Skip to content

[Backport branch/3.2.x] Align local vector storage arrays in vec transform #7163

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
bernhardmgruber merged 1 commit into from
Jan 12, 2026
Merged

[Backport branch/3.2.x] Align local vector storage arrays in vec transform #7163

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
8 changes: 6 additions & 2 deletions cub/cub/device/dispatch/kernels/kernel_transform.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -252,7 +252,9 @@ _CCCL_DEVICE void transform_kernel_vectorized(

// if we can vectorize, we convert f's return type to the output type right away, so we can reinterpret later
using THRUST_NS_QUALIFIER::cuda_cub::core::detail::uninitialized_array;
uninitialized_array<::cuda::std::conditional_t<can_vectorize_store, output_t, result_t>, items_per_thread> output;
using output_array_t = ::cuda::std::conditional_t<can_vectorize_store, output_t, result_t>;
constexpr auto output_array_alignment = can_vectorize_store ? sizeof(output_t) * vec_size : alignof(result_t);
uninitialized_array<output_array_t, items_per_thread, output_array_alignment> output;

auto provide_array = [&](auto... inputs) {
// load inputs
Expand Down Expand Up @@ -299,7 +301,9 @@ _CCCL_DEVICE void transform_kernel_vectorized(
output[i] = f(inputs[i]...);
}
};
provide_array(uninitialized_array<it_value_t<RandomAccessIteratorsIn>, items_per_thread>{}...);
provide_array(uninitialized_array<it_value_t<RandomAccessIteratorsIn>,
items_per_thread,
sizeof(it_value_t<RandomAccessIteratorsIn>) * vec_size>{}...);

// write output
if constexpr (can_vectorize_store)
Expand Down
4 changes: 2 additions & 2 deletions thrust/thrust/system/cuda/detail/core/util.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -510,12 +510,12 @@ struct uninitialized
// uninitialized_array
// --------------
// allocates uninitialized data on stack
template <class T, size_t N>
template <class T, size_t N, size_t Alignment = alignof(T)>
struct uninitialized_array
{
using value_type = T;
static constexpr ::cuda::std::integral_constant<size_t, N> size{};
alignas(T) char data_[N * sizeof(T)];
alignas(Alignment) char data_[N * sizeof(T)];

_CCCL_HOST_DEVICE T* data()
{
Expand Down