TECHNIQUES FOR COMMAND ARBITATION IN SYMMETRIC MULTIPROCESSOR SYSTEMS

• US 20190042486A1
• Filed: 08/02/2017
• Published: 02/07/2019
• Est. Priority Date: 08/02/2017
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: GUTHRIE, GUY L., MARINO, CHARLES, REDDY, PRAVEEN S.
• Granted Patent: US 10,642,760 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 12/0811 with multilevel cache hierarchies
• CPC Class Codes: G06F 12/0813 with a network or matrix configuration
• CPC Class Codes: G06F 12/0831 using a bus scheme, e.g. with bus monitoring or watching means
• CPC Class Codes: G06F 13/161 with latency improvement
• CPC Class Codes: G06F 2212/621 Coherency control relating to peripheral accessing, e.g. from DMA or I/O device

Abstract

A technique for operating a data processing system includes determining, by an arbiter of a processing unit of the data processing system, whether an over-commit has occurred. In response to determining that the over-commit has occurred, the arbiter selects a broadcast command to be dropped based on a number of hops traversed through the data processing system by the broadcast command.

Claims

1. A method of operating a data processing system, comprising:

determining, by an arbiter of a processing unit of the data processing system, whether an over-commit has occurred; and

in response to determining that the over-commit has occurred, selecting, by the arbiter, a broadcast command to be dropped based on a number of hops traversed through the data processing system by the broadcast command.

2. The method of claim 1, wherein multi-hop broadcast commands are weighted such that the arbiter is less likely to drop the multi-hop broadcast commands.

3. The method of claim 1, wherein the arbiter is configured to determine the number of hops based on information in the broadcast command.

4. The method of claim 3, wherein the information includes a source of the broadcast command from which the arbiter is configured to determine the number of hops currently traversed by the broadcast command.

5. The method of claim 3, wherein the information includes an indicator of the number of hops traversed through the data processing system by the broadcast command.

6. The method of claim 1, wherein the broadcast command includes a tag that specifies a chip identifier (ID) that identifies a chip that is a source of the broadcast command and a group ID that identifies a group for the source chip.

7. The method of claim 1, wherein the arbiter is configured to achieve fairness by considering a total number of processing units with current broadcast commands that are being arbitrated by the arbiter.

8. A data processing system, comprising:

a memory; and

a processing unit coupled to the memory, wherein the processing unit includes an arbiter that is configured to;

determine whether an over-commit has occurred; and

in response to determining that the over-commit has occurred, select a broadcast command to be dropped based on a number of hops traversed through the data processing system by the broadcast command.

9. The data processing system of claim 8, wherein multi-hop broadcast commands are weighted such that the arbiter is less likely to drop the multi-hop broadcast commands.

10. The data processing system of claim 8, wherein the arbiter is configured to determine the number of hops based on information in the broadcast command.

11. The data processing system of claim 10, wherein the information includes a source of the broadcast command from which the arbiter is configured to determine the number of hops currently traversed by the broadcast command.

12. The data processing system of claim 10, wherein the information includes an indicator of the number of hops traversed through the data processing system by the broadcast command.

13. The data processing system of claim 8, wherein the broadcast command includes a tag that specifies a chip identifier (ID) that identifies a chip that is a source of the broadcast command and a group ID that identifies a group for the source chip.

14. The data processing system of claim 8, wherein the arbiter is configured to achieve fairness by considering a total number of processing units with current broadcast commands that are being arbitrated by the arbiter.

15. A data processing system, comprising:

a first processing node including one or more first processing units; and

a second processing node coupled to the first processing node, wherein the second processing node includes one or more second processing units each having an arbiter that is configured to;

determine whether an over-commit has occurred; and

in response to determining that the over-commit has occurred, select a broadcast command to be dropped based on a number of hops traversed through the data processing system by the broadcast command.

16. The data processing system of claim 15, wherein multi-hop broadcast commands are weighted such that the arbiter is less likely to drop the multi-hop broadcast commands.

17. The data processing system of claim 15, wherein the arbiter is configured to determine the number of hops based on information in the broadcast command.

18. The data processing system of claim 17, wherein the information includes a source of the broadcast command from which the arbiter is configured to determine the number of hops currently traversed by the broadcast command.

19. The data processing system of claim 17, wherein the information includes an indicator of the number of hops traversed through the data processing system by the broadcast command.

20. The data processing system of claim 15, wherein the broadcast command includes a tag that specifies a chip identifier (ID) that identifies a chip that is a source of the broadcast command and a group ID that identifies a group for the source chip.