Skip to content

CUDA: Fix CUB's argsort when nrows % block_size == 0 CCCL < 3.1 #21181

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

Merged
CISC merged 2 commits into from
Mar 30, 2026
Merged

CUDA: Fix CUB's argsort when nrows % block_size == 0 CCCL < 3.1 #21181

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
28c3c54
CUDA: Fix CUB's argsort when nrows % block_size == 0 CCCL < 3.1
ORippler Mar 30, 2026
d601b1b
Reduce nrows in test case to 256, don't need 768
ORippler Mar 30, 2026
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
6 changes: 4 additions & 2 deletions ggml/src/ggml-cuda/argsort.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,9 +47,11 @@ void argsort_f32_i32_cuda_cub(ggml_cuda_pool & pool,
#ifdef STRIDED_ITERATOR_AVAILABLE
auto offset_iterator = cuda::make_strided_iterator(cuda::make_counting_iterator(0), ncols);
#else
ggml_cuda_pool_alloc<int> offsets_alloc(pool, nrows + 1);
// offset_iterator needs to populate nrows + 1 elements, so we also have to ceildiv nrows + 1 by block_size
const int nrows_offset = nrows + 1;
ggml_cuda_pool_alloc<int> offsets_alloc(pool, nrows_offset);
int * offset_iterator = offsets_alloc.get();
const dim3 offset_grid((nrows + block_size - 1) / block_size);
const dim3 offset_grid((nrows_offset + block_size - 1) / block_size);
init_offsets<<<offset_grid, block_size, 0, stream>>>(offset_iterator, ncols, nrows);
#endif
CUDA_CHECK(cudaMemcpyAsync(temp_keys, x, ncols * nrows * sizeof(float), cudaMemcpyDeviceToDevice, stream));
Expand Down
1 change: 1 addition & 0 deletions tests/test-backend-ops.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8424,6 +8424,7 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {1023, 2, 1, 3}, order));
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {1024, 2, 1, 3}, order));
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {1025, 2, 1, 3}, order));
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {1025, 256, 1, 1}, order)); // test ceildiv in CUDA's CUB's DeviceSegmentedSort
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {2047, 2, 1, 3}, order));
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {2048, 2, 1, 3}, order));
test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {2049, 2, 1, 3}, order));
Expand Down
Loading