This is an archive of the discontinued LLVM Phabricator instance.

Differential D44660

[libcxx] unroll the loops in <experimental/simd> for Clang, until LLVM bugs are fixed
Needs ReviewPublic

Authored by timshen on Mar 19 2018, 5:44 PM.

Details

Reviewers
mclow.lists
EricWF

Diff Detail

Revision Contents

PathSize
libcxx/
include/
experimental/
23 lines

Diff 139052

libcxx/include/experimental/simd

Show First 20 Lines • Show All 605 Lines • ▼ Show 20 Lines
#elif defined(_LIBCPP_MICROARCH_NEON) #elif defined(_LIBCPP_MICROARCH_NEON)
#include <arm_neon.h> #include <arm_neon.h>
#endif #endif
#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER) #if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
#pragma GCC system_header #pragma GCC system_header
#endif #endif
#if !defined(_LIBCPP_COMPILER_CLANG)
#define _LIBCPP_UNROLL
#else
// See LLVM PR/36359 for context of this workaround.
#define _LIBCPP_UNROLL _Pragma("unroll")
#endif
_LIBCPP_BEGIN_NAMESPACE_EXPERIMENTAL_SIMD _LIBCPP_BEGIN_NAMESPACE_EXPERIMENTAL_SIMD
enum class _StorageKind { enum class _StorageKind {
_Scalar, _Scalar,
_Array, _Array,
_VecExt, _VecExt,
}; };
▲ Show 20 Lines • Show All 1,482 Lines • ▼ Show 20 Lines
template <class _Tp, class _Abi> template <class _Tp, class _Abi>
_Tp hmax(const simd<_Tp, _Abi>& __v) { _Tp hmax(const simd<_Tp, _Abi>& __v) {
return __hmax(__v); return __hmax(__v);
} }
// algorithms [simd.alg] // algorithms [simd.alg]
template <class _Tp, class _Abi> template <class _Tp, class _Abi>
simd<_Tp, _Abi> min(const simd<_Tp, _Abi>& __a, // Add `inline` keyword until LLVM PR/36495 is fixed
inline simd<_Tp, _Abi> min(const simd<_Tp, _Abi>& __a,
const simd<_Tp, _Abi>& __b) noexcept { const simd<_Tp, _Abi>& __b) noexcept {
simd<_Tp, _Abi> __v; simd<_Tp, _Abi> __v;
for (size_t __i = 0; __i < __v.size(); __i++) { _LIBCPP_UNROLL for (size_t __i = 0; __i < __v.size(); __i++) {
__v[__i] = std::min(__a[__i], __b[__i]); __v[__i] = std::min(__a[__i], __b[__i]);
} }
return __v; return __v;
} }
template <class _Tp, class _Abi> template <class _Tp, class _Abi>
simd<_Tp, _Abi> max(const simd<_Tp, _Abi>& __a, // Add `inline` keyword until LLVM PR/36495 is fixed
inline simd<_Tp, _Abi> max(const simd<_Tp, _Abi>& __a,
const simd<_Tp, _Abi>& __b) noexcept { const simd<_Tp, _Abi>& __b) noexcept {
simd<_Tp, _Abi> __v; simd<_Tp, _Abi> __v;
for (size_t __i = 0; __i < __v.size(); __i++) { _LIBCPP_UNROLL for (size_t __i = 0; __i < __v.size(); __i++) {
__v[__i] = std::max(__a[__i], __b[__i]); __v[__i] = std::max(__a[__i], __b[__i]);
} }
return __v; return __v;
} }
template <class _Tp, class _Abi> template <class _Tp, class _Abi>
std::pair<simd<_Tp, _Abi>, simd<_Tp, _Abi>> std::pair<simd<_Tp, _Abi>, simd<_Tp, _Abi>>
minmax(const simd<_Tp, _Abi>& __a, const simd<_Tp, _Abi>& __b) noexcept { minmax(const simd<_Tp, _Abi>& __a, const simd<_Tp, _Abi>& __b) noexcept {
▲ Show 20 Lines • Show All 921 Lines • ▼ Show 20 Lines return __hmax(__simd_mask_friend::__simd_select(
_SimdType(std::numeric_limits<typename _SimdType::value_type>::lowest()), _SimdType(std::numeric_limits<typename _SimdType::value_type>::lowest()),
__w.__v_, __w.__m_)); __w.__v_, __w.__m_));
} }
// TODO: Implement <cmath> functions for simd. // TODO: Implement <cmath> functions for simd.
_LIBCPP_END_NAMESPACE_EXPERIMENTAL_SIMD _LIBCPP_END_NAMESPACE_EXPERIMENTAL_SIMD
#undef _LIBCPP_UNROLL
#endif /* _LIBCPP_EXPERIMENTAL_SIMD */ #endif /* _LIBCPP_EXPERIMENTAL_SIMD */