REGISTER SPILL MANAGEMENT FOR GENERAL PURPOSE REGISTERS (GPRs)

摘要

Techniques are described for copying data only from a subset of memory locations allocated to a set of instructions to free memory locations for higher priority instructions to execute. Data from a dynamic portion of one or more general purpose registers (GPRs) allocated to the set of instructions may be copied and stored to another memory unit while data from a static portion of the one or more GPRs allocated to the set of instructions may not be copied and stored to another memory unit.

主权项

1. A method of executing instructions on a graphics processing unit (GPU), the method comprising:
executing, with one or more processing elements of one or more cores of the GPU, a first set of instructions; determining, with the GPU, that a second set of instructions to be executed on the GPU is given higher priority than the first set of instructions executing on the GPU; copying, with the GPU, data stored in a subset, and not all, of one or more general purpose register (GPRs) that are allocated to the first set of instructions to another memory unit in response to determining that the second set of instructions are given higher priority, the one or more GPRs are accessible by the one or more cores of the GPU on which the first and second sets of instructions execute; and executing, with the one or more processing elements and after the copying, the second set of instructions utilizing the subset of the one or more GPRs.