Skip to content

[SYCL][COMPAT] Specialize reverse_bits for nvptx #15805

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
merged 2 commits into from
Oct 31, 2024
Merged

[SYCL][COMPAT] Specialize reverse_bits for nvptx #15805

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
794fab3
Optimize reverse_bits for nvptx.
JackAKirk Oct 22, 2024
94df4f3
Only do opt if sizeof(T) == 4
JackAKirk Oct 24, 2024
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
7 changes: 7 additions & 0 deletions sycl/include/syclcompat/util.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -161,6 +161,13 @@ inline double cast_ints_to_double(int high32, int low32) {
template <typename T> inline T reverse_bits(T a) {
static_assert(std::is_unsigned<T>::value && std::is_integral<T>::value,
"unsigned integer required");
#if defined(__NVPTX__)
Copy link
Contributor

@joeatodd joeatodd Oct 24, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you if constexpr(std::is_same_v<T, unsigned> here? Theoretically this API accepts e.g. unsigned short, unsigned char and I guess the PTX wouldn't work for that.

Of course, the ptx for those cases could be defined too, so that'd be an alternative.

Copy link
Contributor Author

@JackAKirk JackAKirk Oct 24, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for pointing this out. I've dealt with this using sizeof

if constexpr (sizeof(T) == 4) {
unsigned result;
asm volatile("brev.b32 %0, %1;" : "=r"(result) : "r"(a));
return result;
}
#endif // __NVPTX__
if (!a)
return 0;
T mask = 0;
Expand Down
Loading