Method and System for Asymmetrical Processing With Managed Data Affinity

• US 20140380003A1
• Filed: 06/25/2013
• Published: 12/25/2014
• Est. Priority Date: 06/25/2013
• Status: Active Grant

• Current Assignee: Advanced Micro Devices, Inc.
• Original Assignee: Advanced Micro Devices, Inc.
• Inventors: Loh, Gabriel H., O'Connor, James Michael, Jayasena, Nuwan S., Hsu, Lisa R.
• Granted Patent: US 9,244,629 B2
• Time in Patent Office: Days
• US Class Current: 711/162
• CPC Class Codes: G06F 12/1036 for multiple virtual address spaces, e.g. segmentation G06F12/1045 takes precedence
• CPC Class Codes: G06F 2212/302 In image processor or graphics adapter
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
• CPC Class Codes: G06F 9/5033 considering data affinity

Abstract

Methods, systems and computer readable storage mediums for more efficient and flexible scheduling of tasks on an asymmetric processing system having at least one host processor and one or more slave processors, are disclosed. An example embodiment includes, determining a data access requirement of a task, comparing the data access requirement to respective local memories of the one or more slave processors selecting a slave processor from the one or more slave processors based upon the comparing, and running the task on the selected slave processor.

Claims

1. A method of processing in an asymmetrical processing system having at least one host processor and one or more slave processors, the method comprising:

selecting a slave processor from one or more slave processors based upon a data access requirement of a task and information about respective local memories of the one or more slave processors; and

running the task on the selected slave processor.

2. The method of claim 1, further comprising:

selecting the slave processor based upon comparing the data access requirement to the information about respective local memories of the one or more slave processors, wherein the comparing comprises;

detecting one or more accesses to a stored data by the task; and

determining one or more memory pages including the stored data, wherein the one or more memory pages are located in the respective local memories.

3. The method of claim 1, wherein the selecting a slave processor comprises:

detecting a data segment corresponding to the data access requirement in a second local memory which is not a local memory of the selected slave processor; and

copying the data segment to the local memory of the selected slave processor.

4. The method of claim 1, further comprising:

configuring the at least one host processor to execute an operating system; and

configuring the one or more slave processors without an operating system.

5. The method of claim 1, wherein the running comprises accessing a chip-stacked memory of the selected slave processor as a local memory.

6. The method of claim 1, further comprising:

configuring the selected slave processor to map virtual pages to physical pages using a page table in the local memory of the selected slave processor.

7. The method of claim 6, wherein the slave processor is further programmed to execute an operating system helper code.

8. The method of claim 1, further comprising:

receiving information regarding a failed memory allocation by the selected slave processor;

allocating a memory portion by the host processor; and

informing the slave processor regarding the allocated memory portion.

9. The method of claim 8, wherein the allocated memory portion is in the local memory of the selected slave processor.

10. The method of claim 8, wherein the allocated memory portion is in a second local memory, wherein the second local memory is local to a second slave processor.

11. The method of claim 1, further comprising:

configuring a pool of memory in the local memory of the selected slave processor for control by an operating system helper code executing on the selected slave processor.

12. The method of claim 1, further comprising:

configuring a subset of a page hierarchy that corresponds to one or more virtual address ranges of data contained within the local memory of the selected slave processor for control by the selected slave processor.

13. The method of claim 1, further comprising:

copying one or more selected data from a second processor to the selected slave processor, wherein the copied one or more selected data corresponds to the data access requirement of the task.

14. The method of claim 1, further comprising:

allocating operating system input/output file buffers in the associated local memory of the selected slave processor, wherein the allocated operating system input/output file buffers are configured for use by the task when executing on the selected slave processor.

15. An asymmetrical processing system, comprising:

at least one host processor executing an operating system;

one or more slave processors communicatively coupled to the at least one host processor, each of the slave processors having a local memory,

wherein the at least one host processor is configured to;

select a slave processor from the one or more slave processors based upon a data access requirement of a task and information about respective local memories of the one or more slave processors; and

run the task on the selected slave processor.

16. The asymmetrical processing system of claim 15, wherein the at least one host processor is further configured to:

detect one or more accesses to a stored data by the task; and

determine one or more memory pages including the stored data, wherein the one or more memory pages are located in the respective local memories.

17. The asymmetrical processing system of claim 15, wherein the respective local memory of each of the one or more slave processors is a chip-stacked memory.

18. The asymmetrical processing system of claim 15, wherein the at least one host processor is further configured to:

copy one or more selected data from a second processor to the selected slave processor, wherein the copied one or more selected data corresponds to the data access requirement of the task.

19. The asymmetrical processing system of claim 15, wherein the at least one host processor is further configured to:

allocate operating system input/output file buffers in the associated local memory of the selected slave processor, wherein the allocated operating system input/output file buffers are configured for use by the task when executing on the selected slave processor.

20. A non-transitory computer readable storage medium having instructions encoded thereon, execution of which by at least one host processor in an asymmetrical processing system having the at least one host processor and one or more slave processors cause execution of a method, the method including operations comprising:

selecting a slave processor from the one or more slave processors based upon a data access requirement of a task and information about respective local memories of the one or more slave processors; and

running the task on the selected slave processor.