Method and apparatus for unifying graphics processing unit computation languages

• US 9,275,426 B2
• Filed: 06/04/2012
• Issued: 03/01/2016
• Est. Priority Date: 06/04/2012
• Status: Active Grant

• Current Assignee: Adobe Inc.
• Original Assignee: Adobe Systems Incorporated (Adobe Inc.)
• Inventors: Hoeg, Steven Paterson
• Primary Examiner(s): Wu, Sing-Wai
• Application Number:
  US13/487,653
• Publication Number:
  US 20130321436A1
• Time in Patent Office: 1,366 Days
• US Class Current: 1/1
• CPC Class Codes: G06T 1/00 General purpose image data processing

Abstract

A method and apparatus for unifying graphics processing unit (GPU) computation languages is disclosed. The method comprises identifying a GPU of a computer system; accessing a plurality of macros representing a difference in source code between a first GPU computation language and a second GPU computation language, expanding each macro in the plurality of macros based on the identified GPU and executing a kernel on the computer system using the expanded macro.

Claims

1. A method for unifying Graphics Processing Unit (GPU) computation languages comprising:

initiating, by at least one processor, execution of an application on a computing device;

in response to initiating execution of the application, running a utility program to identify a computation language associated with a GPU on the computing device, wherein the GPU is associated with a first GPU computation language or a second GPU computation language;

wherein the first GPU computation language includes source code which is not present in the second GPU computation language, and

wherein the second GPU computation language includes source code that is not present in the first GPU computation language;

accessing a plurality of macros, wherein each macro of the plurality of macros comprises source code in the first GPU computation language and source code in the second GPU computation language;

expanding each macro in the plurality of macros based on the identified computation language associated with the GPU;

compiling, based on the expanded plurality of macros, a kernel using source code in the identified computation language associated with the GPU; and

executing the kernel on the computing device.

2. The method of claim 1, wherein the first GPU computation language and the second GPU computation language share code base similarities derived from a common programming language, and wherein each of the plurality of macros corresponds to a difference between a code base of the first GPU computation language and a code base of the second GPU computation language.

3. The method of claim 1, wherein the first GPU computation language is Compute Unified Device Architecture (CUDA) and the second GPU computation language is Open Computing Language (OpenCL).

4. The method of claim 1, wherein each macro of the plurality of macros represents a difference between the first GPU computation language and the second GPU computation language.

5. The method of claim 3, wherein running a utility program to identify a computation language associated with a GPU comprises determining whether a computer hardware is CUDA-based or OpenCL-based.

6. The method of claim 1, wherein expanding each macro comprises substituting language specific code into the kernel wherein the language specific code is compatible with the identified GPU.

7. The method of claim 6 wherein the kernel contains tokens where the language specific code is to be substituted.

8. A non-transitory computer readable medium for storing computer instructions that, when executed by at least one processor cause the at least one processor to perform a method for unifying Graphics Processing Unit (GPU) computation languages, the method comprising:

initiating execution of an application on a computing device;

in response to initiating execution of the application, running a utility program to identify a computation language associated with a GPU on the computing device, wherein the GPU is associated with a first GPU computation language or a second GPU computation language;

wherein the first GPU computation language includes source code which is not present in the second GPU computation language, and

wherein the second GPU computation language includes source code that is not present in the first GPU computation language;

accessing a plurality of macros, wherein each macro of the plurality of macros comprises source code in the first GPU computation language and source code in the second GPU computation language;

expanding each macro in the plurality of macros based on the identified computational language associated with the GPU;

compiling, based on the expanded plurality of macros, a kernel using source code in the identified computation language associated with the GPU; and

executing the kernel on the computing device.

9. The computer readable medium of claim 8, wherein the first GPU computation language and the second GPU computation language share code base similarities derived from a common programming language, and wherein each of the plurality of macros corresponds to a difference between a code base of the first GPU computation language and a code base of the second GPU computation language.

10. The computer readable medium of claim 8, wherein the first GPU computation language is Compute Unified Device Architecture (CUDA) and the second GPU computation language is Open Computing Language (OpenCL).

11. The computer readable medium of claim 8, wherein running a utility program to identify a computation language associated with a GPU comprises determining whether a computer hardware is CUDA-based or OpenCL-based.

12. The computer readable medium of claim 8, wherein expanding each macro comprises substituting language specific code into the kernel wherein the language specific code is compatible with the identified GPU.

13. The computer readable medium of claim 12, wherein the kernel contains tokens where the language specific code is to be substituted.

14. An apparatus for unifying Graphics Processing Unit (GPU) computation languages comprising at least one processor, a memory operatively associated with the at least one processor and having stored therein instructions executable by the at least one processor, wherein the instructions, when executed by the at least one processor, cause the apparatus to:

initiate execution of an application;

in response to initiating execution of the application, run a utility program to identify a computation language associated with a GPU on the computing device, wherein the GPU is associated with a first GPU computation language or a second GPU computation language;

wherein the first GPU computation language includes source code which is not present in the second GPU computation language, and

wherein the second GPU computation language includes source code that is not present in the first GPU computation language;

access a plurality of macros stored in memory, wherein each macro of the plurality of macros comprises source code in the first GPU computation language and source code in the second GPU computation language;

expand each macro in the plurality of macros based on the identity of the identified GPU; and

execute a kernel using the expanded macro;

wherein the first GPU computation language and the second GPU computation language share code base similarities derived from a common programming language, and wherein each of the plurality of macros corresponds to a difference between a code base of the first GPU computation language and a code base of the second GPU computation language.

15. The apparatus of claim 14, further comprising compiling, based on the expanded plurality of macros, the kernel using source code in the identified computation language associated with the GPU.

16. The apparatus of claim 14, wherein the first GPU computation language is Compute Unified Device Architecture (CUDA) and the second GPU computation language is Open Computing Language (OpenCL).

17. The apparatus of claim 14, wherein the kernel is configured to execute operations on one or more hardware and/or software modules responsive to execution of the kernel using the expanded macro.

18. The apparatus of claim 16, wherein the instructions, when executed by the at least one processor, further cause the apparatus to identify a computation language associated with the GPU by determining whether a computer hardware is CUDA-based or OpenCL-based.

19. The apparatus of claim 18, wherein the instructions, when executed stored in memory and executable by the at least one processor, further cause the apparatus to expand each macro by substituting language specific code into the kernel, wherein the language specific code is compatible with the identified computation language associated with the GPU.

20. The apparatus of claim 14, wherein the kernel contains tokens where the language specific code is to be substituted.