Constant versioning for multi-threaded processing

• US 7,877,565 B1
• Filed: 01/31/2006
• Issued: 01/25/2011
• Est. Priority Date: 01/31/2006
• Status: Active Grant

• Current Assignee: NVIDIA Corporation
• Original Assignee: NVIDIA Corporation
• Inventors: Allen, Roger L., Kong, Thomas H., Moy, Simon S., Everitt, Cass W., Moreton, Henry Packard
• Primary Examiner(s): PATEL, HETUL B
• Application Number:
  US11/344,445
• Time in Patent Office: 1,820 Days
• Field of Search: 711/165
• US Class Current: 711/165
• CPC Class Codes: G06F 9/383 Operand prefetching cache prefetching G06F12/0862
• CPC Class Codes: G06F 9/3851 from multiple instruction streams, e.g. multistreaming

Abstract

Systems and methods for using multiple versions of programmable constants within a multi-threaded processor allow a programmable constant to be changed before a program using the constants has completed execution. Processing performance may be improved since programs using different values for a programmable constant may execute simultaneously. The programmable constants are stored in a constant buffer and an entry of a constant buffer table is bound to the constant buffer. When a programmable constant is changed it is copied to an entry in a page pool and address translation for the page pool is updated to correspond to the old version (copy) of the programmable constant. An advantage is that the constant buffer stores the newest version of the programmable constant.

Claims

1. A method of storing multiple versions of a programmable constant for use during multi-threaded processing, comprising:

storing a first value of the programmable constant in an entry of a constant buffer located at a base address;

storing the base address of the constant buffer in a first entry of a constant buffer table;

receiving a second value of the programmable constant;

incrementing a version value stored in the first entry of the constant buffer table;

copying the first value stored in the entry of the constant buffer to a location in a page pool; and

storing the second value of the programmable constant in the entry of the constant buffer.

2. The method of claim 1, further comprising storing a pointer to the location in the page pool that stores the first value of the programmable constant in an address translation table, wherein the pointer is accessed using an index corresponding to the first entry of the constant buffer table and the version value.

3. The method of claim 1, further comprising deallocating the location in the page pool when the first value of the programmable constant is no longer used by the multi-threaded processing.

4. The method of claim 1, wherein the page pool and the constant buffer are stored in a shared memory.

5. The method of claim 1, further comprising performing multi-threaded processing of a first set of data using the first value of the programmable constant and performing multi-threaded processing of a second set of data using the second value of the programmable constant.

6. The method of claim 5, wherein the first set of data and the second set of data are at least one of vertex data, geometry data, and pixel data.

7. The method of claim 1, further comprising invalidating a cache entry corresponding to the first value of the programmable constant.

8. The method of claim 1, further comprising:

storing an index corresponding to the first entry of the constant buffer table in a slot table; and

incrementing a reference count stored in the first entry of the constant buffer table.

9. The method of claim 1, further comprising outputting a retire version feedback command including the index and version value.

10. A system for storing multiple versions of a programmable constant for use during multi-threaded processing, comprising:

a multiprocessor pipeline configured to execute shader program data processing commands using a set of programmable constants, wherein the set of programmable constants includes a current version of the programmable constant and an old version of the programmable constant;

a memory configured to store a constant buffer that includes the current programmable constant;

a page pool configured to store the old version of the programmable constant; and

a streaming processor array cache collector coupled between the memory and the multiprocessor pipeline and configured to generate the old version of the programmable constant by copying the current version of the programmable constant from the constant buffer to the page pool and changing the value stored as the current version of the programmable constant in the constant buffer.

11. The system of claim 10, wherein the streaming processor array cache collector is configured to update a version value corresponding to the constant buffer when the current programmable constant is generated.

12. The system of claim 11, further comprising a constant buffer table configured to store a base address and configured to store the version value in an entry, wherein the base address corresponds to a location of the constant buffer in the memory.

13. The system of claim 12, further comprising a slot table configured to bind a slot in the slot table to the constant buffer by storing an index corresponding to the entry in the slot.

14. The system of claim 10, wherein the streaming processor array cache collector is configured to retire a location in the page pool when the old version of the programmable constant stored in the location is no longer needed for the multi-threaded processing.

15. The system of claim 10, wherein the streaming processor array cache collector includes a cache configured to store one or more programmable constants in the set of programmable constants used for the multi-threaded processing.

16. The system of claim 15, wherein the streaming processor array cache collector is configured to invalidate an entry in the cache that stores the current version of the programmable constant when the old version of the programmable constant is generated.

17. The system of claim 10, wherein the multiprocessor pipeline includes a cache configured to store one or more programmable constants in the set of programmable constants used for the multi-threaded processing.

18. The system of claim 10, wherein the streaming processor array cache collector includes a write buffer configured to accumulate one or more programmable constants in the set of programmable constants to be stored in the page pool or to be stored in the constant buffer.

19. The system of claim 10, wherein the page pool is stored in the memory.

20. The system of claim 10, wherein the page pool is stored in the streaming processor array cache collector.