Latency guaranteed network on chip

• US 10,355,975 B2
• Filed: 10/19/2016
• Issued: 07/16/2019
• Est. Priority Date: 10/19/2016
• Status: Active Grant

• Current Assignee: REX Computing, Inc.
• Original Assignee: REX Computing, Inc.
• Inventors: Sebexen, Paul Michael, Sohmers, Thomas Rex, Cote, Edmond A., Moezzi Madani, Nariman, Kasat, Piyush Shrinivas
• Primary Examiner(s): Sciacca, Scott M
• Application Number:
  US15/298,180
• Publication Number:
  US 20180109452A1
• Time in Patent Office: 1,000 Days
• CPC Class Codes: H04L 45/127 based on intermediate node capabilities
• CPC Class Codes: H04L 49/109 Integrated on microchip, e.g. switch-on-chip
• CPC Class Codes: H04L 49/3063 Pipelined operation
• CPC Class Codes: H04L 49/901 using storage descriptor, e.g. read or write pointers

Abstract

A method and system including a set of processor cores; and a set of routers each including a set of input ports and a set of output ports, where: each processor core of the set of processor cores corresponds to a different router of the set of routers and is communicatively coupled with a corresponding router via the router'"'"'s set of input ports and set of output ports, based on a physical destination address of a data packet, each router is operable to send one or more data packets to the one or more adjacent routers or the processor core corresponding to the router, where each router is operable to retain a data packet in the event of a traffic condition, and each router implements a deterministic routing policy.

Claims

1. A system comprising:

a set of processor cores;

a set of routers each including a set of input ports and a set of output ports, wherein;

each processor core of the set of processor cores corresponds to a different router of the set of routers, wherein each processor core and corresponding router form a tile,

each processor core is communicatively coupled with a corresponding router via the router'"'"'s set of input ports and set of output ports,

each router is communicatively coupled with one or more adjacent routers via the set of input ports and the set of output ports, and wherein each processor core is communicatively coupled with the other processor cores via the set of routers,

each router is operable to receive one or more data packets from the one or more adjacent routers or the processor core corresponding to the router,

based on a physical destination address of a data packet, each router is operable to send one or more data packets to the one or more adjacent routers or the processor core corresponding to the router, wherein each router is operable to retain a data packet in the event of a traffic condition, and

each router implements a static priority routing policy; and

an optimization module configured to;

determine optimal function assignment configurations for groups of tiles, and

assign two or more functions, which communicate at least unilaterally more frequently with one another than with other functions, to groups of adjacent tiles based on an optimal function assignment configuration determination, wherein the two or more functions are assigned to groups of tiles communicatively coupled in square configurations when the function executes optimally when executed by the groups of tiles communicatively coupled in the square configurations and are assigned to groups of tiles communicatively coupled in linear configurations when the function executes optimally when executed by the groups of tiles communicatively coupled in the linear configurations.

2. The system of claim 1, wherein:

a traffic condition occurs when packets of more than one input port of a particular router are to be sent via the same output port of the particular router; and

the static priority routing policy assigns priority levels to each input port of the particular router, causing data packets from the higher priority input port to be sent before data packets from the lower priority input ports in the event of the traffic condition.

3. The system of claim 1, wherein a traffic condition occurs when a particular router is ready to send a data packet to another router adjacent to the particular router that is not ready to receive the data packet.

4. The system of claim 1, wherein each router utilizes a set of FlFO memory elements, and wherein:

each FIFO memory element of a particular router is operable to store a data packet for subsequent sending to a different router of one or more routers adjacent to the particular router or the processor core corresponding to the router; and

each FIFO memory element is operable to retain a data packet in the event of a traffic condition.

5. The system of claim 4, wherein each FIFO memory element is a single stage memory element.

6. The system of claim 4, wherein each FIFO memory element is a D flip-flop.

7. The system of claim 1, wherein each router of the set of routers does not include SRAM (Static Random-Access Memory) and does not utilize SRAM for routing processes.

8. The system of claim 1, wherein the set of routers are arranged in a grid configuration.

9. The system of claim 8, wherein:

each processor core includes a physically addressable memory module, and the memory module of adjacent cores include sequential physical addressing with one another; and

the routers determine to which adjacent router to send a data packet based on a physical destination address of the data packet and the sequential physical addressing.

10. The system of claim 1, wherein the static priority routing policy assigns static priority to designated input ports of a router.

11. The system of claim 1, wherein each router is operable to simultaneously receive data packets from one or more input ports and simultaneously send data packets on one or more output ports.

12. The system of claim 1, wherein the circuit design defining each router is identical and the circuit design defining each processor core is identical.

13. A network-on-chip microprocessor chip, comprising:

a set of scratchpad memory modules;

a set of tiles arranged in a grid configuration, each tile including a processor core and a router communicatively coupled with one another, wherein;

each processor core corresponds to and is communicatively coupled with a different scratchpad memory module of the set of scratchpad memory modules,

each router includes a set of input ports and a set of output ports, wherein each output port includes a FIFO memory element operable to store a data packet for subsequent sending to a router of an adjacent tile,

based on a physical destination address of a data packet, each router is operable to send one or more data packets to routers of one or more adjacent tiles or the processor core corresponding to the router, and

each router implements a static priority routing policy in the event of a traffic condition; and

an optimization module configured to;

determine optimal function assignment configurations for groups of tiles of the set of tiles, and

assign two or more functions, which communicate at least unilaterally more frequently with one another than with other functions, to groups of adjacent tiles based on an optimal function assignment configuration determination, wherein the two or more functions are assigned to groups of tiles communicatively coupled in square configurations when the function executes optimally when executed by the groups of tiles communicatively coupled in the square configurations and are assigned to groups of tiles communicatively coupled in linear configurations when the function executes optimally when executed by the groups of tiles communicatively coupled in the linear configurations.

14. The network-on-chip microprocessor chip of claim 13, wherein:

a traffic condition occurs when packets of more than one input port of a particular router are to be sent via the same output port of the particular router; and

the static priority routing policy assigns priority levels to each input port of the particular router, causing data packets from the higher priority input port to be sent before data packets from the lower priority input ports in the event of the traffic condition.

15. The network-on-chip microprocessor chip of claim 13, wherein a traffic condition occurs when a particular router is ready to send a data packet to a router of an adjacent tile that is not ready to receive the data packet.

16. The network-on-chip microprocessor chip of claim 13, wherein each FIFO memory element is operable to retain a data packet in the event of a traffic condition.

17. The network-on-chip microprocessor chip of claim 13, wherein:

adjacent scratchpad memory modules include sequential physical addressing with one another; and

the routers determine to which adjacent router to send a data packet based on a physical destination address of the data packet and the sequential physical addressing.

18. The network-on-chip microprocessor chip of claim 13, wherein each router is operable to provide a data packet to a router of an adjacent tile in one clock cycle.

19. The network-on-chip microprocessor chip of claim 13, wherein the circuit design defining each tile is identical.