Register spill/fill using shared local memory space
A mechanism is described for facilitating using of a shared local memory for register spilling/filling relating to graphics processors at computing devices. A method of embodiments, as described herein, includes reserving one or more spaces of a shared local memory (SLM) to perform one or more of spilling and filling relating to registers associated with a graphics processor of a computing device.
1. An apparatus comprising:
one or more processors including a graphics processor, the graphics processor including a plurality of hardware threads;
a plurality of registers associated with the graphics processor; and
a shared local memory (SLM), wherein the one or more processors are to:
estimate an amount of space required for one or more of spilling or filling for the plurality of registers in execution of an application,
allocate a first portion of the SLM to perform the one or more of spilling and filling relating to the plurality of registers associated with the graphics processor, wherein the first portion of the SLM includes a plurality of spaces for the plurality of hardware threads, and allocate a second portion of the SLM to load and hold one or more constants, the second portion of the SLM being a global space for access by the plurality of hardware threads, and
utilize the first portion of the SLM as a register buffer for the one or more of spilling and filling relating to the plurality of registers store one or more constants in the second portion of the SLM during execution of the application.
2. The apparatus of claim 1 , wherein the one or more processors are to detect the plurality of registers and generate a code to specify the plurality of spaces in the first portion of the SLM.
3. The apparatus of claim 2 , wherein the one or more processors are further to communicate the code specifying the plurality of spaces in the first portion of the SLM to a graphics driver associated with the graphics processor and to facilitate the graphics driver to communicate information relating to the plurality of spaces in the first portion of the SLM to the graphics processor.
4. The apparatus of claim 1 , wherein the first portion of the SLM is to be shared between three-dimensional (3D) shader threads and compute threads when both types of threads are run.
5. The apparatus of claim 1 , wherein the one or more processors are to continuously or periodically load constants in the second portion of the SLM during execution of the application.
6. The apparatus of claim 1 , wherein the graphics processor is co-located with an application processor on a common semiconductor package.
7. A method comprising:
estimating an amount of space required for one or more of spilling or filling for a plurality of registers associated with a graphics processor of a computing device in execution of an application, the graphics processor including a plurality of hardware threads;
allocating a first portion of a shared local memory (SLM) to perform the one or more of spilling and filling relating to the plurality of registers associated with the graphics processor, wherein the first portion of the SLM includes a plurality of spaces, and allocating a second portion of the SLM to load and hold one or more constants, the second portion of the SLM being a global space for access by the plurality of hardware threads; and
utilizing the first portion of the SLM as a register spill/fill buffer for the one or more of spilling and filling relating to the plurality of registers storing one or more constants in the second portion of the SLM during execution of the application.
8. The method of claim 7 , further comprising:
detecting the plurality of registers; and
generating a code to specify the plurality of spaces in the first portion of the SLM.
9. The method of claim 8 , further comprising:
communicating the code specifying the plurality of spaces in the first portion of the SLM to a graphics driver associated with the graphics processor, wherein the graphics driver is to communicate information relating to the plurality of spaces in the first portion of the SLM to the graphics processor.
10. The method of claim 7 , wherein the first portion of the SLM is to be shared between three-dimensional (3D) shader threads and compute threads when both types of threads are run.
11. The method of claim 7 , wherein the one or more processors are to continuously or periodically load constants in the second portion of the SLM during execution of the application.
12. The method of claim 7 , wherein the graphics processor is co-located with an application processor on a common semiconductor package.
13. At least one non-transitory machine-readable medium comprising instructions that when executed by a computing device, cause the computing device to perform operations comprising:
estimating an amount of space required for one or more of spilling or filling for a plurality of registers associated with a graphics processor of the computing device in execution of an application, the graphics processor including a plurality of hardware threads;
allocating a first portion of a shared local memory (SLM) to perform the one or more of spilling and filling relating to the plurality of registers associated with the graphics processor, wherein the first portion of the SLM includes a plurality of spaces, and allocating a second portion of the SLM to load and hold one or more constants, the second portion of the SLM being a global space for access by the plurality of hardware threads; and
utilizing the first portion of the SLM as a register spill/fill buffer for the one or more of spilling and filling relating to the plurality of registers and storing one or more constants in the second portion of the SLM during execution of the application.
14. The machine-readable medium of claim 13 , wherein the operations further comprise:
detecting the plurality of registers; and
generating a code to specify the plurality of spaces in the first portion of the SLM.
15. The machine-readable medium of claim 14 , wherein the operations further comprise:
communicating the code specifying the plurality of spaces in the first portion of the SLM to a graphics driver associated with the graphics processor, wherein the graphics driver is to communicate information relating to the plurality of spaces in the first portion of the SLM to the graphics processor, wherein the graphics processor is co-located with an application processor on a common semiconductor package.
16. The machine-readable medium of claim 13 , wherein the first portion of the SLM is to be shared between three-dimensional (3D) shader threads and compute threads when both types of threads are run.
17. The machine-readable medium of claim 13 , wherein the one or more processors are to continuously or periodically load constants in the second portion of the SLM during execution of the application.