be distributed and divided in different ways, depending on circumstances.
Furthermore, although subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that subject matter claimed in appended claims is not necessarily limited to specific features or acts described. Rather, specific features and acts are disclosed as exemplary forms of implementing the claims.
