System and method for network on chip construction through machine learning

• US 10,313,269 B2
• Filed: 12/26/2016
• Issued: 06/04/2019
• Est. Priority Date: 12/26/2016
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Netspeed Systems, Inc. (Intel Corporation)
• Inventors: Raponi, Pier Giorgio
• Primary Examiner(s): Wasel, Mohamed
• Application Number:
  US15/390,705
• Publication Number:
  US 20180183715A1
• Time in Patent Office: 890 Days
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: H04L 49/109 Integrated on microchip, e.g. switch-on-chip
• CPC Class Codes: H04L 49/15 Interconnection of switching modules
• CPC Class Codes: H04L 49/254 Centralised controller, i.e. arbitration or scheduling

Abstract

In example implementations, the specification is processed to determine the characteristics of the NoC to be generated, the characteristics of the flow (e.g. number of hops, bandwidth requirements, type of flow such as request/response, etc.), flow mapping decision strategy (e.g., limit on number of new virtual channels to be constructed, using of existing VCs, yx/xy mapping), and desired strategy to be used for how the flows are to be mapped to the network. In such processing, the machine learning algorithm can provide a determination as to if a flow is acceptable or not in view of the specification (e.g., via a Q score). In example implementations, the machine learning decisions can be applied on a flow by flow basis, and can involve supervised learning and unsupervised learning algorithms.

Claims

1. A method of generating a Network on Chip (NoC), the method comprising:

applying a process on a NoC specification to determine, from a plurality of NoC mapping strategies, ones of the plurality of NoC mapping strategies that meet a threshold for a cost function;

executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings;

scoring the one or more NoC mappings based on the cost function; and

generating the NoC from an implementation of a selected mapping from the one or more NoC mappings;

wherein the process is a machine learning algorithm trained against the cost function for the plurality of NoC mapping strategies.

2. The method of claim 1, wherein the cost function is based on at least one of:

a NoC bandwidth, NoC buffer size, NoC latency, and NoC link cost.

3. The method of claim 1, wherein the plurality of mapping strategies are represented as a vector indicative of which of the plurality of mapping strategies meet the threshold of the cost function;

wherein the ones of the plurality of mapping strategies are selected based on the vector.

4. The method of claim 3, wherein the ones of the plurality of mapping strategies is selected based on a machine learning process.

5. The method of claim 1, wherein the executing the ones of the plurality of mapping strategies is conducted on a flow by flow basis for each of the ones of the plurality of mapping strategies, wherein each flow is kept or discarded for the mapping based on a minimization of the cost function.

6. The method of claim 1, wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a first sorting function configured to specify an order in which traffic flows are to be mapped.

7. The method of claim 6, wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a second sorting function configured to modify the order in which the traffic flows are to be mapped.

8. The method of claim 6, further comprising, for each flow in the order in which traffic flow is mapped, selecting a mapping strategy, based on a NoC state, mapping the each flow according to the selected mapping strategy, and updating the NoC state after mapping the each flow.

9. The method of claim 1, wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a machine learning process to determine an order in which traffic flows are to be mapped.

10. A non-transitory computer readable medium storing instructions for generating a Network on Chip (NoC), the instructions comprising:

applying a process on a NoC specification to determine, from a plurality of NoC mapping strategies, ones of the plurality of NoC mapping strategies that meet a threshold for a cost function;

executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings;

scoring the one or more NoC mappings based on the cost function; and

generating the NoC from an implementation of a selected mapping from the one or more NoC mappings;

wherein the process is a machine learning algorithm trained against the cost function for the plurality of NoC mapping strategies.

11. The non-transitory computer readable medium of claim 10, wherein the cost function is based on at least one of:

a NoC bandwidth, NoC buffer size, NoC latency, and NoC link cost.

12. The non-transitory computer readable medium of claim 10, wherein the plurality of mapping strategies are represented as a vector indicative of which of the plurality of mapping strategies meet the threshold of the cost function;

wherein the ones of the plurality of mapping strategies are selected based on the vector.

13. The non-transitory computer readable medium of claim 12, wherein the ones of the plurality of mapping strategies is selected based on a machine learning process.

14. The non-transitory computer readable medium of claim 10, wherein the executing the ones of the plurality of mapping strategies is conducted on a flow by flow basis for each of the ones of the plurality of mapping strategies, wherein each flow is kept or discarded for the mapping based on a minimization of the cost function.

15. The non-transitory computer readable medium of claim 10, wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a first sorting function configured to specify an order in which traffic flows are to be mapped.

16. The non-transitory computer readable medium of claim 15, wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a second sorting function configured to modify the order in which the traffic flows are to be mapped.

17. The non-transitory computer readable medium of claim 15, further comprising, for each flow in the order in which traffic flow is mapped, selecting a mapping strategy, based on a NoC state, mapping the each flow according to the selected mapping strategy, and updating the NoC state after mapping the each flow.

18. The non-transitory computer readable medium of claim 15, wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a machine learning process to determine an order in which traffic flows are to be mapped.

19. A method of generating a Network on Chip (NoC), the method comprising:

applying a process on a NoC specification to determine, from a plurality of NoC mapping strategies, ones of the plurality of NoC mapping strategies that meet a threshold for a cost function;

executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings;

scoring the one or more NoC mappings based on the cost function; and

generating the NoC from an implementation of a selected mapping from the one or more NoC mappings;

wherein the executing the ones of the plurality of NoC mapping strategies to generate one or more NoC mappings comprises executing a machine learning process to determine an order in which traffic flows are to be mapped.