Skip to content
Snippets Groups Projects
Markus Holzer's avatar
deleted branch CUDA at pycodegen / pystencils-benchmark
Markus Holzer's avatar
accepted merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark
Markus Holzer's avatar
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 879ee872 · removed the unused `cuda_block_size` for the `gpu.generate_benchmark`
Frederik Hennig's avatar
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • d38a9324 · using pytest skip if there is no nvcc or gpu available
Christoph Alt's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

Good point, I think skipping is the the most reasonable in that case

Jan Hönig's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

This is dangerous, as it makes the test pass, although it was not tested completely. I'd use pytest.fail, pytest.skip, or pytest.exit, depending on...

Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 1e542f17 · Skipping compiling and running cuda kernels if cuda or gpu is not
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 82ce1d7d · Fix the missing constants for the gpu main file and added a kernel ...
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 6e88d389 · Using cuda as a base for the docker container to also test the gpu
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 4b1f3f53 · fixed the _add_launch_bounds and also added some small tests
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 4ee400e9 · added the new packages to the setup.cfg and the new templates to the
Christoph Alt's avatar
pushed to branch CUDA at pycodegen / pystencils-benchmark
  • 9140da63 · Added a parameter to insert a launch bounds to the kernel
Markus Holzer's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

Alright, fine for me 😄

Christoph Alt's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

I think it would be a bit cleaner to do that in another MR

Markus Holzer's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

Would it make sense to also add ROCm support? Mostly this would be just a renaming for example: #include <cuda_runtime.h> --> #include <hip_runtime.h>...

Markus Holzer's avatar
approved merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark
Markus Holzer's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

It means this:...

Christoph Alt's avatar
commented on merge request !1 "Add CUDA support" at pycodegen / pystencils-benchmark

Thank you and thank you for your review 😄 I am not really sure what it is meant there. In principle it is possible to configure the cuda_block_size...