Shader pipelines and hierarchical shader resources
View Patent ↗Shader resources may be specified for input to a shader using a hierarchical data structure which may be referred to as a descriptor set. The descriptor set may be bound to a bind point of the shader and may contain slots with pointers to memory containing shader resources. The shader may reference a particular slot of the descriptor set using an offset, and may change shader resources by referencing a different slot of the descriptor set or by binding or rebinding a new descriptor set. A graphics pipeline may be specified by creating a pipeline object which specifies a shader and a rendering context object, and linking the pipeline object. Part or all of the pipeline may be validated, cross-validated, or optimized during linking.
1. A method for inputting a shader resource to a shader or a pipeline in a computing device, comprising:
creating, in a computer memory, by a processing device, a data structure having at least one slot which contains a pointer to a shader resource stored in a graphics processing unit (GPU) memory;
binding the data structure to a GPU register;
referencing the at least one slot by an offset to input the shader resource to the shader or pipeline;
determining a GPU overhead associated with inputting a different shader resource to the shader or pipeline by referencing a different slot of the data structure;
determining a CPU overhead associated with inputting the different shader resource to the shader or pipeline by binding a different data structure to the GPU register; and
inputting a different shader resource to the shader or pipeline by conditionally either referencing a different slot of the data structure or binding a different data structure to the GPU register, wherein the condition is based on the GPU overhead and the CPU overhead.
2. The method of claim 1 , wherein the data structure includes a plurality of slots, each containing a pointer to a shader resource.
3. The method of claim 2 , wherein the data structure is bound to a single bind point.
4. The method of claim 2 , wherein the different shader resource is specified by referencing a different slot by a different offset.
5. The method of claim 2 , wherein the different shader resource is specified by binding the different data structure and referencing a slot of the different data structure which contains a pointer to the different shader resource.
6. The method of claim 5 , wherein the data structure is unbound from the GPU register and the different data structure is bound to the GPU register.
7. The method of claim 5 , wherein the different data structure is bound to a different bind point from the data structure.
8. The method of claim 1 , wherein the at least one slot contains a pointer to a second data structure having at least one slot containing a pointer to the shader resource.
9. The method of claim 1 , wherein the at least one slot contains a pointer to another slot of the data structure which contains a pointer to the shader resource.
10. The method of claim 1 , further comprising storing the data structure in a memory.
11. The method of claim 10 , further comprising displaying data stored in or pointed to by the data structure on a display.
12. A method for inputting a graphics pipeline resource in a computing device, comprising:
creating, in a computer memory, by a processing device, a data structure having at least one slot which contains a pointer to a pipeline resource;
binding the data structure to a GPU register;
referencing the at least one slot by an offset to input the graphics pipeline resource to the pipeline;
determining a GPU overhead associated with inputting a different shader resource to the shader or pipeline by referencing a different slot of the data structure;
determining a CPU overhead associated with inputting the different shader resource to the shader or pipeline by binding a different data structure to the GPU register; and
inputting a different pipeline resource to the pipeline by conditionally either referencing a different slot of the data structure or binding a different data structure to the GPU register, wherein the condition is based on the GPU overhead and the CPU overhead.
13. The method of claim 12 , wherein the pipeline resource includes any one or more of: a parameter for a fixed function block or a descriptor for a render target.
14. A method for implementing a graphics pipeline in a computing device, comprising:
creating, in a computer memory, a pipeline object and a descriptor set having at least one slot which contains a pointer to a shader resource;
attaching at least one shader object to the pipeline object, which specifies at least one shader;
attaching at least one rendering context object to the pipeline object which specifies at least part of a rendering context; and
binding the descriptor set to a GPU register; and
determining a GPU overhead associated with replacing the shader resource by referencing a different slot of the descriptor set;
determining a CPU overhead associated with replacing the shader resource by binding a different descriptor set to the GPU register;
replacing the shader resource by conditionally either referencing a different slot of the descriptor set or binding a different descriptor set to the GPU register, wherein the condition is based on the GPU overhead and the CPU overhead.
15. The method of claim 14 , further comprising attaching at least one descriptor set which maps the at least one shader to a shader resource.
16. The method of claim 14 , wherein at least one rendering context objects specifies a fixed function state.
17. The method of claim 14 , wherein at least one rendering context objects specifies constant data.
18. The method of claim 14 , further comprising attaching a data flow object to the pipeline object, which specifies data flow among the at least one shader.
19. The method of claim 18 , wherein the data flow object includes a directed acyclic graph.
20. The method of claim 14 , wherein at least one shader object includes at least one shader fragment.
21. The method of claim 14 , further comprising:
linking the pipeline object; and
validating, cross-validating, or optimizing the pipeline during linking.
22. The method of claim 21 , wherein the validating, cross-validating, or optimizing is performed using the specified rendering context.