Skip to content
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

Jump to bottom

<bit>: popcount() utilizes cnt instruction on arm64 #2127

Merged
merged 16 commits into from
Sep 11, 2021
Merged

<bit>: popcount() utilizes cnt instruction on arm64 #2127

Changes from 13 commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
80377ff
utilize cnt instruction on arm64
fsb4000 Aug 15, 2021
c3678ca
fix error
fsb4000 Aug 15, 2021
d5727b0
use `neon_cnt` with `_M_ARM64`
fsb4000 Aug 16, 2021
9e1028c
missing inline
fsb4000 Aug 16, 2021
3b9f984
nodiscard must be before inline
fsb4000 Aug 16, 2021
37859da
tests stuck. try again
fsb4000 Aug 16, 2021
090cfea
Disable the optimization for Clang
fsb4000 Aug 16, 2021
906a844
Merge branch 'main' into fix1924
fsb4000 Aug 17, 2021
7130b98
address review comments
fsb4000 Aug 21, 2021
a554c85
don't include inside namespace std
fsb4000 Aug 21, 2021
9c68dad
rename _HAS_CNT_INTRINSICS to _HAS_NEON_INTRINSICS
fsb4000 Aug 26, 2021
032220a
add a reminder
fsb4000 Aug 26, 2021
0d4dd74
try without extra ifdefs
fsb4000 Aug 26, 2021
f9fca84
Merge branch 'main' into fix1924
StephanTLavavej Sep 8, 2021
54a7bf1
Code review feedback.
StephanTLavavej Sep 8, 2021
193ee4c
Fix whitespace in comment.
StephanTLavavej Sep 8, 2021
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
26 changes: 24 additions & 2 deletions stl/inc/limits
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,18 @@
#include <isa_availability.h>
#include <xstddef>

// TRANSITION, GH-2129, move down to _Arm64_popcount
#if defined(_M_ARM64) && !defined(_M_ARM64EC) && !defined(_M_CEE_PURE) && !defined(__CUDACC__) \
&& !defined(__INTEL_COMPILER) && !defined(__clang__) // TRANSITION, LLVM-51488
#define _HAS_NEON_INTRINSICS 1
#else // ^^^ intrinsics available ^^^ / vvv intrinsics unavailable vvv
#define _HAS_NEON_INTRINSICS 0
#endif // ^^^ intrinsics unavailable ^^^

#if _HAS_NEON_INTRINSICS
#include <arm64_neon.h> // TRANSITION, GH-2129
#endif

#pragma pack(push, _CRT_PACKING)
#pragma warning(push, _STL_WARNING_LEVEL)
#pragma warning(disable : _STL_DISABLED_WARNINGS)
Expand Down Expand Up @@ -1116,6 +1128,13 @@ _NODISCARD int _Checked_x86_x64_popcount(const _Ty _Val) noexcept {
}
#endif // _HAS_POPCNT_INTRINSICS

#if _HAS_NEON_INTRINSICS
_NODISCARD inline int _Arm64_popcount(const unsigned long long _Val) noexcept {
const __n64 _Temp = neon_cnt(__uint64ToN64_v(_Val));
return neon_addv8(_Temp).n8_i8[0];
}
#endif // _HAS_NEON_INTRINSICS

template <class _Ty>
constexpr bool _Is_standard_unsigned_integer =
_Is_any_of_v<remove_cv_t<_Ty>, unsigned char, unsigned short, unsigned int, unsigned long, unsigned long long>;
Expand All @@ -1135,18 +1154,21 @@ _NODISCARD constexpr int _Countr_zero(const _Ty _Val) noexcept {

template <class _Ty, enable_if_t<_Is_standard_unsigned_integer<_Ty>, int> _Enabled = 0>
_NODISCARD _CONSTEXPR20 int _Popcount(const _Ty _Val) noexcept {
#if _HAS_POPCNT_INTRINSICS
#if _HAS_CXX20
if (!_STD is_constant_evaluated())
#endif // _HAS_CXX20
{
StephanTLavavej marked this conversation as resolved.
Show resolved Hide resolved
#if _HAS_POPCNT_INTRINSICS
return _Checked_x86_x64_popcount(_Val);
#elif _HAS_NEON_INTRINSICS // ^^^ X86_x64 intrinsics available ^^^ / vvv ARM64 intrinsics available vvv
StephanTLavavej marked this conversation as resolved.
Show resolved Hide resolved
return _Arm64_popcount(_Val);
#endif // ^^^ ARM64 intrinsics available ^^^
}
#endif // _HAS_POPCNT_INTRINSICS
return _Popcount_fallback(_Val);
}

#undef _HAS_POPCNT_INTRINSICS
#undef _HAS_NEON_INTRINSICS

_STD_END
#pragma pop_macro("new")
Expand Down