ENFORCING CENTRAL PROCESSING UNIT QUALITY OF SERVICE GUARANTEES WHEN SERVICING ACCELERATOR REQUESTS

• US 20190317807A1
• Filed: 04/16/2018
• Published: 10/17/2019
• Est. Priority Date: 04/16/2018
• Status: Active Grant

• Current Assignee: Advanced Micro Devices, Inc.
• Original Assignee: Advanced Micro Devices, Inc.
• Inventors: Basu, Arkaprava, Greathouse, Joseph Lee
• Granted Patent: US 11,275,613 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 11/3024 where the computing system component is a central processing unit [CPU]
• CPC Class Codes: G06F 11/3419 by assessing time
• CPC Class Codes: G06F 11/3433 for load management allocation of a server based on load conditions G06F9/505; load rebalancing G06F9/5083; redistributing the load in a network by a load balancer H04L67/1029
• CPC Class Codes: G06F 2201/81 Threshold
• CPC Class Codes: G06F 2201/865 Monitoring of software
• CPC Class Codes: G06F 9/466 Transaction processing
• CPC Class Codes: G06F 9/48 Program initiating; Program switching, e.g. by interrupt
• CPC Class Codes: G06F 9/4887 involving deadlines, e.g. rate based, periodic
• CPC Class Codes: G06F 9/5005 to service a request
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

Systems, apparatuses, and methods for enforcing processor quality of service guarantees when servicing system service requests (SSRs) are disclosed. A system includes a first processor executing an operating system and a second processor executing an application which generates SSRs for the first processor to service. The first processor monitors the number of cycles spent servicing SSRs over a previous time interval, and if this number of cycles is above a threshold, the first processor starts delaying the servicing of subsequent SSRs. In one implementation, if the previous delay was non-zero, the first processor increases the delay used in the servicing of subsequent SSRs. If the number of cycles is less than or equal to the threshold, then the first processor services SSRs without delay. As the delay is increased, the second processor begins to stall and its SSR generation rate falls, reducing the load on the first processor.

Claims

1. A system comprising:

a first processor comprising circuitry configured to execute a plurality of threads of an operating system kernel; and

a second processor coupled to the first processor, wherein the second processor comprises circuitry configured to execute an application and send system service requests to the first processor for servicing;

wherein the first processor is configured to;

monitor a number of cycles that threads executing on the first processor have spent servicing system service requests over a previous time interval; and

dynamically adjust an amount of delay that is added to servicing of a given system service request based on the number of cycles.

2. The system as recited in claim 1, wherein dynamically adjusting the amount of delay that is added to servicing of the given system service request comprises:

adding a first amount of delay responsive to determining that the number of cycles is greater than a threshold; and

adding a second amount of delay responsive to determining that the number of cycles is less than or equal to the threshold, wherein the second amount of delay is less than the first amount of delay.

3. The system as recited in claim 2, wherein the circuitry of the first processor is configured to set the first amount of delay to a value greater than a previous amount of delay responsive to the number of cycles being greater than the threshold and the previous amount of delay being greater than zero.

4. The system as recited in claim 1, wherein dynamically adjusting the amount of delay that is added to servicing of the given system service request comprises:

waiting a first duration prior to initiating servicing of the given system service request responsive to determining that the number of cycles is greater than a threshold; and

waiting a second duration prior to initiating servicing of the given system service request responsive to determining that the number of cycles is less than or equal to the threshold, wherein the second duration is less than the first duration.

5. The system as recited in claim 4, wherein the threshold is dynamically adjusted by the operating system kernel.

6. The system as recited in claim 1, wherein the first processor is further configured to:

calculate an overhead associated with servicing system service requests over the previous time interval; and

dynamically adjust an amount of delay that is added to servicing of a given system service request based on the overhead.

7. The system as recited in claim 1, wherein the circuitry of the first processor is further configured to:

calculate an overhead associated with servicing system service requests over the previous time interval, wherein the overhead comprises one or more of cycles spent servicing system service requests, cycles spent transitioning between user mode and kernel mode, and cycles spent in user mode running at a lower instruction per cycle (IPC) rate; and

dynamically adjust an amount of delay that is added to servicing of a given system service request based on the overhead.

8. A method comprising:

monitoring, by a first processor, a number of cycles that threads executing on the first processor have spent servicing system service requests generated by a second processor over a previous time interval; and

dynamically adjusting an amount of delay that is added to servicing of a given system service request based on the number of cycles that threads of the first processor have spent servicing system service requests over the previous time interval.

9. The method as recited in claim 8, wherein dynamically adjusting the amount of delay that is added to servicing of the given system service request comprises:

adding a first amount of delay responsive to determining that the number of cycles is greater than a threshold; and

adding a second amount of delay responsive to determining that the number of cycles is less than or equal to the threshold, wherein the second amount of delay is less than the first amount of delay.

10. The method as recited in claim 9, further comprising setting the first amount of delay to a value greater than a previous amount of delay responsive to the number of cycles being greater than the threshold and the previous amount of delay being greater than zero.

11. The method as recited in claim 8, wherein dynamically adjusting the amount of delay that is added to servicing of the given system service request comprises:

waiting a first duration prior to initiating servicing of the given system service request responsive to determining that the number of cycles is greater than a threshold; and

waiting a second duration prior to initiating servicing of the given system service request responsive to determining that the number of cycles is less than or equal to the threshold, wherein the second duration is less than the first duration.

12. The method as recited in claim 11, wherein the threshold is set by an operating system kernel.

13. The method as recited in claim 8, further comprising:

calculating an overhead associated with servicing system service requests over the previous time interval; and

dynamically adjusting an amount of delay that is added to servicing of a given system service request based on the overhead.

14. The method as recited in claim 8, further comprising:

calculating an overhead associated with servicing system service requests over the previous time interval, wherein the overhead comprises one or more of cycles spent servicing system service requests, cycles spent transitioning between user and kernel mode, and cycles spent in user mode running at a lower instruction per cycle (IPC) rate; and

dynamically adjusting an amount of delay that is added to servicing of a given system service request based on the overhead.

15. An apparatus comprising:

a memory;

one or more execution units coupled to the memory;

wherein the apparatus is configured to;

monitor a number of cycles that threads executing on the one or more execution units have spent servicing system service requests over a previous time interval; and

dynamically adjust an amount of delay that is added to servicing of a given system service request based on the number of cycles.

16. The apparatus as recited in claim 15, wherein dynamically adjusting the amount of delay that is added to servicing of the given system service request comprises:

adding a first amount of delay responsive to determining that the number of cycles is greater than a threshold; and

adding a second amount of delay responsive to determining that the number of cycles is less than or equal to the threshold, wherein the second amount of delay is less than the first amount of delay.

17. The apparatus as recited in claim 16, wherein the apparatus is further configured to set the first amount of delay to a value greater than a previous amount of delay responsive to the number of cycles being greater than the threshold and the previous amount of delay being greater than zero.

18. The apparatus as recited in claim 15, wherein dynamically adjusting the amount of delay that is added to servicing of the given system service request comprises:

waiting a first duration prior to initiating servicing of the given system service request responsive to determining that the number of cycles is greater than a threshold; and

waiting a second duration prior to initiating servicing of the given system service request responsive to determining that the number of cycles is less than or equal to the threshold, wherein the second duration is less than the first duration.

19. The apparatus as recited in claim 18, wherein the threshold is set by the operating system.

20. The apparatus as recited in claim 15, wherein the the apparatus is further configured to:

calculate an overhead associated with servicing system service requests over the previous time interval; and

dynamically adjust an amount of delay that is added to servicing of a given system service request based on the overhead.