Optimization of cache configuration for application design

• US 8,595,439 B1
• Filed: 04/17/2012
• Issued: 11/26/2013
• Est. Priority Date: 09/28/2007
• Status: Active Grant

• Current Assignee: The MathWorks, Inc.
• Original Assignee: The MathWorks, Inc.
• Inventors: Koh, David, Belge, Murat, Weixel, James K.
• Primary Examiner(s): Chery, Mardochee
• Application Number:
  US13/449,073
• Time in Patent Office: 588 Days
• Field of Search: 711/118, 711/122, 711/123
• US Class Current: 711/118
• CPC Class Codes: G06F 11/3442 for planning or managing the needed capacity
• CPC Class Codes: G06F 11/3466 Performance evaluation by tracing or monitoring
• CPC Class Codes: G06F 12/0802 Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
• CPC Class Codes: G06F 2201/885 Monitoring specific for caches
• CPC Class Codes: G06F 8/30 Creation or generation of source code

Abstract

A device may execute application code in a first cache environment to obtain a first result. The first cache environment may be based on a first cache configuration that is associated with the application code. The device may determine a second cache configuration based on the first result and execute the application code in a second cache environment to obtain a second result. The second cache environment may be based on the second cache configuration. The device may select one of the first cache configuration or the second cache configuration as a selected cache configuration for the application code based on comparing the first result with the second result, and may configure the one or more caches based on the selected one of the first cache configuration or the second cache configuration.

Claims

1. A device comprising:

a multi-core processor communicatively coupled to one or more caches, each core, of the multi-core processor, to;

execute application code in a first cache environment to obtain a first result,

the first cache environment being based on a first cache configuration that is associated with the application code,

the first result including a first execution time associated with executing the application code in the first cache environment,

determine a second cache configuration based on the first result,

execute the application code in a second cache environment to obtain a second result,

the second cache environment being based on the second cache configuration,

the second result including a second execution time associated with executing the application code in the second cache environment,

compare the first result, including the first execution time, with the second result including the second execution time,

select one of the first cache configuration or the second cache configuration as a selected cache configuration for the application code based on comparing the first result with the second result,

the first cache configuration being selected as the selected cache configuration for the application code when the first execution time is less than the second execution time, and

the second cache configuration being selected as the selected cache configuration for the application code when the second execution time is less than the first execution time, and

configure the one or more caches based on the selected one of the first cache configuration or the second cache configuration.

2. The device of claim 1, where, when comparing the first result with the second result, each core, of the multi-core processor, is further to:

determine that the first execution time is less than the second execution time, and

determine that the first execution time is less than a threshold execution time based on the first execution time being less than the second execution time, and

where, when selecting the one of the first cache configuration or the second cache configuration, each core, of the multi-core processor is further to;

select the first cache configuration as the selected cache configuration for the application code based on the first execution time being less than the threshold execution time.

3. The device of claim 1, where each core, of the multi-core processor, is further to:

receive one or more of the application code, information identifying a cache condition, or information identifying one or more rules associated with configuring the one or more caches, and

determine the first cache configuration based on the one or more of the application code, the information identifying the cache condition, or the information identifying the one or more rules.

4. The device of claim 3, where at least one core, of the multi-core processor, receives the information identifying the cache condition, and

where the information identifying the cache condition includes one or more of a quantity of levels of cache available, a cache size associated with each level of the levels of cache available, or a quantity of execution times associated with a portion of the application code.

5. The device of claim 1, where each core, of the multi-core processor, is further to:

generate an intermediate representation based on a model, and

determine the first cache configuration based on the generated intermediate representation.

6. The device of claim 1, where each core, of the multi-core processor, is further to:

perform a static cache analysis on the application code to obtain static cache performance results, and

determine the second cache configuration based on the static cache performance results.

7. A method comprising:

executing application code in a first cache environment to obtain a first result,

the first cache environment being based on a first cache configuration that is associated with the application code,

the first result including a first execution time associated with executing the application code in the first cache environment, and

the executing the application code in the first cache environment being performed by a multi-core processor;

executing the application code in a second cache environment to obtain a second result,

the second cache environment being based on a second cache configuration that is determined based on the first result,

the second result including a second execution time associated with executing the application code in the second cache environment,

the second cache environment differing from the first cache environment, and

the executing the application code in the second cache environment being performed by the multi-core processor;

comparing the first result, including the first execution time, with the second result including the second execution time,

the comparing the first result with the second result being performed by the multi-core processor,

where comparing the first result with the second result includes;

determining whether the first execution time is less than the second execution time; and

selecting one of the first cache configuration or the second cache configuration as a selected cache configuration for the application code, based on comparing the first result with the second result,

the selecting being performed by the multi-core processor,

the first cache configuration being selected as the selected cache configuration for the application code when the first execution time is less than the second execution time, and

the second cache configuration being selected as the selected cache configuration for the application code when the first execution time is not less than the second execution time.

8. The method of claim 7, further comprising:

determining that the first execution time is greater than a threshold execution time; and

executing the application code in the second cache environment based on the first execution time being greater than the threshold execution time.

9. The method of claim 7, further comprising:

receiving, from a client via a network, one or more of the application code, information identifying a cache condition, or information identifying one or more rules associated with configuring one or more caches associated with the multi-core processor; and

determining the first cache configuration based on the one or more of the application code, the information identifying the cache condition, or the information identifying the one or more rules.

10. The method of claim 7, further comprising:

outputting specifications for building the selected one of the first cache configuration or the second cache configuration to a client via a network.

11. The method of claim 10, where outputting the specifications further includes:

outputting information to align one or more portions of the application code at one or more cache line boundaries of a cache of the client.

12. The method of claim 7, further comprising:

generating an intermediate representation based on a block diagram model; and

determining the first cache configuration based on the generated intermediate representation.

13. A device comprising:

a multi-core processor communicatively coupled to one or more caches, the multi-core processor to;

execute the application code in a first cache environment to obtain a first result,

the first cache environment being based on a first cache configuration,

the first result including a first execution time associated with executing the application code in the first cache environment,

determine a second cache configuration based on the first result,

execute the application code in a second cache environment to obtain a second result,

the second cache environment being based on the second cache configuration,

the second result including a second execution time associated with executing the application code in the second cache environment,

compare the first result with the second result, and

select one of the first cache environment or the second cache environment as a selected cache environment based on comparing the first result with the second result,

the first cache environment being selected as the selected cache environment when the first execution time is less than the second execution time, and

the second cache environment being selected as the selected cache environment when the first execution time is not less than the second execution time.

14. The device of claim 13, where the multi-core processor is further to:

generate the first cache environment based on hardware description language (HDL) code.

15. The device of claim 13, where the multi-core processor is further to:

obtain a model associated with the application code,

generate, based on the model, information associated with an intermediate representation of the model, and

generate the first cache environment based on the generated information.

16. The device of claim 15, where the generated information includes one or more of:

information identifying a frequency of execution of a section of the application code,

information identifying a frequency of access of a data section associated with the application code, or

information identifying a timeline of execution and data access of the section of the application code.

17. The device of claim 13, where the multi-core processor is further to:

align or group sections of the application code in the one or more caches based on the selected one of the first cache environment or the second cache environment.

18. The device of claim 13, where the multi-core processor is further to:

output, to a client, information to align or group sections of the application code in a cache of the client based on the selected one of the first cache environment or the second cache environment.

19. A non-transitory computer-readable medium storing instructions, the instructions comprising:

one or more instructions executable by a multi-core processor to;

execute application code in a first cache environment to obtain a first result,

the first cache environment being based on a first cache configuration that is associated with the application code, and

the first result including a first execution time associated with executing the application code in the first cache environment,

determine a second cache configuration based on the first result,

execute the application code in a second cache environment to obtain a second result,

the second cache environment being based on the second cache configuration, and

the second result including a second execution time associated with executing the application code in the second cache environment,

compare the first result with the second result,

select one of the first cache configuration or the second cache configuration as a selected cache configuration for the application code based on comparing the first result with the second result,

the first cache configuration being selected as the selected cache configuration for the application code when the first execution time is less than the second execution time, and

the second cache configuration being selected as the selected cache configuration for the application code when the first execution time is not less than the second execution time, and

configure one or more caches based on the selected cache configuration.

20. The non-transitory computer-readable medium of claim 19, where the instructions further comprise:

one or more instructions to align or group sections of the application code in one or more caches based on the selected cache environment.

21. A non-transitory computer-readable medium storing instructions, the instructions comprising:

one or more instructions executable by a multi-core processor to;

perform a static cache analysis on application code to obtain a first result,

execute the application code in a first cache environment to obtain a second result,

the first cache environment being based on a first cache configuration,

the second result including a first execution time associated with executing the application code in the first cache environment,

determine a second cache configuration based on the first result and the second result,

execute the application code in a second cache environment to obtain a third result,

the second cache environment being based on the second cache configuration,

the third result including a second execution time associated with executing the application code in the second cache environment,

compare the second result with the third result, and

select one of the first cache environment or the second cache environment as a selected cache environment based on comparing the second result with the third result,

the first cache environment being selected as the selected cache environment when the first execution time is less than the second execution time, and

the second cache environment being selected as the selected cache environment when the first execution time is not less than the second execution time.

22. The non-transitory computer-readable medium of claim 21, where the instructions further comprise:

one or more instructions to align or group sections of the application code in one or more caches based on the selected cache environment.