Optimized alpha blend for anti-aliased render
AssigneeNVIDIA Corporation
InventorsGary C. King, Luke Y. Chang, Steven E. Molnar, David K. McAllister
Life of this patent
Estimating the term…
Abstract
Circuits, methods, and apparatus that reduce the amount of data transferred between a graphics processor integrated circuit and graphics memory. Various embodiments of the present invention further improve the efficiency of blenders that are included on a graphics processor. One embodiment provides for the storage of a reduced number of subsamples of a pixel when the storage of a larger number of subsamples would be redundant. The number of subsamples that are blended with source data are compressed, thereby reducing the task load on the blenders increasing their efficiency. These methods can be disabled to avoid errors that may arise in certain applications.
Claims24
A method of generating a pixel for display comprising:
- receiving a first source data primitive, the first source data primitive covering at least one pixel including a first pixel;
- reading a first number of subsamples of the first pixel from a first number of memory locations;
- reducing the first number of subsamples of the first pixel to a second number of subsamples, the second number of subsamples less than the first number of subsamples;
- blending the second number of subsamples of the first pixel with a second number of subsamples of the received first source data primitive to generate a second number of updated subsamples;
- and
- storing the second number of updated subsamples in a second number of memory locations.
A method of generating pixels for display comprising:
- receiving a first source data primitive, the first source data primitive completely covering at least one pixel including a first pixel;
- storing a first number of subsamples for the first pixel;
- receiving a second source data primitive, the second source data primitive partially covering one or more pixels including a second pixel;
- and
- storing a second number of subsamples for the second pixel, wherein the first number is less than the second number.
A method of generating a pixel for display comprising:
- if compression is enabled:
- receiving a source data primitive, the source data primitive covering at least one pixel including a first pixel;
- reading a first number of subsamples of the first pixel from a first number of memory locations;
- reducing the first number of subsamples of the first pixel to a second number of subsamples, the second number of subsamples less than the first number of subsamples;
- blending the second number of subsamples of the first pixel with a second number of subsamples of the received source data primitive to generate a second number of updated subsamples;
- and
- storing the second number of updated subsamples in a second number of memory locations;
- and
- if compression is disabled:
- receiving the source data primitive, the source data primitive covering at least one pixel including the first pixel;
- reading the first number of subsamples of the first pixel from the first number of memory locations;
- blending the first number of subsamples of the first pixel with a first number of subsamples of the received source data primitive to generate a first number of updated subsamples;
- and
- storing the first number of updated subsamples in a first number of memory locations,
- wherein the first number of subsamples of the received source data primitive is greater than the second number of subsamples of the received source data primitive, and the first number of updated subsamples is greater than the second number of updated subsamples.
Dependencies are inferred from the claim text.