Skip to content

This is a follow-up pr to make some changes suggested after approval #1937

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
oleksandr-pavlyk merged 1 commit into from
Dec 17, 2024
Merged

This is a follow-up pr to make some changes suggested after approval #1937

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
5 changes: 4 additions & 1 deletion dpctl/tensor/libtensor/include/utils/sycl_utils.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -166,7 +166,10 @@ T custom_reduce_over_group(const GroupT &wg,
const T &local_val,
const OpT &op)
{
// value experimentally tuned to achieve best runtime on Iris Xe,
// Arc A140V integrated Intel GPUs, and discrete Intel Max GPU.
constexpr std::uint32_t low_sz = 8u;
// maximal work-group size
constexpr std::uint32_t high_sz = 1024u;
const std::uint32_t wgs = wg.get_local_linear_range();
const std::uint32_t lid = wg.get_local_linear_id();
Expand All @@ -192,7 +195,7 @@ T custom_reduce_over_group(const GroupT &wg,
#pragma unroll
for (std::uint32_t sz = high_sz; sz >= low_sz; sz >>= 1) {
if (n_witems >= sz) {
n_witems = (n_witems + 1) >> 1;
n_witems >>= 1;
_fold(local_mem_acc, lid, n_witems, op);
sycl::group_barrier(wg, sycl::memory_scope::work_group);
}
Expand Down
Loading