Method and apparatus for compression of multi-sampled anti-aliasing color data

• US 8,774,535 B2
• Filed: 01/18/2012
• Issued: 07/08/2014
• Est. Priority Date: 02/13/2003
• Status: Active Grant

• Current Assignee: ATI Technologies ULC (Advanced Micro Devices, Inc.)
• Original Assignee: ATI Technologies ULC (Advanced Micro Devices, Inc.)
• Inventors: Van Hook, Timothy J., Fouladi, Farhad, Elder, Gordon III
• Primary Examiner(s): MOTSINGER, SEAN T
• Application Number:
  US13/352,455
• Publication Number:
  US 20120183215A1
• Time in Patent Office: 902 Days
• Field of Search: None
• US Class Current: 382/232
• CPC Class Codes: G06T 11/40 Filling a planar surface by adding surface attributes, e.g. colour or texture
• CPC Class Codes: G06T 9/00 Image coding bandwidth or redundancy reduction for static pictures H04N1/41; coding or decoding of static colour picture signals H04N1/64; methods or arrangements for coding, decoding, compressing or decompressing digital video signals H04N19/00
• CPC Class Codes: H04N 1/644 using a reduced set of representative colours, e.g. each representing a particular range in a colour space
• CPC Class Codes: H04N 19/186 the unit being a colour or a chrominance component
• CPC Class Codes: H04N 19/423 characterised by memory arrangements H04N19/433 takes precedence

Abstract

The present invention provides a scheme for compressing the color components of image data, and in particular, data used in multi-sampled anti-aliasing applications. Adjacent pixels are grouped into rectangular tiles, with the sample colors stored in compressed formats accessible via an encoded pointer. In one embodiment, duplicate colors are stored once. Unlike prior compression schemes that rely on pixel to pixel correlation, the present invention takes advantages of the sample to sample correlation that exists within the pixels. A memory and graphics processor configuration incorporating the tile compression schemes is also provided. The configuration defines the tile sizes in main memory and cache memory. In one embodiment, graphics processor relies on a Tile Format Table (TFT) to process incoming tiles in compressed formats. The present invention reduces memory consumption and speeds up essential and oft-repeated operations in rendering. Thus it is valuable in the design and manufacture of graphic sub-systems.

Claims

1. A method for compressing image data, comprising:

retrieving tiles of pixel data from a main memory;

resizing said tiles;

compressing said resized tiles, wherein said compressed tiles are stored in cache memory after compression;

processing said compressed tiles in a graphics processor

returning said compressed tiles from said graphics processor to cache memory;

resizing said compressed tiles in cache memory for storage in said main memory; and

returning said resized tiles to main memory.

2. The method of claim 1 wherein said graphics processor comprises:

a first Tile Format Table (TFT), wherein said graphics processor uses information from said TFT to process said compressed tiles.

3. The method of claim 2 wherein said first TFT comprises:

a compression format entry for each of said tiles in said main memory.

4. The method of claim 3 wherein said entry comprises an encoding for a plurality of states for each of said resized tiles, said states comprising:

a full compression state;

a partial compression state;

an uncompressed state; and

a clear state.

5. The method of claim 2 wherein said graphics processor comprises:

a second TFT, wherein said second TFT comprises a compression format entry for each of said compressed tiles in said cache memory.

6. The method of claim 5 wherein said entry in said second TFT comprises an encoding for a plurality of states for each of said resized tiles, said states comprising:

a full compression state;

a partial compression state;

an uncompressed state; and

a clear state.

7. The method of claim 6 wherein said size of tiles in said main memory and size of compressed tiles in said cache memory differ.

8. The method of claim 6 wherein tiles in said main memory is of size 4×

4 and compressed tiles in said cache memory is of size 2×

2.

9. The method of claim 8 wherein said resizing said compressed tiles in cache memory for storage in said main memory further comprises:

combining four 2×

2 tiles in said cache memory into a 4×

4 tile;

choosing a lowest common compression level from among said four 2×

2 tiles; and

using said lowest common compression level to compress said 4×

4 tile, wherein said 4×

4 is returned for storage in said main memory.

10. The method of claim 6 wherein said processing further comprises:

clearing entries in said cache memory by setting each entry of said second TFT to said clear state; and

writing a clear color storied on said graphics processor in said entries in said cache memory.

11. The method of claim 1 wherein said processing does not decompress said compressed tiles prior to processing said compressed tiles in a graphics processor.

12. The method of claim 1 further comprising:

accessing color information of samples in each of said pixel in said resized tiles, wherein duplicate color information is accessed once; and

combining color information samples to obtain one color for said pixel.

13. The method of claim 1 wherein said compressing further comprise:

evaluating said resized tiles for compression suitability, wherein said compression recognizes duplicate data and reduces amount of duplicate data stored within said resized tiles; and

compressing said resized tiles if said resized tiles are deemed suitable for said compression.

14. The method of claim 13 wherein said evaluating further comprises:

determining whether a resized tile is suitable for full compression; and

designating said resized tile for full compression if it is deemed to be suitable.

15. The method of claim 14 wherein said determining further comprises:

determining whether said resized tile is wholly covered by a triangle primitive.

16. The method of claim 14 wherein said compressing further comprises:

storing a single color entry for each pixel in said resized tile.

17. The method of claim 13 wherein said evaluating further comprises:

determining whether a resized tile is suitable for partial compression; and

designating said resized tile for partial compression if it is deemed to be suitable.

18. The method of claim 17 wherein said determining further comprises:

determining whether said resized tile is covered by less than two triangle primitives.

19. The method of claim 17 wherein said compressing further comprises:

assigning an order to triangle primitives covering said resized tile;

determining the color type of each sample of said resized tile;

creating a compressed format of color entries out of said pixel data; and

creating a pointer to said compressed format.

20. A graphics processing apparatus comprising:

a retrieving unit configured to retrieve tiles of pixel data from a main memory;

a resizing unit configured to resize said tiles;

a compression unit configured to compress said resized tiles if said resized tiles are deemed suitable for said compression, wherein said compressed tiles are stored in cache memory after compression;

a processing unit configured to process said compressed tiles in a graphics processor; and

a first TFT comprising a compression format entry for each of said tiles in said main memory, wherein said entry comprises an encoding for a plurality of states for each of said resized tiles, said states comprising;

a full compression state;

a partial compression state;

an uncompressed state; and

a clear state.

21. The graphics processing apparatus of claim 20 wherein said graphics processor comprises:

a second TFT, wherein said second TFT comprises a compression format entry for each of said compressed tiles in said cache memory.

22. The graphics processing apparatus of claim 21 wherein said entry in said second TFT comprises an encoding for a plurality of states for each of said resized tiles, said states comprising:

a full compression state;

a partial compression state;

an uncompressed state; and

a clear state.

23. The graphics processing apparatus of claim 22 wherein said size of tiles in said main memory and size of compressed tiles in said cache memory differ.

24. The graphics processing apparatus of claim 22 wherein tiles in said main memory is of size 4×

4 and compressed tiles in said cache memory is of size 2×

2.

25. The graphics processing apparatus of claim 24 wherein said compression unit is configured to:

combine four 2×

2 tiles in said cache memory into a 4×

4 tile;

choose a lowest common compression level from among said four 2×

2 tiles; and

use said lowest common compression level to compress said 4×

4 tile, wherein said 4×

4 is returned for storage in said main memory.

26. The graphics processing apparatus of claim 22 wherein said graphics processor is configured to clear entries in said cache memory by setting each entry of said second TFT to said clear state and write a clear color storied on said graphics processor in said entries in said cache memory.

27. The graphics processing apparatus of claim 20 wherein said compressed tiles are not decompressed prior to being processed by said graphics processor.

28. The graphics processing apparatus of claim 20 wherein said evaluation unit is further configured to determine whether a resized tile is suitable for full compression and designate said resized tile for full compression if it is deemed to be suitable.

29. The graphics processing apparatus of claim 28 wherein said evaluation unit is further configured to determine whether said resized tile is wholly covered by a triangle primitive.

30. The graphics processing apparatus of claim 29 wherein said compression unit is further configured to store a single color entry for each pixel in said resized tile.

31. The graphics processing apparatus of claim 20 wherein said evaluation unit is further configured to determine whether a resized tile is suitable for partial compression and designate said resized tile for partial compression if it is deemed to be suitable.

32. The graphics processing apparatus of claim 31 wherein said evaluation unit is further configured to determine whether said resized tile is covered by less than two triangle primitives.

33. The graphics processing apparatus of claim 31 wherein said compression unit is further configured to:

assign an order to triangle primitives covering said resized tile;

determine the color type of each sample of said resized tile;

create a compressed format of color entries out of said pixel data; and

create a pointer to said format.

34. The graphics processing apparatus of claim 33 wherein said pointer comprises a bit encoding associated with each sample in said resized tile, wherein each bit represents an index to entries in said compressed format.

35. The graphics processing apparatus of claim 34 wherein said pointer further comprises a short-hand bit encoding scheme for encoding a pixel with no replacement colors.

55. The graphics processing apparatus of claim 20 wherein said graphics processor further comprises:

an evaluation unit configured to evaluate said resized tiles for compression suitability, wherein said compression recognizes duplicate data and reduces amount of duplicate data stored within said resized tiles.

36. A computer program product comprising:

A non-transitory computer readable medium having computer readable program code embodied therein configured for data compression, comprising;

computer readable code configured to cause a computer to retrieve tiles of pixel data from a main memory;

computer readable code configured to cause a computer to resize said tiles;

computer readable code configured to cause a computer to compress said resized tiles, wherein said compressed tiles are stored in cache memory after compression;

computer readable code configured to cause a computer to process said compressed tiles in a graphics processor;

computer readable code configured to cause a computer to return said compressed tiles from said graphics processor to cache memory;

computer readable code configured to cause a computer to resize said compressed tiles in cache memory for storage in said main memory; and

computer readable code configured to cause a computer to return said resized tiles to main memory.

37. The computer program product of claim 36 wherein said graphics processor comprises:

a first Tile Format Table (TFT), wherein said graphics processor uses information from said TFT to process said compressed tiles.

38. The computer program product of claim 37 wherein said first TFT comprises:

a compression format entry for each of said tiles in said main memory.

39. The computer program product of claim 38 wherein said entry comprises an encoding for a plurality of states for each of said resized tiles, said states comprising:

a full compression state;

a partial compression state;

an uncompressed state; and

a clear state.

40. The computer program product of claim 36 wherein said graphics processor comprises:

a second TFT, wherein said second TFT comprises a compression format entry for each of said compressed tiles in said cache memory.

41. The computer program product of claim 40 wherein said entry in said second TFT comprises an encoding for a plurality of states for each of said resized tiles, said states comprising:

a full compression state;

a partial compression state;

an uncompressed state; and

a clear state.

42. The computer program product of claim 41 wherein said size of tiles in said main memory and size of compressed tiles in said cache memory differ.

43. The computer program product of claim 41 wherein tiles in said main memory is of size 4×

4 and compressed tiles in said cache memory is of size 2×

2.

44. The computer program product of claim 43 wherein said computer readable code configured to cause a computer to resize said compressed tiles in cache memory for storage in said main memory further comprises:

computer readable code configured to cause a computer to combine four 2×

2 tiles in said cache memory into a 4×

4 tile;

computer readable code configured to cause a computer to choose a lowest common compression level from among said four 2×

2 tiles; and

computer readable code configured to cause a computer to use said lowest common compression level to compress said 4×

4 tile, wherein said 4×

4 is returned for storage in said main memory.

45. The computer program product of claim 41 wherein said computer readable code configured to cause a computer to process further comprises:

computer readable code configured to cause a computer to clear entries in said cache memory by setting each entry of said second TFT to said clear state; and

computer readable code configured to cause a computer to write a clear color storied on said graphics processor in said entries in said cache memory.

46. The computer program product of claim 36 wherein said computer readable code configured to cause a computer to process does not decompress said compressed tiles prior to processing said compressed tiles in a graphics processor.

47. The computer program product of claim 36 further comprising:

computer readable code configured to cause a computer to access color information of samples in each of said pixel in said resized tiles, wherein duplicate color information is accessed once; and

computer readable code configured to cause a computer to combine color information samples to obtain one color for said pixel.

48. The computer program product of claim 36 wherein said computer readable code configured to cause a computer to compress further comprise:

computer readable code configured to cause a computer to evaluate said resized tiles for compression suitability, wherein said compression recognizes duplicate data and reduces amount of duplicate data stored within said resized tiles; and

computer readable code configured to cause a computer to compress said resized tiles if said resized tiles are deemed suitable for said compression.

49. The computer program product of claim 48 wherein said computer readable code configured to cause a computer to evaluate further comprises:

computer readable code configured to cause a computer to determine whether a resized tile is suitable for full compression; and

computer readable code configured to cause a computer to designate said resized tile for full compression if it is deemed to be suitable.

50. The computer program product of claim 49 wherein said computer readable code configured to cause a computer to determine further comprises:

computer readable code configured to cause a computer to determine whether said resized tile is wholly covered by a triangle primitive.

51. The computer program product of claim 49 wherein said computer readable code configured to cause a computer to compress further comprises:

computer readable code configured to cause a computer to store a single color entry for each pixel in said resized tile.

52. The computer program product of claim 48 wherein said computer readable code configured to cause a computer to evaluate further comprises:

computer readable code configured to cause a computer to determine whether a resized tile is suitable for partial compression; and

computer readable code configured to cause a computer to designate said resized tile for partial compression if it is deemed to be suitable.

53. The computer program product of claim 52 wherein said computer readable code configured to cause a computer to determine further comprises:

computer readable code configured to cause a computer to determine whether said resized tile is covered by less than two triangle primitives.

54. The computer program product of claim 52 wherein said computer readable code configured to cause a computer to compress further comprises:

computer readable code configured to cause a computer to assign an order to triangle primitives covering said resized tile;

computer readable code configured to cause a computer to determine the color type of each sample of said resized tile;

computer readable code configured to cause a computer to create a compressed format of color entries out of said pixel data; and

computer readable code configured to cause a computer to create a pointer to said compressed format.

56. A graphics processing apparatus comprising:

a graphics processor operative to;

retrieve tiles of pixel data from a main memory;

resize said tiles;

compress said resized tiles if said resized tiles are deemed suitable for said compression, wherein said compressed tiles are stored in cache memory after compression;

process said compressed tiles in a graphics processor;

return said compressed tiles from said graphics processor to cache memory;

resize said compressed tiles in cache memory for storage in said main memory; and

return said resized tiles to main memory.

57. A method for compressing image data, comprising:

retrieving tiles of pixel data from a main memory;

resizing said tiles;

compressing said resized tiles, wherein said compressed tiles are stored in cache memory after compression;

processing said compressed tiles in a graphics processor;

accessing color information of samples in each of said pixel in said resized tiles, wherein duplicate color information is accessed once; and

combining color information samples to obtain one color for said pixel.