Skip to content

Fix floating minmax performance #5016

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
Stephan T. Lavavej (StephanTLavavej) merged 6 commits into from
Oct 21, 2024
Merged

Fix floating minmax performance #5016

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
a36d1e3
Fix floating minmax performance
AlexGuteniev Oct 14, 2024
716b2b7
all the floats
AlexGuteniev Oct 17, 2024
b2b4336
Merge branch 'main' into perf-floats-away
StephanTLavavej Oct 17, 2024
c65ac95
Fix typo.
StephanTLavavej Oct 17, 2024
2d7679e
Add `noexcept`.
StephanTLavavej Oct 17, 2024
135d8ca
Add `_M_ARM64EC` guard.
StephanTLavavej Oct 17, 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
16 changes: 15 additions & 1 deletion stl/src/vector_algorithms.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2222,11 +2222,25 @@ namespace {
}
}

#ifndef _M_ARM64EC
// TRANSITION, DevCom-10767462
template <_Min_max_mode _Mode, class _Traits, bool _Sign>
auto __std_minmax_impl_wrap(const void* const _First, const void* const _Last) noexcept {
auto _Rx = __std_minmax_impl<_Mode, _Traits, _Sign>(_First, _Last);
_mm256_zeroupper();
Comment thread
StephanTLavavej marked this conversation as resolved.
return _Rx;
}
#endif // ^^^ !defined(_M_ARM64EC) ^^^

template <_Min_max_mode _Mode, class _Traits, bool _Sign>
auto __std_minmax_disp(const void* const _First, const void* const _Last) noexcept {
#ifndef _M_ARM64EC
if (_Byte_length(_First, _Last) >= 32 && _Use_avx2()) {
return __std_minmax_impl<_Mode, typename _Traits::_Avx, _Sign>(_First, _Last);
if constexpr (_Traits::_Avx::_Is_floating) {
return __std_minmax_impl_wrap<_Mode, typename _Traits::_Avx, _Sign>(_First, _Last);
} else {
return __std_minmax_impl<_Mode, typename _Traits::_Avx, _Sign>(_First, _Last);
}
}

if (_Byte_length(_First, _Last) >= 16 && _Use_sse42()) {
Expand Down