Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Test and fix failing nightly libcudacxx + CUB jobs #1847

Merged
merged 1 commit into from
Sep 3, 2024
Merged

Test and fix failing nightly libcudacxx + CUB jobs #1847

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 10 additions & 33 deletions ci/matrix.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
workflows:

Check notice on line 1 in ci/matrix.yaml

View workflow job for this annotation

GitHub Actions / Build workflow from matrix 

Merging consumers for duplicate producer '[C++17 GCC13] Build(amd64)' in 'CUB CTK12.5 nvcc GCC'

Check notice on line 1 in ci/matrix.yaml

View workflow job for this annotation

GitHub Actions / Build workflow from matrix 

Original consumers: [C++17 GCC13] HostLaunch(amd64, V100), [C++17 GCC13] GraphCapture(amd64, V100), [C++17 GCC13] DeviceLaunch(amd64, V100), [C++17 GCC13] TestGPU(amd64, V100)

Check notice on line 1 in ci/matrix.yaml

View workflow job for this annotation

GitHub Actions / Build workflow from matrix 

Duplicate consumers: [C++17 GCC13] SmallGMem(amd64, V100)

Check notice on line 1 in ci/matrix.yaml

View workflow job for this annotation

GitHub Actions / Build workflow from matrix 

Merged consumers: [C++17 GCC13] HostLaunch(amd64, V100), [C++17 GCC13] GraphCapture(amd64, V100), [C++17 GCC13] DeviceLaunch(amd64, V100), [C++17 GCC13] TestGPU(amd64, V100), [C++17 GCC13] SmallGMem(amd64, V100)
# If any jobs appear here, they will be executed instead of `pull_request' for PRs.
# This is useful for limiting resource usage when a full matrix is not needed.
# The branch protection checks will fail when using this override workflow.
Expand Down Expand Up @@ -48,44 +48,21 @@
- {jobs: ['limited'], project: 'cub', std: 17}

nightly:
# libcudacxx build fails, CUB tests fail:
- {jobs: ['build'], ctk: '11.1', gpu: 'v100', sm: 'gpu', cxx: 'gcc6', std: [11], project: ['cub']}
- {jobs: ['test'], ctk: '11.1', gpu: 'v100', sm: 'gpu', cxx: 'gcc6', std: [11], project: ['thrust']}
# - {jobs: ['test'], ctk: '11.1', gpu: 'v100', sm: 'gpu', cxx: 'gcc6', std: [11] }

# libcudacxx build fails, CUB tests fail:
- {jobs: ['build'], ctk: '11.1', gpu: 't4', sm: 'gpu', cxx: 'clang9', std: [17], project: ['cub']}
- {jobs: ['test'], ctk: '11.1', gpu: 't4', sm: 'gpu', cxx: 'clang9', std: [17], project: ['thrust']}
# - {jobs: ['test'], ctk: '11.1', gpu: 't4', sm: 'gpu', cxx: 'clang9', std: [17] }

# CUB + libcudacxx tests fails:
- {jobs: ['build'], ctk: '11.8', gpu: 'rtx2080', sm: 'gpu', cxx: 'gcc11', std: [17], project: ['libcudacxx', 'cub']}
- {jobs: ['test'], ctk: '11.8', gpu: 'rtx2080', sm: 'gpu', cxx: 'gcc11', std: [17], project: ['thrust']}
# - {jobs: ['test'], ctk: '11.8', gpu: 'rtx2080', sm: 'gpu', cxx: 'gcc11', std: [17] }

# libcudacxx tests fail:
- {jobs: ['build'], ctk: 'curr', gpu: 'rtxa6000', sm: 'gpu', cxx: 'gcc7', std: [14], project: ['libcudacxx']}
- {jobs: ['build'], ctk: 'curr', gpu: 'l4', sm: 'gpu', cxx: 'gcc12', std: 'all', project: ['libcudacxx']}
- {jobs: ['build'], ctk: 'curr', gpu: 'rtx4090', sm: 'gpu', cxx: 'clang9', std: [11], project: ['libcudacxx']}
# H100 runners are currently flakey, only build since those use CPU-only runners:
- {jobs: ['build'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'gcc12', std: [11, 20]}
- {jobs: ['build'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'clang16', std: [17]}

- {jobs: ['test'], ctk: 'curr', gpu: 'rtxa6000', sm: 'gpu', cxx: 'gcc7', std: [14], project: ['cub', 'thrust']}
- {jobs: ['test'], ctk: 'curr', gpu: 'l4', sm: 'gpu', cxx: 'gcc12', std: 'all', project: ['cub', 'thrust']}
- {jobs: ['test'], ctk: 'curr', gpu: 'rtx4090', sm: 'gpu', cxx: 'clang9', std: [11], project: ['cub', 'thrust']}
# - {jobs: ['test'], ctk: 'curr', gpu: 'rtxa6000', sm: 'gpu', cxx: 'gcc7', std: [14] }
# - {jobs: ['test'], ctk: 'curr', gpu: 'l4', sm: 'gpu', cxx: 'gcc12', std: 'all' }
# - {jobs: ['test'], ctk: 'curr', gpu: 'rtx4090', sm: 'gpu', cxx: 'clang9', std: [11] }
# - {jobs: ['test'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'gcc12', std: [11, 20] }
# - {jobs: ['test'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'clang16', std: [17] }
- {jobs: ['test'], ctk: '11.1', gpu: 'v100', sm: 'gpu', cxx: 'gcc6', std: [11]}
- {jobs: ['test'], ctk: '11.1', gpu: 't4', sm: 'gpu', cxx: 'clang9', std: [17]}
- {jobs: ['test'], ctk: '11.8', gpu: 'rtx2080', sm: 'gpu', cxx: 'gcc11', std: [17]}
- {jobs: ['test'], ctk: 'curr', gpu: 'rtxa6000', sm: 'gpu', cxx: 'gcc7', std: [14]}
- {jobs: ['test'], ctk: 'curr', gpu: 'l4', sm: 'gpu', cxx: 'gcc12', std: 'all'}
- {jobs: ['test'], ctk: 'curr', gpu: 'rtx4090', sm: 'gpu', cxx: 'clang9', std: [11]}
# H100 runners are currently flakey, only build since those use CPU-only runners:
- {jobs: ['build'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'gcc12', std: [11, 20]}
- {jobs: ['build'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'clang16', std: [17]}

# nvrtc:
- {jobs: ['nvrtc'], ctk: 'curr', gpu: 't4', sm: 'gpu', cxx: 'gcc12', std: [20], project: ['libcudacxx']}
- {jobs: ['nvrtc'], ctk: 'curr', gpu: 'rtxa6000', sm: 'gpu', cxx: 'gcc12', std: [20], project: ['libcudacxx']}
- {jobs: ['nvrtc'], ctk: 'curr', gpu: 'l4', sm: 'gpu', cxx: 'gcc12', std: 'all', project: ['libcudacxx']}
# Fails on h100:
# - {jobs: ['nvrtc'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'gcc12', std: [11, 20], project: ['libcudacxx']}
- {jobs: ['nvrtc'], ctk: 'curr', gpu: 'h100', sm: 'gpu', cxx: 'gcc12', std: [11, 20], project: ['libcudacxx']}

# Any generated jobs that match the entries in `exclude` will be removed from the final matrix for all workflows.
exclude:
Expand Down
5 changes: 3 additions & 2 deletions cub/test/catch2_test_nvrtc.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -226,10 +226,11 @@ TEST_CASE("Test nvrtc", "[test][nvrtc]")
int ptx_version{};
cub::PtxVersion(ptx_version);
const std::string arch = std::string("-arch=sm_") + std::to_string(ptx_version / 10);
const std::string std = std::string("-std=c++") + std::to_string(_CCCL_STD_VER - 2000);

constexpr int num_includes = 5;
constexpr int num_includes = 6;
const char* includes[num_includes] = {
NVRTC_CUB_PATH, NVRTC_THRUST_PATH, NVRTC_LIBCUDACXX_PATH, NVRTC_CTK_PATH, arch.c_str()};
NVRTC_CUB_PATH, NVRTC_THRUST_PATH, NVRTC_LIBCUDACXX_PATH, NVRTC_CTK_PATH, arch.c_str(), std.c_str()};

std::size_t log_size{};
nvrtcResult compile_result = nvrtcCompileProgram(prog, num_includes, includes);
Expand Down
Loading