Enhanced page locality in network-on-chip (NoC) architectures

Abstract

Aspects of the present disclosure relate to page locality based memory access request processing in a network-on-chip (NoC) architecture. In an example implementation, the proposed method includes determining, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and re-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

Description

TECHNICAL FIELD

Methods and example implementations described herein are generally directed to memory access in Network-on-Chip (NoC) architectures, and more specifically, to reduction in access latency due to enhanced page locality in NoC architectures.

RELATED ART

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 FIG. 1A, 2-D (two dimensional) mesh (as shown in FIG. 1B), and 2-D Torus (as shown in FIG. 1C) are examples of topologies in the related art. Mesh and Torus can also be extended to 2.5-D (two and half dimensional) or 3-D (three dimensional) organizations. FIG. 1D shows a 3D mesh NoC, where there are three layers of 3×3 2D mesh NoC shown over each other. The NoC routers have up to two additional ports, one connecting to a router in the higher layer, and another connecting to a router in the lower layer. Router 111 in the middle layer of the example has its ports used, one connecting to the router 112 at the top layer and another connecting to the router 110 at the bottom layer. Routers 110 and 112 are at the bottom and top mesh layers respectively and therefore have only the upper facing port 113 and the lower facing port 114 respectively connected.

Packets are message transport units for intercommunication between various components. Routing involves identifying a path that is 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 identification (ID). Packets can 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.

FIG. 2A pictorially illustrates an example of XY routing in a two dimensional mesh. More specifically, FIG. 2A illustrates XY routing from node ‘34’ to node ‘00’. In the example of FIG. 2A, each component is connected to only one port of one router. A packet is first routed over the X-axis till the packet reaches node ‘04’ where the X-coordinate of the node is the same as the X-coordinate of the destination node. The packet is next routed over the Y-axis until the packet reaches the destination node.

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 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 a 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 header flit, containing remaining payload of data. The final flit is a tail flit, which, in addition to containing 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 exist multiple route paths of equal length or number of hops between any pair of source and destination nodes. For example, in FIG. 2B, in addition to the standard XY route between nodes 34 and 00, there are additional routes available, such as YX route 203 or a multi-turn route 202 that makes more than one turn from source to destination.

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. FIG. 3A illustrates a two layer NoC. Here the two NoC layers are shown adjacent to each other on the left and right, with the hosts connected to the NoC replicated in both left and right diagrams. A host is connected to two routers in this example—a router in the first layer shown as R1, and a router is the second layer shown as R2. In this example, the multi-layer NoC is different from the 3D NoC, i.e. multiple layers are on a single silicon die and are used to meet the high bandwidth demands of the communication between hosts on the same silicon die. Messages do not go from one layer to another. For purposes of clarity, the present application will utilize such a horizontal left and right illustration for multi-layer NoC to differentiate from the 3D NoCs, which are illustrated by drawing the NoCs vertically over each other.

In FIG. 3B, a host connected to a router from each layer, R1 and R2 respectively, is illustrated. Each router is connected to other routers in its layer using directional ports 301, and is connected to the host using injection and ejection ports 302. A bridge-logic 303 may sit between the host and the two NoC layers to determine the NoC layer for an outgoing message and sends the message from host to the NoC layer, and also perform the arbitration and multiplexing between incoming messages from the two NoC layers and delivers them to the host.

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.

System on Chips (SoCs) are becoming increasingly sophisticated, feature rich, and high performance by integrating a growing number of standard processor cores, memory and I/O subsystems, and specialized acceleration IPs. To address this complexity, NoC approach of connecting SoC components is gaining popularity. A NoC can provide connectivity to a plethora of components and interfaces and simultaneously enable rapid design closure by being automatically generated from a high level specification. The specification describes interconnect requirements of SoC in terms of connectivity, bandwidth, and latency. The specification can include constraints such as Bandwidth/Quality of Service (QoS)/latency attributes that are to be met by the NoC, and can be, in various software formats, depending on the design tools, utilized. Once NoC is generated through the use of design tools on the specification to meet specification requirements, physical architecture can be implemented either by manufacturing a chip layout to facilitate NoC or by generation of a register transfer level (RTL) for execution on a chip to emulate the generated NoC, depending on desired implementation. Specifications may be in common power format (CPF), Unified Power Format (UPF), or others according to the desired specification. Specifications can be in the form of traffic specifications indicating the traffic, bandwidth requirements, latency requirements, interconnections, etc depending on the desired implementation. Specifications can also be in the form of power specifications to define power domains, voltage domains, clock domains, and so on, depending on the desired implementation.

Specification can include parameters for bandwidth, traffic, jitter, dependency information, and attribute information depending on desired implementation. In addition to this, information such as position of various components, protocol information, clocking and power domains, etc. may be supplied. A NoC compiler can then use this specification to automatically design a NoC for the SoC. A number of NoC compilers were introduced in the related art that automatically synthesize a NoC to fit a traffic specification. In such design flows, synthesized NoC is simulated to evaluate performance under various operating conditions and to determine whether the specifications are met. This may be necessary because NoC-style interconnects are distributed systems and their dynamic performance characteristics under load are difficult to predict statically and can be very sensitive to a wide variety of parameters.

One of the key resources in any SOC is the memory bandwidth, wherein available bandwidth is traffic dependent. For a DRAM/memory controller to achieve high bandwidth, it must avoid impediments that arise from switching pages or switching between reads and writes. These impediments/stalls create absolute bandwidth loss in the system, and most memory controllers in real system fall far short of their theoretical bandwidth (something like 65-75% efficiency). To avoid these stalls and achieve higher bandwidth, memory controllers must be able to take advantage of page locality.

Page locality is a kind of spatial locality, wherein when a DRAM page is opened, further accesses to that page are low latency and require no overhead. Since the page is defined by an address range, requests within that address range will have page locality. Therefore, request streams with good spatial locality have the potential to have good page locality. Page locality has a temporal component. For a page hit to occur, requests within the same spatial block of physical memory (often 4 KB or 8 KB) must arrive at the memory controller and be processed before the page is closed.

There is therefore a need for methods, systems, and computer readable mediums for overcoming the above-mentioned issues with existing implementations of memory access so as to increase efficiency of memory controllers and reduce overall memory access latency.

SUMMARY

Methods and example implementations described herein are generally directed to quality-of-service (QOS) enhancement pertaining to packet routing in Network-on-Chip (NoC) architectures, and more specifically, to implementation of a bandwidth weighting mechanism based NoC configuration/constructions for packet routing.

Aspects of the present disclosure relate to methods, systems, and computer readable mediums for page locality based memory access request processing in a network-on-chip (NoC) architecture. In an aspect, method of the present disclosure comprises determining, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and re-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

In an aspect, if said packet stream processing is completed, said arbitrator selects the NoC agent having the next highest priority for the forthcoming round, and said current NoC agent is allocated lowest priority.

In another aspect, said arbitrator is a round-robin based packet processing arbitrator.

In an aspect, the arbitrator selects the NoC agent so as to enhance page locality without compromising on unfairness to said plurality of NoC agents beyond a defined threshold.

In an example implementation, the step of determining whether said packet stream is completely processed is performed based on a stream indicator that forms part of said packet stream, such that said current NoC agent is selected in each forthcoming round till end of packet stream indicator is received. In another example implementation, the step of determining whether said packet stream is completely processed is performed based on destination address mentioned in each packet of said packet stream such that said current NoC agent is selected in each forthcoming round till the same destination address is mentioned in each packet of said packet stream. In yet another example implementation, the step of determining whether said packet stream is completely processed is performed based on an intimation from said current NoC agent that it is a part of said packet stream. In another example implementation, the step of determining whether said packet stream is completely processed is done by a master bridge by analyzing address stream that forms part of said packet stream. In yet another example implementation, the step of determining whether said packet stream is completely processed is done based on receipt of new stream information, said new stream information being shared with each output port of router associated with said arbitrator.

In an aspect, the present disclosure relates to a system for page locality based memory access request processing in a network-on-chip (NoC) architecture, said system comprising: an current agent packet stream processing status determination module configured to determine, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and a packet stream processing status based agent selection module configured to re-select, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

In another aspect, the present disclosure further relates to a non-transitory computer readable storage medium storing instructions for executing a process, the instructions comprising: determining, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and re-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

BRIEF DESCRIPTION OF DRAWINGS

FIGS. 1A, 1B, 1C, and 1D illustrate examples of Bidirectional ring, 2D Mesh, 2D Torus, and 3D Mesh NoC Topologies.

FIG. 2A illustrates an example of XY routing in a related art two dimensional mesh.

FIG. 2B illustrates three different routes between a source and destination nodes.

FIG. 3A illustrates an example of a related art two layer NoC interconnect.

FIG. 3B illustrates the related art bridge logic between host and multiple NoC layers.

FIGS. 4A and 4B illustrate existing implementations of how determination of NoC agent for each round is current done.

FIG. 5 illustrates an exemplary architecture showing manner in which determination of NoC agent for each round is performed in accordance in accordance with an example implementation.

FIG. 6 illustrates a flow diagram showing working of the present invention in accordance with an example implementation.

FIG. 7 illustrates an example computer system on which example embodiments may be implemented.

DETAILED DESCRIPTION

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.

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. In example implementations, a NoC interconnect is generated from a specification by utilizing design tools. The specification can include constraints such as bandwidth/Quality of Service (QoS)/latency attributes that is to be met by the NoC, and can be in various software formats depending on the design tools utilized. Once the NoC is generated through the use of design tools on the specification to meet the specification requirements, the physical architecture can be implemented either by manufacturing a chip layout to facilitate the NoC or by generation of a register transfer level (RTL) for execution on a chip to emulate the generated NoC, depending on the desired implementation. Specifications may be in common power format (CPF), Unified Power Format (UPF), or others according to the desired specification. Specifications can be in the form of traffic specifications indicating the traffic, bandwidth requirements, latency requirements, interconnections, etc. depending on the desired implementation. Specifications can also be in the form of power specifications to define power domains, voltage domains, clock domains, and so on, depending on the desired implementation.

Methods and example implementations described herein are generally directed to quality-of-service (QOS) enhancement pertaining to packet routing in Network-on-Chip (NoC) architectures, and more specifically, to implementation of a bandwidth weighting mechanism based NoC configuration/constructions for packet routing.

Aspects of the present disclosure relate to methods, systems, and computer readable mediums for page locality based memory access request processing in a network-on-chip (NoC) architecture. In an aspect, method of the present disclosure comprises determining, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and re-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

In an aspect, if said packet stream processing is completed, said arbitrator selects the NoC agent having the next highest priority for the forthcoming round, and said current NoC agent is allocated lowest priority.

In another aspect, said arbitrator is a round-robin based packet processing arbitrator.

In an aspect, the arbitrator selects the NoC agent so as to enhance page locality without compromising on unfairness to said plurality of NoC agents beyond a defined threshold.

In an example implementation, the step of determining whether said packet stream is completely processed is performed based on a stream indicator that forms part of said packet stream, such that said current NoC agent is selected in each forthcoming round till end of packet stream indicator is received. In another example implementation, the step of determining whether said packet stream is completely processed is performed based on destination address mentioned in each packet of said packet stream such that said current NoC agent is selected in each forthcoming round till the same destination address is mentioned in each packet of said packet stream. In yet another example implementation, the step of determining whether said packet stream is completely processed is performed based on an intimation from said current NoC agent that it is a part of said packet stream. In another example implementation, the step of determining whether said packet stream is completely processed is done by a master bridge by analyzing address stream that forms part of said packet stream. In yet another example implementation, the step of determining whether said packet stream is completely processed is done based on receipt of new stream information, said new stream information being shared with each output port of router associated with said arbitrator.

In an aspect, the present disclosure relates to a system for page locality based memory access request processing in a network-on-chip (NoC) architecture, said system comprising: an current agent packet stream processing status determination module configured to determine, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and a packet stream processing status based agent selection module configured to re-select, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

In another aspect, the present disclosure further relates to a non-transitory computer readable storage medium storing instructions for executing a process, the instructions comprising: determining, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and re-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

In an aspect, an exemplary problem being solved by the present invention pertains to how to make arrival of requests to memory resource-efficient. Spatial locality occurs naturally in the request streams of most agents, but there is usually no spatial locality across agents, which means that the memory controller must receive multiple requests from a single agent in order to get any page locality. Consider a system with 100 agents talking to memory, where if the arbitration for these 100 agents uses round-robin technique, only 1 request from an agent every 100 requests would be observed, which would provide almost no page locality, and would substantially limit the number of requests from an agent.

The present disclosure therefore provides an arbitration mechanism that favors clustering of agent's traffic so that several requests from the same master arrive at the memory together, allowing for higher memory bandwidth. Not all traffic will have spatial locality, but any hope of improving memory bandwidth rests on utilizing the spatial locality where it does exist.

Routers and bridges that form part of a NoC have a complex arbitration mechanism, wherein they recognize priority levels, QoS weight mechanisms, catch-up and end-of-round state, and store-and-forward levels, but when the competing traffic is the same priority level, the arbitration uses a round-robin arbiter. While simple, this has an almost worst-case effect on the page locality. Therefore, in an aspect, of the present disclosure, in order to enhance page locality, a modification to the round-robin arbiter is needed. Normally when an agent wins the arbitration, it marks itself as the lowest-priority agent in the round-robin protocol. To provide stickiness to an agent (i.e. requests from the same agent be processed in the subsequent round as well in order to improve page locality), an update of the state needs to be changed.

In an aspect, the present disclosure enables an agent/source request to update its state so that is conditionally either the highest priority or the lowest priority based on one or more inputs. It should be set as the highest priority when it is continuing an existing stream, but should be set as lowest priority if it is trying to start a new stream. As would be appreciated, if an agent is sending packets pertaining to a single stream, its likelihood to access a cached memory resource such as a page would be much higher in the next round as well, which would significantly improve the page locality and reduce latency, and therefore in case, during arbitration in the next round, it is determined that the current agent is sending packets that form part of a single stream, the current agent is again selected/chosen as the active agent through arbitration in the next round as well, else said agent is associated the lowest priority.

In an instance of the present invention, if there are 128 master agents talking to memory in a NoC architecture, expectation due to spatial locality is that, for any given master agent, its requests have fairly high likelihood of hitting the same part of the memory, enabling enhanced page locality, low latency, and efficient access to next cache line. Therefore, it may be more efficient for the same master to continue sending requests in the next round as well even though in the next round, due to round-robin implementation, said master would be allocated the lowest priority in the existing techniques. Such allowance of the same master to continue being served till its stream is over, increases stickiness of the master and bring in unfairness to other master agents, but strongly enhances page locality and, as a result, low latency.

In an aspect, the present disclosure enables the arbitration to be programmable and not rigid so as to allow clustering of streaming packets to be processed in a single go and bring in more stickiness while processing requests from an agent that is processing a packet stream. In an aspect therefore, the present invention enables NoC network to cluster requests from the same master to enable efficient page locality. Alternatively, if the network/arbitrator has better information about the request stream itself, requests in the same stream that are directed to the same page can be clustered. Aspects of the present invention can further programmably modify the number of credits being allocated to each agent based on packets being processed by said agent, streams being processed by said agent, attributes of the memory controller, along with priority and characteristics of the other NoC agents.

FIG. 4A illustrates an existing implementation of round-robin based packet routing wherein, as shown, there are two paths 402 and 404 in existing round-robin priority mechanisms, one path 402 being for high priority requests that utilize a mask value 406 to determine which requests are high priority, and other path 404 being for low priority requests. If any high priority request is active, the output Mux 408 selects the first of the them, else it selects the first low priority request.

FIG. 4B illustrates an exemplary high priority mask where in order to support a state where a request is either highest priority or lowest priority, such a high priority mask can be incorporated.

In aspects of the present disclosure, manner in which the highest priority requests/agents are determined for agent selection in the next round can be modified so as to incorporate a factor of whether the current requests form part of the same a single stream such that in case the requests are of the same stream and are likely to access the same cached page from the memory, agent responsible for such current requests is given the highest priority for the next round as well so that same stream continues to get processed, increasing page locality.

As shown in FIG. 4B, requests from 3 to 7 are all high priority, with 3 being the highest priority. In the prior art systems, if all the requests were active, the find-first of the high priority requests would select request 3. On the other hand, the proposed architecture takes into context status of the current agent i.e. the agent that was granted last. In this case, request 2 was the last winner, because after winning an agent modifies the round-robin mask to indicate that they are the lowest priority. Lowest priority is indicated by the equation ˜req[A] & req[A+1]. This information can be used to allow a request port that won last cycle to act as highest priority. In order to determine whether a request should use high or low priority, a new input, as part of the present invention, can indicate whether the request is part of a new stream or an old stream. The following equation can then be used to determine which, if any, input request should act as high priority:

assign continue_hp=req[n−1:0]&˜new_stream[n−1:0]&˜hp_mask[n−1:0]& hp_mask[n:1]; According to the above equation, a request port should act as high priority when it has a valid request, is the lowest priority, and it is continuing a stream. This information can therefore be factored into the round-robin mechanism as shown in FIG. 5.

FIG. 5 illustrates an architecture of the present invention in accordance with an example implementation, wherein, as shown, an evaluation/calculation can be done to generate a new vector that indicates if the lowest priority agent wants to continue with the same stream. There can only be one bit active in this vector as only one of the agents is the lowest priority. If one of the bits is active, the output mux can select that bit, instead of one of the other find-first vectors, which will make the lowest priority agent, if it is continuing the same stream, to win the arbitration which will lead to better page locality.

In an aspect, any method that can enable a network arbitrator node to detect whether the current agent is processing a stream so as to then arbitrate the next round in order to make the same agent win the next arbitration round as well is well within the scope of the present invention. There can be multiple methods of generating appropriate stream information, based on which an arbitrator node can detect if the current agent is processing a stream.

In an exemplary implementation, aspects of the present invention, in order to enhance page locality, can add a small counter (one per input VC) in each arbiter. When an input VC first wins the arbitration, it could be marked as the start of a new stream and a counter can be reset. If new requests from the same input VC arrive, they will be recognized as part of the same stream and win the arbitration through the new arbitration mechanism. Each granted request would increment the counter such that when the counter reaches a predefined count, the next request will act as a new stream. In an exemplary aspect, the counter can only be increments on an End-of-Packet (EOP), as it is counting full requests. The counter can be parameterized or programmable, and can interchangeably be referred to as a stickiness counter.

In another exemplary implementation, the page locality can be significantly improved if page locality was propagated though the network from the source bridges, wherein instead of a counter in the router arbiters, the counter can exist in each TX bridge, such that the bridge would then pass a signal to the target routers to indicate a new stream, or a continuation of the old stream. This allows each bridge to use a different policy. Some agents would want stickiness. Others would want less or none at all. Each could be independently defined or programmed. With this information sent with the packet, the routers would be in better shape to recognize the real starting and ending of streams to maximize page locality. In an aspect, this approach can be implemented as an upper bound for stream length. For designs without any other indicator of stream length, this would determine where a new stream starts. For bridges with a concept of a stream, this could specify the upper bound of a stream length to prevent too much unfairness within the bandwidth sharing round.

In yet another implementation, instead of having a bridge use a counter to heuristically determine stream boundaries, AXI (Advanced eXtensible Interface) bridges could do a better job by actually looking at the request stream itself. If a large request (say 4 KB) is split into smaller requests (say 64B), it would be easy for the bridge to indicate that the request is part of a single stream and has good page-locality.

In another implementation, a more general-purpose approach would be for the bridge to look at the address of consecutive requests, whether they are from a split request or independent requests. If the address is in the same 4 KB page boundary, the bridge could determine that locality exists and mark the stream information appropriately. In an aspect, this can be implemented in all address-based bridge designs.

In another implementation, while source bridge could use a counter, some agents may have better information about whether a request is part of a stream or not. This is particularly true in the NSIP interface where page locality cannot be gleaned from the packet itself. But the agent itself may be able to pass this information with the packet stream, when it makes sense. If the new stream bit was added as an optional interface addition, the user could control locality directly. This can be an optional input to the NSIP bridge.

In an aspect relating to how new stream information can be passed along, in one exemplary implementation, the new stream information can be sent as an additional sideband bit, and can only be asserted on the Start-of-Packet (SOP) of a packet. Any additional flits for that packet can be marked as a continuation of the stream.

In another exemplary implementation, router can replicate this new stream information such that when a new stream bit arrives at an input VC, it should remember this for each possible output VC. When the first packet is sent from this input VC to an output VC, that packet should be marked as a new stream.

In an aspect, in one scenario when packets are coming on multiple input VCs going to the same output VC, page locality enhancement may attempt to stick with the same input VC until the stream has ended. However, if there are bubbles in the requests, it may switch early, and therefore the question is whether when it switches, the granted packet should be marked as a new stream or not. It is to be appreciated that it is a different stream since it is coming from a different input. Imagine a scenario with two stream, marked A and B. If bubbles created an interleaved traffic stream, we would see: A0, B0, A1, B1, A2, B2, A3, B3 . . . , in such a situation the first two requests in each stream can be marked as a new stream and shown in bold, whereas the later requests are not.

In another implementation, the hardware could choose to mark each of the packets as a new stream, or it could choose to not modify the stream information based on Input VC switching. If it marked it as a new stream, there would appear to be no page locality among these requests. If it left it as is, the requests after B0 would appear to be all from the same stream, even though they are not all related.

FIG. 6 illustrates an exemplary flow diagram showing working of the invention in accordance with an example implementation of the present disclosure. At step 602, the method can include the step of determining, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and at step 604, the method can include the step of re-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

FIG. 7 illustrates an example computer system on which example embodiments may be implemented. This example system is merely illustrative, and other modules or functional partitioning may therefore be substituted as would be understood by those skilled in the art. Further, this system may be modified by adding, deleting, or modifying modules and operations without departing from the scope of the inventive concept.

In an aspect, computer system 700 includes a server 702 that may involve an I/O unit 710, storage 712, and a processor 704 operable to execute one or more units as known to one skilled in the art. The term “computer-readable medium” as used herein refers to any medium that participates in providing instructions to processor 704 for execution, which may come in the form of computer-readable storage mediums, such as, but not limited to optical disks, magnetic disks, read-only memories, random access memories, solid state devices and drives, or any other types of tangible media suitable for storing electronic information, or computer-readable signal mediums, which can include transitory media such as carrier waves. The I/O unit processes input from user interfaces 714 and operator interfaces 716 which may utilize input devices such as a keyboard, mouse, touch device, or verbal command

The server 702 may also be connected to an external storage 718, 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 server may also be connected an output device 720, such as a display to output data and other information to a user, as well as request additional information from a user. The connections from the server 702 to the user interface 714, the operator interface 716, the external storage 718, and the output device 720 may via wireless protocols, such as the 802.11 standards, Bluetooth® or cellular protocols, or via physical transmission media, such as cables or fiber optics. The output device 720 may therefore further act as an input device for interacting with a user.

The processor 704 can include a current agent packet stream processing status determination module 706 that can be configured to determine, at an arbitrator, while selecting a NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if current NoC agent of current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at same part of said memory and enhances page locality; and a packet stream processing status based agent selection module 708 that can be configured to re-select, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round.

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.

Claims

1. A method for page locality based memory access request processing in a network-on-chip (NoC) architecture, said method comprising: determining, at an arbitrator, while selecting an NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if a current NoC agent of a current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at a same part of a memory and enhances page locality; andre-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round, wherein said determining of whether said packet stream is completely processed is performed based on receipt of new stream information, said new stream information being shared with each output port of a router associated with said arbitrator.

2. The method of claim 1, wherein if said packet stream processing is completed, said arbitrator selects the NoC agent having the next highest priority for the forthcoming round, and said current NoC agent is allocated lowest priority.

3. The method of claim 1, wherein said arbitrator is a round-robin based packet processing arbitrator.

4. The method of claim 1, wherein the arbitrator selects the NoC agent so as to enhance page locality without compromising on unfairness to said plurality of NoC agents beyond a defined threshold.

5. The method of claim 1, wherein said determining of whether said packet stream is completely processed is performed based on a stream indicator that forms part of said packet stream, such that said current NoC agent is selected in each forthcoming round until an end of packet stream indicator is received.

6. The method of claim 1, wherein said determining of whether said packet stream is completely processed is performed based on a destination address mentioned in each packet of said packet stream such that said current NoC agent is selected in each forthcoming round while the same destination address is mentioned in each packet of said packet stream.

7. The method of claim 1, wherein said determining of whether said packet stream is completely processed is performed based on an intimation from said current NoC agent that it is a part of said packet stream.

8. The method of claim 1, wherein said determining of whether said packet stream is completely processed is done by a master bridge by analyzing address stream that forms part of said packet stream.

9. A system for page locality based memory access request processing in a network-on-chip (NoC) architecture, said system comprising: a current agent packet stream processing status determination circuitry configured to determine, at an arbitrator, while selecting an NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if a current NoC agent of a current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at a same part of a memory and enhances page locality,a packet stream processing status based agent selection circuitry configured to re-select, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round, wherein said determining of whether said packet stream is completely processed is to be performed based on receipt of new stream information, said new stream information being shared with each output port of a router associated with said arbitrator.

10. The system of claim 9, wherein if said packet stream processing was completed, said arbitrator selects the NoC agent having the next highest priority for the forthcoming round, and said current NoC agent is allocated lowest priority.

11. The system of claim 9, wherein said arbitrator is a round-robin based packet processing arbitrator.

12. The system of claim 9, wherein the arbitrator selects the NoC agent so as to enhance page locality without compromising on unfairness to said plurality of NoC agents beyond a defined threshold.

13. The system of claim 9, wherein said determination of whether said packet stream is completely processed is to be performed based on any or a combination of: a stream indicator that forms part of said packet stream such that said current NoC agent is selected in each forthcoming round until an end of packet stream indicator is received, the destination address mentioned in each packet of said packet stream such that said current NoC agent is selected in each forthcoming round while the same destination address is mentioned in each packet of said packet stream, and an intimation from said current NoC agent that it is a part of said packet stream.

14. The system of claim 9, wherein said determination of whether said packet stream is completely processed is done by a master bridge by analyzing address stream that forms part of said packet stream.

15. A non-transitory computer readable storage medium storing instructions for executing a process, the instructions comprising: determining, at an arbitrator, while selecting an NoC agent from a plurality of NoC agents for request processing for a forthcoming round, if a current NoC agent of a current round is processing a packet stream and if said packet stream is completely processed at the end of said current round, wherein processing of the packet stream enables cluster requests to be processed at a same part of a memory and enhances page locality; andre-selecting, at said arbitrator, said current NoC agent as the NoC agent for the forthcoming round if said packet stream processing is not completed at the end of said current round, so as to enable said current NoC agent to complete processing of said packet stream in said forthcoming round, wherein said determining of whether said packet stream is completely processed is to be performed based on receipt of new stream information, said new stream information being shared with each output port of a router associated with said arbitrator.

16. The non-transitory computer readable storage medium according to claim 15, wherein if said packet stream processing was completed, said arbitrator selects the NoC agent having the next highest priority for the forthcoming round, and said current NoC agent is allocated lowest priority.

17. The non-transitory computer readable storage medium according to claim 15, wherein the arbitrator selects the NoC agent so as to enhance page locality without compromising on unfairness to said plurality of NoC agents beyond a defined threshold.

18. The non-transitory computer readable storage medium according to claim 15, wherein said determining of whether said packet stream is completely processed is performed based on any or a combination of: a stream indicator that forms part of said packet stream such that said current NoC agent is selected in each forthcoming round until an end of packet stream indicator is received, the destination address mentioned in each packet of said packet stream such that said current NoC agent is selected in each forthcoming round while the same destination address is mentioned in each packet of said packet stream, and an intimation from said current NoC agent that it is a part of said packet stream.