Methods and example implementations described herein are directed to interconnect architecture, and more specifically, implementing Quality of Service (QoS) in a system with end-to-end flow control and QoS aware buffer allocation.
The number of components on a chip is rapidly growing due to increasing levels of integration, system complexity and shrinking transistor geometry. Complex System-on-Chips (SoCs) may involve a variety of components e.g., processor cores, DSPs, hardware accelerators, memory and I/O, while Chip Multi-Processors (CMPs) may involve a large number of homogenous processor cores, memory and I/O subsystems. In both SoC and CMP systems, the on-chip interconnect plays a role in providing high-performance communication between the various components. Due to scalability limitations of traditional buses and crossbar based interconnects, Network-on-Chip (NoC) has emerged as a paradigm to interconnect a large number of components on the chip. NoC is a global shared communication infrastructure made up of several routing nodes interconnected with each other using point-to-point physical links.
Messages are injected by the source and are routed from the source node to the destination over multiple intermediate nodes and physical links. The destination node then ejects the message and provides the message to the destination. For the remainder of this application, the terms ‘components’, ‘blocks’, ‘hosts’ or ‘cores’ will be used interchangeably to refer to the various system components which are interconnected using a NoC. Terms ‘routers’ and ‘nodes’ will also be used interchangeably. Without loss of generalization, the system with multiple interconnected components will itself be referred to as a ‘multi-core system’.
There are several topologies in which the routers can connect to one another to create the system network. Bi-directional rings (as shown in
Packets are message transport units for intercommunication between various components. Routing involves identifying a path composed of a set of routers and physical links of the network over which packets are sent from a source to a destination. Components are connected to one or multiple ports of one or multiple routers; with each such port having a unique ID. Packets carry the destination's router and port ID for use by the intermediate routers to route the packet to the destination component.
Examples of routing techniques include deterministic routing, which involves choosing the same path from A to B for every packet. This form of routing is independent from the state of the network and does not load balance across path diversities, which might exist in the underlying network. However, such deterministic routing may implemented in hardware, maintains packet ordering and may be rendered free of network level deadlocks. Shortest path routing may minimize the latency as such routing reduces the number of hops from the source to the destination. For this reason, the shortest path may also be the lowest power path for communication between the two components. Dimension-order routing is a form of deterministic shortest path routing in 2-D, 2.5-D, and 3-D mesh networks. In this routing scheme, messages are routed along each coordinates in a particular sequence until the message reaches the final destination. For example in a 3-D mesh network, one may first route along the X dimension until it reaches a router whose X-coordinate is equal to the X-coordinate of the destination router. Next, the message takes a turn and is routed in along Y dimension and finally takes another turn and moves along the Z dimension until the message reaches the final destination router. Dimension ordered routing may be minimal turn and shortest path routing.
In heterogeneous mesh topology in which one or more routers or one or more links are absent, dimension order routing may not be feasible between certain source and destination nodes, and alternative paths may have to be taken. The alternative paths may not be shortest or minimum turn.
Source routing and routing using tables are other routing options used in NoC. Adaptive routing can dynamically change the path taken between two points on the network based on the state of the network. This form of routing may be complex to analyze and implement.
A NoC interconnect may contain multiple physical networks. Over each physical network, there may exist multiple virtual networks, wherein different message types are transmitted over different virtual networks. In this case, at each physical link or channel, there are multiple virtual channels; each virtual channel may have dedicated buffers at both end points. In any given clock cycle, only one virtual channel can transmit data on the physical channel.
NoC interconnects may employ wormhole routing, wherein, a large message or packet is broken into small pieces known as flits (also referred to as flow control digits). The first flit is the header flit, which holds information about this packet's route and key message level info along with payload data and sets up the routing behavior for all subsequent flits associated with the message. Optionally, one or more body flits follows the head flit, containing the remaining payload of data. The final flit is the tail flit, which in addition to containing the last payload also performs some bookkeeping to close the connection for the message. In wormhole flow control, virtual channels are often implemented.
The physical channels are time sliced into a number of independent logical channels called virtual channels (VCs). VCs provide multiple independent paths to route packets, however they are time-multiplexed on the physical channels. A virtual channel holds the state needed to coordinate the handling of the flits of a packet over a channel. At a minimum, this state identifies the output channel of the current node for the next hop of the route and the state of the virtual channel (idle, waiting for resources, or active). The virtual channel may also include pointers to the flits of the packet that are buffered on the current node and the number of flit buffers available on the next node.
The term “wormhole” plays on the way messages are transmitted over the channels: the output port at the next router can be so short that received data can be translated in the head flit before the full message arrives. This allows the router to quickly set up the route upon arrival of the head flit and then opt out from the rest of the conversation. Since a message is transmitted flit by flit, the message may occupy several flit buffers along its path at different routers, creating a worm-like image.
Based upon the traffic between various end points, and the routes and physical networks that are used for various messages, different physical channels of the NoC interconnect may experience different levels of load and congestion. The capacity of various physical channels of a NoC interconnect is determined by the width of the channel (number of physical wires) and the clock frequency at which it is operating. Various channels of the NoC may operate at different clock frequencies, and various channels may have different widths based on the bandwidth requirement at the channel. The bandwidth requirement at a channel is determined by the flows that traverse over the channel and their bandwidth values. Flows traversing over various NoC channels are affected by the routes taken by various flows. In a mesh or Torus NoC, there may exist multiple route paths of equal length or number of hops between any pair of source and destination nodes. For example, in
In a NoC with statically allocated routes for various traffic slows, the load at various channels may be controlled by intelligently selecting the routes for various flows. When a large number of traffic flows and substantial path diversity is present, routes can be chosen such that the load on all NoC channels is balanced nearly uniformly, thus avoiding a single point of bottleneck. Once routed, the NoC channel widths can be determined based on the bandwidth demands of flows on the channels. Unfortunately, channel widths cannot be arbitrarily large due to physical hardware design restrictions, such as timing or wiring congestion. There may be a limit on the maximum channel width, thereby putting a limit on the maximum bandwidth of any single NoC channel.
Additionally, wider physical channels may not help in achieving higher bandwidth if messages are short. For example, if a packet is a single flit packet with a 64-bit width, then no matter how wide a channel is, the channel will only be able to carry 64 bits per cycle of data if all packets over the channel are similar. Thus, a channel width is also limited by the message size in the NoC. Due to these limitations on the maximum NoC channel width, a channel may not have enough bandwidth in spite of balancing the routes.
To address the above bandwidth concern, multiple parallel physical NoCs may be used. Each NoC may be called a layer, thus creating a multi-layer NoC architecture. Hosts inject a message on a NoC layer; the message is then routed to the destination on the NoC layer, where it is delivered from the NoC layer to the host. Thus, each layer operates more or less independently from each other, and interactions between layers may only occur during the injection and ejection times.
In
In a multi-layer NoC, the number of layers needed may depend upon a number of factors such as the aggregate bandwidth requirement of all traffic flows in the system, the routes that are used by various flows, message size distribution, maximum channel width, etc. Once the number of NoC layers in NoC interconnect is determined in a design, different messages and traffic flows may be routed over different NoC layers. Additionally, one may design NoC interconnects such that different layers have different topologies in number of routers, channels and connectivity. The channels in different layers may have different widths based on the flows that traverse over the channel and their bandwidth requirements.
If a NoC agent 400 processes and accepts requests indiscriminately, problems may arise in optimizing bandwidth allocation. In particular, if a NoC agent 400 is an endpoint (e.g., a resource such as memory or I/O), allocating endpoint bandwidth among a number of NoC agents 400 may not be trivial to resolve.
The present disclosure is directed to Quality of Service (QoS) and handshake protocols to facilitate endpoint bandwidth allocation among one or more agents in a Network on Chip (NoC) for an endpoint agent. The QoS policy and handshake protocols may involve use of credits for buffer allocation, wherein the credits are sent to agents in the NoC to compel acceptance of data and allocation of an appropriate buffer. Messages sent to NoC agents may also have a priority associated with the message, wherein higher priority messages have automatic bandwidth allocation and lower priority messages are processed using a handshake protocol.
Aspects of the present application may include a method, which involves processing a request for accepting data from a requesting agent associated with a Network on Chip (NoC), wherein processing can include receiving data, at receiving NoC agent, associated with the request when the request is associated with a credit for buffer allocation and utilizing a handshake protocol to process the request when the request is not associated with the credit for buffer allocation. In an implementation, handshake protocol can include determination of whether a buffer is available for receiving data associated with the request, wherein in case a determination, indicative of the buffer being available for receiving data, is made, a credit can be issued for buffer allocation to the requesting NoC agent. On the other hand, in case a determination, indicative of the buffer not being available for accepting data, is made, the requesting NoC agent can be notified with an indication of the buffer not being available.
Aspect of present application may include a computer readable storage medium storing instructions for executing a process. The instructions may involve processing a request for accepting data from a requesting agent associated with a Network on Chip (NoC), wherein processing can include receiving data, at receiving NoC agent, associated with the request when the request is associated with a credit for buffer allocation and utilizing a handshake protocol to process the request when the request is not associated with the credit for buffer allocation. In an implementation, handshake protocol can include determination of whether a buffer is available for receiving data associated with the request, wherein in case a determination, indicative of the buffer being available for receiving data, is made, a credit can be issued for buffer allocation to the requesting NoC agent. On the other hand, in case a determination, indicative of the buffer not being available for accepting data, is made, the requesting NoC agent can be notified with an indication of the buffer not being available.
Aspects of present application may include a method, which involves, for a network on chip (NoC) configuration, including a plurality of cores interconnected by a plurality of NoC agents/routers in a heterogenous or heterogenous mesh, ring, or torus arrangement, allocating one or more buffers to one or more agents associated with a Network on Chip (NoC) based on a Quality of Service (QoS), and sending one or more credits for buffer allocation to the one or more agents based on the allocation. In an implementation, one or more buffers can be reserved for one or more NoC agents that require a reserve buffer based on the QoS policy.
Aspects of the present application may include a system, which involves, a processor that can be configured to execute one or more modules including a memory controller module, wherein the memory controller module can be configured to generate instructions for transmitting data/packets into/from memory controllers of one or more agents of NoC. Module can also be configured to facilitate Quality-of-Service (QoS) through various protocols such as a QoS policy, handshaking protocols, and other protocols depending on desired implementations, wherein the module can either configured within each memory controller of respective NoC agent, or in a group/sub-group of controllers, or can be externally implemented to control and be operatively/communicatively coupled with respective memory controllers. Instructions of the module can be configured to facilitate interaction between the data/packet requesting NoC agents and receiving NoC agents. Such instructions, for instance, can be implemented on a non-transitory computer readable medium and configured to process a request for accepting data, wherein the module may determine whether to act on the request or deny the request. Module may also be configured to implement and allocate bandwidth to associated NoC agents based on a QoS policy by issuing credits for buffer allocation, thereby behaving as a QoS policy allocator (QPA).
The following detailed description provides further details of the figures and example implementations of the present application. Reference numerals and descriptions of redundant elements between figures are omitted for clarity. Terms used throughout the description are provided as examples and are not intended to be limiting. For example, the use of the term “automatic” may involve fully automatic or semi-automatic implementations involving user or administrator control over certain aspects of the implementation, depending on the desired implementation of one of ordinary skill in the art practicing implementations of the present application.
Example implementations described herein are directed to end-to-end flow control for data transmitted within a NoC. By implementation of a flow control, agents that submit a request to send data are configured to send data only when allowed by the receiving NoC agent. In another example implementation, requesting agent obtains permission from receiving agent before the requesting agent can send data. In another example implementation, permission can be granted by the requesting agent when the requesting agent has space in memory. In another implementation, all data sent by the requesting agent must be accepted by the receiving agent, and therefore the receiving agent must have pre-allocated buffers to accept data.
In an embodiment, various policies can be configured in memory controller 502 based on desired implementations. For example, QoS can be enforced based on a handshake protocol between NoC agent 500-1 and other NoC agents 500. Each NoC agent in the NoC can also be configured with its own memory controller 502 so that the NoC can keep its buffers empty when possible, thereby reducing traffic congestion. Memory controller 502 can achieve this by using end-to-end flow control as described in implementations below to reduce traffic congestion. QoS policy allocator (QPA) and handshake protocols implemented by/in a controller 502 can facilitate the end-to-end flow control.
In another embodiment, memory controller 502 can be a dedicated hardware for handling requests of other agents of the NoC or may also be in the form of a computer readable medium storing instructions for facilitating the requests. Computer readable medium may take the form of a non-transitory computer readable storage medium or a computer readable signal medium as described below. Memory controller 502 may also be implemented as a processor for its respective NoC agent.
According to one embodiment, the present disclosure is directed to Quality of Service (QoS) and handshake protocols to facilitate endpoint bandwidth allocation among one or more agents in a Network on Chip (NoC) for an endpoint agent. The QoS policy and handshake protocols may involve use of credits for buffer allocation, wherein the credits are sent to agents in the NoC to compel acceptance of data and allocation of an appropriate buffer. Messages sent to NoC agents may also have a priority associated with the message, wherein higher priority messages have automatic bandwidth allocation and lower priority messages can be processed using a handshake protocol.
Aspects of the present application may include a method, which involves processing a request for accepting data from a requesting agent associated with a Network on Chip (NoC), wherein processing can include receiving data, at receiving NoC agent, associated with the request when the request is associated with a credit for buffer allocation and utilizing a handshake protocol to process the request when the request is not associated with the credit for buffer allocation. In an implementation, handshake protocol can include determination of whether a buffer is available for receiving data associated with the request, wherein in case a determination, indicative of the buffer being available for receiving data, is made, a credit can be issued for buffer allocation to the requesting NoC agent. On the other hand, in case a determination, indicative of the buffer not being available for accepting data, is made, the requesting NoC agent can be notified with an indication of the buffer not being available. In an implementation, the step of determining indication relating to availability of buffer can further include the step of instructing the requesting NoC agent to wait for an issuance of the credit for buffer allocation. In another example implementation, such an indication can be associated with a time interval for the requesting NoC agent to resend the request upon an elapse of the time interval, wherein the time interval can be determined by one or a combination of the NoC agent and the indication.
Aspects of present application may also include a method, which involves, for a network on chip (NoC) configuration, including a plurality of cores interconnected by a plurality of NoC agents/routers in a heterogenous or heterogenous mesh, ring, or torus arrangement, allocating one or more buffers to one or more agents associated with a Network on Chip (NoC) based on a Quality of Service (QoS), and sending one or more credits for buffer allocation to the one or more agents based on the allocation. In an implementation, one or more buffers can be reserved for one or more NoC agents that require a reserve buffer based on the QoS policy. In an example implementation, allocation of one or more buffers can be done to one or more NoC agents based on the QoS policy, from a pool of buffers.
In another example implementation, method of the proposed architecture can include processing a request for accepting data from a requesting NoC agent and evaluating priority of the request. The method can further include determining whether the evaluated priority is ‘high’, in which case a credit can be allocated for buffer allocation to the requesting agent. On the other hand, in case it is determined that the priority is ‘not high’, the requesting NoC agent can be notified of the buffer not being available. In such a case a credit can be allocated when a buffer is available or at any other defined criteria. In another embodiment, a run-time check of a ‘not high’ request can be done, and when the status of the same changes, a credit can be allocated. Number of time a ‘not high’ request is being received can also be a parameter for deciding when the credit is to be allocated. One should appreciate that any other possible combination or new criteria for deciding when to allocate a credit is completely within the scope of the instant disclosure.
If a buffer cannot be allocated to accept the data for the request (No), then the flow proceeds to 605, wherein the receiving NoC agent sends a message to the requesting agent that indicates that no buffer is available for the requesting NoC agent. The message may be implemented in various ways, depending on the desired implementation. In one example implementation, the message may be associated with a time interval such that the requesting NoC agent resends the request to the receiving NoC agent once the time interval has elapsed. Time interval can be set based on desired implementation (e.g., based on QoS policy, bandwidth allocated to the agent, etc.). Alternatively, time interval can be determined by the requesting NoC agent based on desired implementation of the requesting NoC agent, and the requesting agent can be configured to automatically resend the request after the time interval has elapsed.
In another example implementation, the message may be associated with instructions to the requesting agent to wait until a credit for buffer allocation is issued by the receiving agent. In this example implementation, the requesting NoC agent does not attempt to resend the request until it receives a credit, wherein the requesting agent resends the request along with the credit to compel the allocation of a buffer for the data.
In an example implementation, receiving NoC agent can also be configured to issue one or more credits for buffer allocation to one or more requesting NoC agents based on QoS policy. In an example implementation, receiving NoC agent can preemptively issue one or more credits to agents that are well known to be potential requesting NoC agents in view of the receiving NoC agent. In this manner, requesting agents can thereby send a request and use one of the preemptively received credit and avoid the handshake protocol. After a requesting agent has used up all of its associated credits, requesting agent can be mandated to obtain additional credits via the handshake protocol. Issuance of credits by the receiving agent can also depend on the QoS policy. For example, a known requesting agent that requires a higher bandwidth may be issued credits, whereas agents that have smaller bandwidth requirements may not be issued credits, or may be issued fewer credits.
In an example embodiment, receiving NoC agent can also be configured to issue credits to requesting agents dynamically. For example, while the receiving NoC agent is receiving data and allocating a buffer from a requesting agent, the receiving agent can issue additional credits to the requesting agent or to other agents in accordance with the QoS policy. The receiving agent can also dynamically allocate buffers based on the credits outstanding (e.g., issued but not yet received).
Furthermore, credits for buffer allocation can be preemptively provided based on expectation of receiving a message such as a reply to a request.
During the next three transactions, the receiving agent 704 sends (
In an implementation, number of transactions in
In example implementations of the present disclosure, buffers in memory of a receiving agent 804 can be used to form a common pool, or can be reserved for use only by specific requesting agents 802 based on the QoS policy. In either implementation, buffer can be allocated according to the QoS policy and the handshake protocols. For example, for implementations that reserve buffers for specific requesting agents, the receiving agent 804 can decide to preemptively allocate buffers to specific agents based on the QoS policy (e.g., 10 buffers for agent 1, 20 buffers for agent 2, etc.), wherein the buffers allocated are only utilized by specific agents as determined by the receiving agent.
In another example implementation, a common pool of buffers can be utilized, wherein buffers are allocated based on QoS policy. In such an implementation, receiving agent can be configured to keep track of how many credits were given to each requestor during a predetermined period (e.g., last T cycles) depending on the desired implementation.
To illustrate an example implementation involving a pool of buffers, let the number of buffers allocated to each of the requesting agents associated with the receiving agent be denoted as b1, b2, b3, . . . , bn, where 1, . . . , and n denotes the requesting agent. For each cycle T, the receiving agent can be configured to determine how many buffers has the receiving agent given to each requestor. Let C1, C2, C3, . . . , Cn be a constant indicative of a threshold of buffer allocations for agents 1, . . . , and n. In one example, the receiving agent can automatically issue a credit if bi<Ci, otherwise the handshake protocol can be used to determine allocation of a credit.
Other variations of this implementation are also possible. For example, a function f (b1, b2, . . . , bn) can be implemented such that if bi<[Ci*function (b1, b2, . . . , bn)], a credit can be automatically issued. Function can be predicated on any factor as needed for the desired implementation. For example, the function can optimize credit allocation such that credits are issued if other agents are idle. In such an implementation, function can be configured such that if other b1, b2, . . . , bn are zero, the function is a high number (e.g. (summation of Ci/summation of bi). This is because when a value of b is zero, the requesting agent is idle as the requesting agent has not asked for anything for last T cycles. Other functions are also possible based on the desired policy.
In an example embodiment, an implementation involving a common pool of buffers can be used in conjunction with or separately from an implementation involving reserving buffers for specific receiving agents.
In example implementations where buffers can be reserved for specific requesting agents, receiving agent can send pre-credits to the specific requesting agents based on reserved buffers. Similar implementations can also be used with the implementation involving a common pool of buffers, wherein number of pre-credits can be based on some algorithm according to a desired implementation. For example, when a buffer is available in the pool, pre-credit can be automatically sent based on determining requesting agents that need a credit (e.g., are either waiting, were previously denied, etc.)), (Ci−bi) can be computed, requesting agents having highest value of (Ci−bi) can be determined, and a credit can be issued to requesting agent when a buffer is available. Other implementations are also possible, such as sending a credit to the requesting agent that has the highest value of [(Ci−bi)/bi], or by other weighted allocation schemes depending on the desired implementation.
In other example implementations, certain requests can be prioritized over other requests. In such an implementation, the receiving agent can be configured to process a request for accepting data from a requesting agent associated with the NoC and determine a priority of the request. When the priority of the request is determined to be high, the receiving agent can automatically allocate a credit for buffer allocation to the agent. In this manner, high priority requests can be processed more quickly by ensuring that a credit is sent to the requesting agent without requiring the agent to wait or resend the request at a later time interval. For requests that are not high priority, the receiving agent can utilize a handshake protocol as described above to either notify the agent of buffer not being available or allocating a credit when buffer is available. The notification can take the form of an indication such as a message, or can be implemented in other ways depending on the desired implementation.
Further, the priority scheme can be implemented based on desired implementation of the NoC. For example, if requests from the Central Processing Unit (CPU) agent of the NoC are considered to be high priority, then requests from the CPU can be indicated as high priority by either a flag or by other implementations. In an example implementation, a high priority request can always be serviced before a request that is not high priority and no arbitration is needed for the high priority requests. If multiple high priority requests are received by the requesting agent, arbitration implementations such as the credit system and handshake protocol as described above can be applied to arbitrate between multiple high priority requests.
Computer 905 may also be connected to an external storage 950, which can contain removable storage such as a portable hard drive, optical media (CD or DVD), disk media or any other medium from which a computer can read executable code. The computer may also be connected an output device 955, such as a display to output data and other information to a user, as well as request additional information from a user. Connections from computer 905 to user interface 940, operator interface 945, external storage 950, and output device 955 may be through wireless protocols, such as the 802.11 standards, Bluetooth® or cellular protocols, or via physical transmission media, such as cables or fiber optics. Output device 955 may therefore further act as an input device for interacting with a user.
Processor 910 may execute one or more modules including a memory controller module 911 that is configured to generate instructions for transmitting data/packets into/from memory controllers of one or more agents of NoC. Module 911 can also be configured to facilitate Quality-of-Service (QoS) through various protocols such as a QoS policy, handshaking protocols, and other protocols depending on desired implementations, wherein the module 911 can either configured within each memory controller of respective NoC agent, or in a group/sub-group of controllers, or can be externally implemented to control and be operatively/communicatively coupled with respective memory controllers. Instructions of the module 911 can be configured to facilitate interaction between the data/packet requesting NoC agents and receiving NoC agents. Such instructions, for instance, can be implemented on a non-transitory computer readable medium and configured to process a request for accepting data, wherein the module 911 may determine whether to act on the request or deny the request. Module 911 may also be configured to implement and allocate bandwidth to associated NoC agents based on a QoS policy by issuing credits for buffer allocation, thereby behaving as a QoS policy allocator (QPA).
In some example implementations, the computer system 900 can be implemented in a computing environment such as a cloud. Such a computing environment can include the computer system 900 being implemented as or communicatively connected to one or more other devices by a network and also connected to one or more storage devices. Such devices can include movable user equipment (UE) (e.g., smartphones, devices in vehicles and other machines, devices carried by humans and animals, and the like), mobile devices (e.g., tablets, notebooks, laptops, personal computers, portable televisions, radios, and the like), and devices designed for stationary use (e.g., desktop computers, other computers, information kiosks, televisions with one or more processors embedded therein and/or coupled thereto, radios, and the like).
Furthermore, some portions of the detailed description are presented in terms of algorithms and symbolic representations of operations within a computer. These algorithmic descriptions and symbolic representations are the means used by those skilled in the data processing arts to most effectively convey the essence of their innovations to others skilled in the art. An algorithm is a series of defined steps leading to a desired end state or result. In the example implementations, the steps carried out require physical manipulations of tangible quantities for achieving a tangible result.
Moreover, other implementations of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the example implementations disclosed herein. Various aspects and/or components of the described example implementations may be used singly or in any combination. It is intended that the specification and examples be considered as examples, with a true scope and spirit of the application being indicated by the following claims.
The present application is a continuation of U.S. patent application Ser. No. 15/264,381, filed on Sep. 13, 2016, which is a continuation of U.S. Pat. No. 9,473,415, issued on Oct. 18, 2016 (U.S. patent application Ser. No. 14/185,811 filed on Feb. 20, 2014), the disclosures of which are incorporated by reference in its entirety for all purposes.
Number | Name | Date | Kind |
---|---|---|---|
4409838 | Schomberg | Oct 1983 | A |
4933933 | Dally et al. | Jun 1990 | A |
5105424 | Flaig et al. | Apr 1992 | A |
5163016 | Har'El et al. | Nov 1992 | A |
5355455 | Hilgendorf et al. | Oct 1994 | A |
5432785 | Ahmed et al. | Jul 1995 | A |
5563003 | Suzuki et al. | Oct 1996 | A |
5583990 | Birrittella et al. | Dec 1996 | A |
5588152 | Dapp et al. | Dec 1996 | A |
5764741 | Holender | Jun 1998 | A |
5859981 | Levin et al. | Jan 1999 | A |
5991308 | Fuhrmann et al. | Nov 1999 | A |
6003029 | Agrawal et al. | Dec 1999 | A |
6029220 | Iwamura et al. | Feb 2000 | A |
6058385 | Koza et al. | May 2000 | A |
6101181 | Passint et al. | Aug 2000 | A |
6108739 | James et al. | Aug 2000 | A |
6249902 | Igusa et al. | Jun 2001 | B1 |
6314487 | Hahn et al. | Nov 2001 | B1 |
6347337 | Shah | Feb 2002 | B1 |
6415282 | Mukherjea et al. | Jul 2002 | B1 |
6674720 | Passint et al. | Jan 2004 | B1 |
6711717 | Nystrom et al. | Mar 2004 | B2 |
6925627 | Longway et al. | Aug 2005 | B1 |
6967926 | Williams, Jr. et al. | Nov 2005 | B1 |
6983461 | Hutchison et al. | Jan 2006 | B2 |
7046633 | Carvey | May 2006 | B2 |
7065730 | Alpert et al. | Jun 2006 | B2 |
7143221 | Bruce et al. | Nov 2006 | B2 |
7318214 | Prasad et al. | Jan 2008 | B1 |
7379424 | Krueger | May 2008 | B1 |
7437518 | Tsien | Oct 2008 | B2 |
7461236 | Wentzlaff | Dec 2008 | B1 |
7509619 | Miller et al. | Mar 2009 | B1 |
7564865 | Radulescu | Jul 2009 | B2 |
7590959 | Tanaka | Sep 2009 | B2 |
7639037 | Eberle | Dec 2009 | B1 |
7693064 | Thubert et al. | Apr 2010 | B2 |
7701252 | Chow et al. | Apr 2010 | B1 |
7724735 | Locatelli et al. | May 2010 | B2 |
7725859 | Lenahan et al. | May 2010 | B1 |
7774783 | Toader | Aug 2010 | B2 |
7808968 | Kalmanek, Jr. et al. | Oct 2010 | B1 |
7853774 | Wentzlaff | Dec 2010 | B1 |
7917885 | Becker | Mar 2011 | B2 |
7957381 | Clermidy et al. | Jun 2011 | B2 |
7973804 | Mejdrich et al. | Jul 2011 | B2 |
8018249 | Koch et al. | Sep 2011 | B2 |
8020163 | Nollet et al. | Sep 2011 | B2 |
8020168 | Hoover et al. | Sep 2011 | B2 |
8050256 | Bao et al. | Nov 2011 | B1 |
8059551 | Milliken | Nov 2011 | B2 |
8099757 | Riedle et al. | Jan 2012 | B2 |
8136071 | Solomon | Mar 2012 | B2 |
8203938 | Gibbings | Jun 2012 | B2 |
8261025 | Mejdrich et al. | Sep 2012 | B2 |
8281297 | Dasu et al. | Oct 2012 | B2 |
8306042 | Abts | Nov 2012 | B1 |
8312402 | Okhmatovski et al. | Nov 2012 | B1 |
8352774 | Elrabaa | Jan 2013 | B2 |
8407425 | Gueron et al. | Mar 2013 | B2 |
8412795 | Mangano et al. | Apr 2013 | B2 |
8438578 | Hoover et al. | May 2013 | B2 |
8448102 | Komachuk et al. | May 2013 | B2 |
8490110 | Hoover et al. | Jul 2013 | B2 |
8492886 | Or-Bach et al. | Jul 2013 | B2 |
8514889 | Jayasimha | Aug 2013 | B2 |
8541819 | Or-Bach et al. | Sep 2013 | B1 |
8543964 | Ge et al. | Sep 2013 | B2 |
8601423 | Philip et al. | Dec 2013 | B1 |
8619622 | Harrand et al. | Dec 2013 | B2 |
8635577 | Kazda et al. | Jan 2014 | B2 |
8661455 | Mejdrich et al. | Feb 2014 | B2 |
8667439 | Kumar et al. | Mar 2014 | B1 |
8671220 | Garg | Mar 2014 | B1 |
8705368 | Abts et al. | Apr 2014 | B1 |
8711867 | Guo et al. | Apr 2014 | B2 |
8717875 | Bejerano et al. | May 2014 | B2 |
8726295 | Hoover et al. | May 2014 | B2 |
8738860 | Griffin et al. | May 2014 | B1 |
8793644 | Michel et al. | Jul 2014 | B2 |
8798038 | Jayasimha et al. | Aug 2014 | B2 |
8819611 | Philip et al. | Aug 2014 | B2 |
20020071392 | Grover et al. | Jun 2002 | A1 |
20020073380 | Cooke et al. | Jun 2002 | A1 |
20020083159 | Ward et al. | Jun 2002 | A1 |
20020095430 | Egilsson et al. | Jul 2002 | A1 |
20030088602 | Dutta et al. | May 2003 | A1 |
20030145314 | Nguyen et al. | Jul 2003 | A1 |
20040049565 | Keller et al. | Mar 2004 | A1 |
20040103218 | Blumrich et al. | May 2004 | A1 |
20040216072 | Alpert et al. | Oct 2004 | A1 |
20050147081 | Acharya et al. | Jul 2005 | A1 |
20050203988 | Nollet et al. | Sep 2005 | A1 |
20060031615 | Bruce et al. | Feb 2006 | A1 |
20060075169 | Harris et al. | Apr 2006 | A1 |
20060161875 | Rhee | Jul 2006 | A1 |
20060206297 | Ishiyama et al. | Sep 2006 | A1 |
20060209846 | Clermidy et al. | Sep 2006 | A1 |
20060268909 | Langevin et al. | Nov 2006 | A1 |
20070038987 | Ohara et al. | Feb 2007 | A1 |
20070088537 | Lertora et al. | Apr 2007 | A1 |
20070118320 | Luo et al. | May 2007 | A1 |
20070147379 | Lee et al. | Jun 2007 | A1 |
20070162903 | Babb, II et al. | Jul 2007 | A1 |
20070244676 | Shang et al. | Oct 2007 | A1 |
20070256044 | Coryer et al. | Nov 2007 | A1 |
20070267680 | Uchino et al. | Nov 2007 | A1 |
20070274331 | Locatelli et al. | Nov 2007 | A1 |
20080072182 | He et al. | Mar 2008 | A1 |
20080120129 | Seubert et al. | May 2008 | A1 |
20080126569 | Rhim et al. | May 2008 | A1 |
20080184259 | Lesartre et al. | Jul 2008 | A1 |
20080186998 | Rijpkema | Aug 2008 | A1 |
20080211538 | Lajolo et al. | Sep 2008 | A1 |
20080232387 | Rijpkema et al. | Sep 2008 | A1 |
20090037888 | Tatsuoka et al. | Feb 2009 | A1 |
20090043934 | Bjerregaard | Feb 2009 | A1 |
20090046727 | Towles | Feb 2009 | A1 |
20090070726 | Mehrotra et al. | Mar 2009 | A1 |
20090100148 | Murali | Apr 2009 | A1 |
20090122703 | Gangwal et al. | May 2009 | A1 |
20090172304 | Gueron et al. | Jul 2009 | A1 |
20090187716 | Comparan et al. | Jul 2009 | A1 |
20090187756 | Nollet et al. | Jul 2009 | A1 |
20090210184 | Medardoni et al. | Aug 2009 | A1 |
20090231348 | Mejdrich et al. | Sep 2009 | A1 |
20090268677 | Chou et al. | Oct 2009 | A1 |
20090285222 | Hoover et al. | Nov 2009 | A1 |
20090300292 | Fang et al. | Dec 2009 | A1 |
20090307714 | Hoover et al. | Dec 2009 | A1 |
20090313592 | Murali et al. | Dec 2009 | A1 |
20100040162 | Suehiro | Feb 2010 | A1 |
20100158005 | Mukhopadhyay et al. | Jun 2010 | A1 |
20100211718 | Gratz et al. | Aug 2010 | A1 |
20100223505 | Andreev et al. | Sep 2010 | A1 |
20110022754 | Cidon et al. | Jan 2011 | A1 |
20110035523 | Feero et al. | Feb 2011 | A1 |
20110060831 | Ishii et al. | Mar 2011 | A1 |
20110072407 | Keinert et al. | Mar 2011 | A1 |
20110085550 | Lecler et al. | Apr 2011 | A1 |
20110103799 | Shacham et al. | May 2011 | A1 |
20110154282 | Chang et al. | Jun 2011 | A1 |
20110191774 | Hsu et al. | Aug 2011 | A1 |
20110235531 | Vangal | Sep 2011 | A1 |
20110276937 | Waller | Nov 2011 | A1 |
20110302345 | Boucard et al. | Dec 2011 | A1 |
20110307734 | Boesen et al. | Dec 2011 | A1 |
20110320854 | Elrabaa | Dec 2011 | A1 |
20120022841 | Appleyard | Jan 2012 | A1 |
20120023473 | Brown et al. | Jan 2012 | A1 |
20120026917 | Guo et al. | Feb 2012 | A1 |
20120079147 | Ishii et al. | Mar 2012 | A1 |
20120099475 | Tokuoka | Apr 2012 | A1 |
20120110106 | De Lescure et al. | May 2012 | A1 |
20120110541 | Ge et al. | May 2012 | A1 |
20120155250 | Carney et al. | Jun 2012 | A1 |
20120173846 | Wang et al. | Jul 2012 | A1 |
20120209944 | Mejdrich et al. | Aug 2012 | A1 |
20130028090 | Yamaguchi et al. | Jan 2013 | A1 |
20130028261 | Lee | Jan 2013 | A1 |
20130051397 | Guo | Feb 2013 | A1 |
20130054811 | Harrand | Feb 2013 | A1 |
20130080073 | de Corral | Mar 2013 | A1 |
20130103369 | Huynh et al. | Apr 2013 | A1 |
20130103912 | Jones et al. | Apr 2013 | A1 |
20130117543 | Venkataramanan et al. | May 2013 | A1 |
20130148506 | Lea | Jun 2013 | A1 |
20130151215 | Mustapha | Jun 2013 | A1 |
20130159669 | Comparan | Jun 2013 | A1 |
20130159944 | Uno et al. | Jun 2013 | A1 |
20130163615 | Mangano et al. | Jun 2013 | A1 |
20130174113 | Lecler et al. | Jul 2013 | A1 |
20130179613 | Boucard et al. | Jul 2013 | A1 |
20130179902 | Hoover et al. | Jul 2013 | A1 |
20130191572 | Nooney et al. | Jul 2013 | A1 |
20130207801 | Barnes | Aug 2013 | A1 |
20130219148 | Chen et al. | Aug 2013 | A1 |
20130250792 | Yoshida et al. | Sep 2013 | A1 |
20130254488 | Kaxiras et al. | Sep 2013 | A1 |
20130263068 | Cho et al. | Oct 2013 | A1 |
20130268990 | Urzi et al. | Oct 2013 | A1 |
20130326458 | Kazda et al. | Dec 2013 | A1 |
20140052938 | Kim | Feb 2014 | A1 |
20140068132 | Philip et al. | Mar 2014 | A1 |
20140068134 | Philip et al. | Mar 2014 | A1 |
20140092740 | Wang et al. | Apr 2014 | A1 |
20140098683 | Kumar | Apr 2014 | A1 |
20140112149 | Urzi et al. | Apr 2014 | A1 |
20140115218 | Philip et al. | Apr 2014 | A1 |
20140115298 | Philip et al. | Apr 2014 | A1 |
20140211622 | Kumar et al. | Jul 2014 | A1 |
20140254388 | Kumar et al. | Sep 2014 | A1 |
20140301205 | Harrand | Oct 2014 | A1 |
20140328172 | Kumar | Nov 2014 | A1 |
20150043575 | Kumar et al. | Feb 2015 | A1 |
20150109024 | Abdelfattah et al. | Apr 2015 | A1 |
20150159330 | Weisman et al. | Jun 2015 | A1 |
Number | Date | Country |
---|---|---|
103684961 | Mar 2014 | CN |
10-2013-0033898 | Apr 2013 | KR |
2010074872 | Jul 2010 | WO |
2013063484 | May 2013 | WO |
2014059024 | Apr 2014 | WO |
Entry |
---|
QNoC: QoS architecture and design process for network on chip: Bolotin et al. 2004. |
Ababei, C., et al., Achieving Network on Chip Fault Tolerance by Adaptive Remapping, Parallel & Distributed Processing, 2009, IEEE International Symposium, 4 pgs. |
Abts, D., et al., Age-Based Packet Arbitration in Large-Radix k-ary n-cubes, Supercomputing 2007 (SC07), Nov. 10-16, 2007, 11 pgs. |
Beretta, I, et al., A Mapping Flow for Dynamically Reconfigurable Multi-Core System-on-Chip Design, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, Aug. 2011, 30(8), pp. 1211-1224. |
Das, R., et al., Aergia: Exploiting Packet Latency Slack in On-Chip Networks, 37th International Symposium on Computer Architecture (ISCA '10), Jun. 19-23, 2010, 11 pgs. |
Ebrahimi, E., et al., Fairness via Source Throttling: A Configurable and High-Performance Fairness Substrate for Multi-Core Memory Systems, ASPLOS '10, Mar. 13-17, 2010, 12 pgs. |
Gindin, R., et al., NoC-Based FPGA: Architecture and Routing, Proceedings of the First International Symposium on Networks-on-Chip (NOCS'07), May 2007, pp. 253-262. |
Grot, B., Preemptive Virtual Clock: A Flexible, Efficient, and Cost-Effective QOS Scheme for Networks-on-Chip, Micro '09, Dec. 12-16, 2009, 12 pgs. |
Grot, B., Kilo-NOC: A Heterogeneous Network-on-Chip Architecture for Scalability and Service Guarantees, ISCA 11, Jun. 4-8, 2011, 12 pgs. |
Grot, B., Topology-Aware Quality-of-Service Support in Highly Integrated Chip Multiprocessors, 6th Annual Workshop on the Interaction between Operating Systems and Computer Architecture, Jun. 2006, 11 pgs. |
Hestness, J., et al., Netrace: Dependency-Tracking for Efficient Network-on-Chip Experimentation, The University of Texas at Austin, Dept. of Computer Science, May 2011, 20 pgs. |
Jiang, N., et al., Performance Implications of Age-Based Allocations in On-Chip Networks, CVA MEMO 129, May 24, 2011, 21 pgs. |
Lee, J. W., et al., Globally-Synchronized Frames for Guaranteed Quality-of-Service in On-Chip Networks, 35th IEEE/ACM International Symposium on Computer Architecture (ISCA), Jun. 2008, 12 pgs. |
Lee, M. M., et al., Approximating Age-Based Arbitration in On-Chip Networks, PACT '10, Sep. 11-15, 2010, 2 pgs. |
Li, B., et al. CoQoS: Coordinating QoS-Aware Shared Resources in NoC-based SoCs, J. Parallel Distrib. Comput., 71(5), May 2011, 14 pgs. |
Lin, S., et al., Scalable Connection-Based Flow Control Scheme for Application-Specific Network-on-Chip, The Journal of China Universities of Posts and Telecommunications, Dec. 2011, 18(6), pp. 98-105. |
Munirul, H.M., et al., Evaluation of Multiple-Valued Packet Multiplexing Scheme for Network-on-Chip Architecture, Proceedings of the 36th International Symposium on Multiple-Valued Logic (ISMVL '06), 2006, 6 pgs. |
Yang, J., et al., Homogeneous NoC-based FPGA: The Foundation for Virtual FPGA, 10th IEEE International Conference on Computer and Information Technology (CIT 2010), Jun. 2010, pp. 62-67. |
Zaman, Aanam, “Formal Verification of Circuit-Switched Network on Chip (NoC) Architectures using SPIN”, Oosman Hasan, IEEE © 2014, 8 pages. |
International Search Report and Written Opinion for PCT/US2014/037902, dated Sep. 30, 2014, 14 pgs. |
International Search Report and Written Opinion for PCT/US2014/060745, dated Jan. 21, 2015, 10 pgs. |
International Search Report and Written Opinion for PCT/US2014/060879, dated Jan. 21, 2015, 10 pgs. |
International Search Report and Written Opinion for PCT/US2013/064140, dated Jan. 22, 2014, 9 pgs. |
International Search Report and Written Opinion for PCT/US2014/048190, dated Nov. 28, 2014, 11 pgs. |
International Search Report and Written Opinion for PCT/US2014/023625, dated Jul. 10, 2014, 9 pgs. |
International Search Report and Written Opinion for PCT/US2014/012012, dated May 14, 2014, 9 pgs. |
International Search Report and Written Opinion for PCT/US2014/060892, dated Jan. 27, 2015, 10 pgs. |
International Search Report and Written Opinion for PCT/US2014/060886, dated Jan. 26, 2015, 10 pgs. |
International Search Report and Written Opinion for PCT/US2014/012003, dated Mar. 26, 2014, 9 pgs. |
Office Action for Korean Patent Application No. 10-2016-7019093 dated Sep. 8, 2016, 3 pages plus 1 page English translation. KIPO, Korea. |
Notice of Allowance for for Korean Patent Application No. 10-2016-7019093 dated Dec. 5, 2016, 5 pages. KIPO, Korea. |
Bolotin, Evgency, et al., “QNoC: QoS Architecture and Design Process for Network on Chip” 2004, 24 pages, Journal of Systems Architecture 50 (2004) 105-128 Elsevier. |
Number | Date | Country | |
---|---|---|---|
20180019949 A1 | Jan 2018 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 15264381 | Sep 2016 | US |
Child | 15671052 | US | |
Parent | 14185811 | Feb 2014 | US |
Child | 15264381 | US |