Branch: refs/heads/main
Revision: 7226c35
Author: stonea
Link: Refactor gpuTransforms.cpp by stonea · Pull Request #19995 · chapel-lang/chapel · GitHub
Log Message:
Merge pull request #19995 from stonea/gputransform_refactor
Refactor gpuTransforms.cpp
This PR includes some refactoring changes to GPUTransforms.cpp (originally proposed in #19802)
Specifically, I've moved various functions into two classes:
GpuizableLoop: Which is concerned with determining if a loop is eligible for GPU code generation and extracting the bounds of the loop
GpuKernel: Which is responsible for generating the kernel function itself and populating it (copying the body of the loop into it)
I've manually verified that our gpu/native tests pass on Osprey when CHPL_COMM=none.
[Reviewed by @e-kayrakli]
Modified Files:
M compiler/optimizations/gpuTransforms.cpp
Compare: https://github.com/chapel-lang/chapel/compare/e38d9e4021ee...7226c355ebf5