STOCHASTIC RASTERIZATION USING ENHANCED STENCIL OPERATIONS ON A GRAPHICS PROCESSING UNIT (GPU)

• US 20150294498A1
• Filed: 04/14/2015
• Published: 10/15/2015
• Est. Priority Date: 04/15/2014
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Mei, Chunhui, Wang, Tao, Yeo, Young In, Goel, Vineet
• Granted Patent: US 9,626,762 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06T 1/20 Processor architectures; Processor configuration, e.g. pipelining
• CPC Class Codes: G06T 1/60 Memory management
• CPC Class Codes: G06T 13/80 2D [Two Dimensional] animation, e.g. using sprites
• CPC Class Codes: G06T 15/005 General purpose rendering architectures
• CPC Class Codes: G06T 15/80 Shading
• CPC Class Codes: G06T 2210/12 Bounding box
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80

Abstract

Techniques are described for stochastic rasterization. A graphics processing unit (GPU) may discard samples of bounding polygons that together indicate movement of one or more primitives before a pixel shader process the samples. The GPU may leverage a stencil buffer and stencil test for discarding of such samples.

Claims

1. A method of graphics processing, the method comprising:

determining a plurality of bounding polygons that together cover an animation timeline of one or more primitives from a first frame to a second frame, wherein each bounding polygon covers a portion of the animation timeline, and wherein the animation timeline corresponds to a movement of the one or more primitives;

determining temporal values for samples within each of the bounding polygons based on a sampling pattern;

prior to samples of respective bounding polygons being processed by a pixel shader of a graphics pipeline of a graphics processing unit (GPU), determining which samples within respective bounding polygons are to be further processed based on temporal values for the samples within respective bounding polygons and respective different threshold conditions for each of the respective bounding polygons;

discarding from further processing a first set of the samples within a bounding polygon of the plurality of bounding polygons and maintaining a second set of the samples within the bounding polygon of the plurality of bounding polygons based on the determination, from a threshold condition for that bounding polygon, of which samples within the bounding polygon are to be further processed; and

executing one or more instances of the pixel shader to process the second set of samples of the bounding polygon.

2. The method of claim 1,

wherein determining which samples within respective bounding polygons are to be further processed comprises:

prior to samples of a first bounding polygon of the plurality of bounding polygons being processed by the pixel shader stage, determining which samples within the first bounding polygon are to be further processed based on temporal values for the samples within the first bounding polygon and a first threshold condition for the first bounding polygon; and

prior to samples of a second bounding polygon of the plurality of bounding polygons being processed by the pixel shader stage, determining which samples within the second bounding polygon are to be further processed based on temporal values for the samples within the second bounding polygon and a second threshold condition for the second bounding polygon, and wherein discarding from further processing comprises;

discarding from further processing a first set of the samples within the first bounding polygon of the plurality of bounding polygons and maintaining a second set of the samples within the first bounding polygon of the plurality of bounding polygons based on the determination, from the first threshold condition for the first bounding polygon, of which samples within the first bounding polygon are to be further processed; and

discarding from further processing a first set of the samples within the second bounding polygon of the plurality of bounding polygons and maintaining a second set of the samples within the second bounding polygon of the plurality of bounding polygons based on the determination, from the second threshold condition for the second bounding polygon, of which samples within the second bounding polygon are to be further processed.

3. The method of claim 1, wherein determining the plurality of bounding polygons comprises:

determining a single bounding polygon that covers the entire animation timeline; and

segmenting the single bounding polygon to generate the plurality of bounding polygons.

4. The method of claim 1,

wherein determining temporal values for samples within the bounding polygon based on the sampling pattern comprises:

loading a stochastic sampling pattern; and

determining values of a stencil buffer based on the stochastic sampling pattern,

wherein a temporal value of a sample equals a value in a storage location of the stencil buffer that corresponds to a location of the sample in the bounding polygon, and

wherein determining which samples within the respective bounding polygons are to be further processed comprises;

performing a stencil test, prior to executing one or more instances of the pixel shader, to compare values in the stencil buffer that correspond to the bounding polygon of the plurality of bounding polygons to the threshold condition for that bounding polygon; and

determining which samples within the bounding polygon are to be further processed based on the stencil test.

5. The method of claim 4, further comprising:

performing a depth test separate from performing the stencil test after the execution of the one or more instances of the pixel shader.

6. The method of claim 4, wherein performing the stencil test to compare values in the stencil buffer that correspond to the bounding polygon of the plurality of bounding polygons to the threshold condition for that bounding polygon comprises determining whether values in the stencil buffer that correspond to the bounding polygon are greater than a first threshold value and less than a second threshold value.

7. The method of claim 1, wherein executing the one or more instances of the pixel shader comprises:

determining which samples of the second set of samples are to be shaded based on a location of the second set of samples within the bounding polygon;

determining relative positions within a time instance of the one or more primitives that cover the second set of samples; and

shading one or more samples of the second set of samples based on the determination of the relative positions.

8. The method of claim 1, further comprising:

timeline segmenting the animation timeline into a plurality of timeline segments, wherein the plurality of timeline segments form the animation timeline,

wherein determining the plurality of bounding polygons comprises determining the plurality of bounding polygons based on the plurality of timeline segments.

9. The method of claim 1, wherein the one or more primitives includes a cluster of a plurality of primitives.

10. The method of claim 1, wherein executing one or more instances of the pixel shader to process the second set of samples of the bounding polygon comprises executing one or more instances of the pixel shader to add motion blur to the second frame as part of stochastic rasterization.

11. A device for graphics processing, the device comprising:

a memory unit configured to store a sampling pattern; and

a graphics processing unit (GPU) configured to;

determine a plurality of bounding polygons that together cover an animation timeline of one or more primitives from a first frame to a second frame, wherein each bounding polygon covers a portion of the animation timeline, and wherein the animation timeline corresponds to a movement of the one or more primitives;

determine temporal values for samples within each of the bounding polygons based on the stored sampling pattern;

prior to samples of respective bounding polygons being processed by a pixel shader of a graphics pipeline of GPU, determine which samples within respective bounding polygons are to be further processed based on temporal values for the samples within respective bounding polygons and respective different threshold conditions for each of the respective bounding polygons;

discard from further processing a first set of the samples within a bounding polygon of the plurality of bounding polygons and maintain a second set of the samples within the bounding polygon of the plurality of bounding polygons based on the determination, from a threshold condition for that bounding polygon, of which samples within the bounding polygon are to be further processed; and

execute one or more instances of the pixel shader to process the second set of samples of the bounding polygon.

12. The device of claim 11,

wherein to determine which samples within respective bounding polygons are to be further processed, the GPU is configured to:

prior to samples of a first bounding polygon of the plurality of bounding polygons being processed by the pixel shader stage, determine which samples within the first bounding polygon are to be further processed based on temporal values for the samples within the first bounding polygon and a first threshold condition for the first bounding polygon; and

prior to samples of a second bounding polygon of the plurality of bounding polygons being processed by the pixel shader stage, determine which samples within the second bounding polygon are to be further processed based on temporal values for the samples within the second bounding polygon and a second threshold condition for the second bounding polygon, and wherein to discard from further processing, the GPU is configured to;

discard from further processing a first set of the samples within the first bounding polygon of the plurality of bounding polygons and maintain a second set of the samples within the first bounding polygon of the plurality of bounding polygons based on the determination, from the first threshold condition for the first bounding polygon, of which samples within the first bounding polygon are to be further processed; and

discard from further processing a first set of the samples within the second bounding polygon of the plurality of bounding polygons and maintain a second set of the samples within the second bounding polygon of the plurality of bounding polygons based on the determination, from the second threshold condition for the second bounding polygon, of which samples within the second bounding polygon are to be further processed.

13. The device of claim 11, wherein to determine the plurality of bounding polygons, the GPU is configured to:

determine a single bounding polygon that covers the entire animation timeline; and

segment the single bounding polygon to generate the plurality of bounding polygons.

14. The device of claim 11,

wherein the sampling pattern comprises a stochastic sampling pattern,

wherein the memory unit comprises a stencil buffer,

wherein to determine temporal values for samples within the bounding polygon based on the sampling pattern, the GPU is configured to:

load the stochastic sampling pattern into the stencil buffer; and

determine values of the stencil buffer based on the stochastic sampling pattern,

wherein a temporal value of a sample equals a value in a storage location of the stencil buffer that corresponds to a location of the sample in the bounding polygon, and

wherein to determine which samples within the respective bounding polygons are to be further processed, the GPU is configured to;

perform a stencil test, prior to executing one or more instances of the pixel shader, to compare values in the stencil buffer that correspond to the bounding polygon of the plurality of bounding polygons to the threshold condition for that bounding polygon; and

determine which samples within the bounding polygon are to be further processed based on the stencil test.

15. The device of claim 14, wherein the GPU is configured to:

perform a depth test separate from performing the stencil test after the execution of the one or more instances of the pixel shader.

16. The device of claim 14, wherein to perform the stencil test to compare values in the stencil buffer that correspond to the bounding polygon of the plurality of bounding polygons to the threshold condition for that bounding polygon, the GPU is configured to determine whether values in the stencil buffer that correspond to the bounding polygon are greater than a first threshold value and less than a second threshold value.

17. The device of claim 11, wherein to execute the one or more instances of the pixel shader, the GPU is configured to:

determine which samples of the second set of samples are to be shaded based on a location of the second set of samples within the bounding polygon;

determine relative positions within a time instance of the one or more primitives that cover the second set of samples; and

shade one or more samples of the second set of samples based on the determination of the relative positions.

18. The device of claim 11, wherein the GPU is configured to:

timeline segment the animation timeline into a plurality of timeline segments, wherein the plurality of timeline segments form the animation timeline,

wherein to determine the plurality of bounding polygons, the GPU is configured to determine the plurality of bounding polygons based on the plurality of timeline segments.

19. The device of claim 11, wherein the one or more primitives includes a cluster of a plurality of primitives.

20. The device of claim 11, wherein to execute one or more instances of the pixel shader to process the second set of samples of the bounding polygon, the GPU is configured to execute one or more instances of the pixel shader to add motion blur to the second frame as part of stochastic rasterization.

21. A computer-readable storage medium having instructions stored thereon that when executed cause one or more processors of a device for graphics processing to:

determine a plurality of bounding polygons that together cover an animation timeline of one or more primitives from a first frame to a second frame, wherein each bounding polygon covers a portion of the animation timeline, and wherein the animation timeline corresponds to a movement of the one or more primitives;

determine temporal values for samples within each of the bounding polygons based on a sampling pattern;

prior to samples of respective bounding polygons being processed by a pixel shader of a graphics pipeline of a graphics processing unit (GPU), determine which samples within respective bounding polygons are to be further processed based on temporal values for the samples within respective bounding polygons and respective different threshold conditions for each of the respective bounding polygons;

discard from further processing a first set of the samples within a bounding polygon of the plurality of bounding polygons and maintain a second set of the samples within the bounding polygon of the plurality of bounding polygons based on the determination, from a threshold condition for that bounding polygon, of which samples within the bounding polygon are to be further processed; and

execute one or more instances of the pixel shader to process the second set of samples of the bounding polygon.

22. A device for graphics processing, the device comprising:

means for determining a plurality of bounding polygons that together cover an animation timeline of one or more primitives from a first frame to a second frame, wherein each bounding polygon covers a portion of the animation timeline, and wherein the animation timeline corresponds to a movement of the one or more primitives;

means for determining temporal values for samples within each of the bounding polygons based on a sampling pattern;

means for determining which samples within respective bounding polygons are to be further processed based on temporal values for the samples within respective bounding polygons and respective different threshold conditions for each of the respective bounding polygons, prior to samples of respective bounding polygons being processed by a pixel shader of a graphics pipeline of a graphics processing unit (GPU);

means for discarding from further processing a first set of the samples within a bounding polygon of the plurality of bounding polygons and means for maintaining a second set of the samples within the bounding polygon of the plurality of bounding polygons based on the determination, from a threshold condition for that bounding polygon, of which samples within the bounding polygon are to be further processed; and

means for executing one or more instances of the pixel shader to process the second set of samples of the bounding polygon.