Skip to content

Commit

Permalink
neon/get_low: use __builtin_shufflevector if available
Browse files Browse the repository at this point in the history 
Fixes #779
  • Loading branch information
@ngzhian @nemequ
ngzhian authored and nemequ committed May 18, 2021
1 parent 4003afa commit ea3f75e
Showing 1 changed file with 80 additions and 40 deletions.
120 changes: 80 additions & 40 deletions simde/arm/neon/get_low.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -43,10 +43,14 @@ simde_vget_low_f32(simde_float32x4_t a) {
simde_float32x2_private r_;
simde_float32x4_private a_ = simde_float32x4_to_private(a);

SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif

return simde_float32x2_from_private(r_);
#endif
Expand All @@ -65,10 +69,14 @@ simde_vget_low_f64(simde_float64x2_t a) {
simde_float64x1_private r_;
simde_float64x2_private a_ = simde_float64x2_to_private(a);

SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif

return simde_float64x1_from_private(r_);
#endif
Expand All @@ -90,10 +98,14 @@ simde_vget_low_s8(simde_int8x16_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1, 2, 3, 4, 5, 6, 7);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_int8x8_from_private(r_);
Expand All @@ -116,10 +128,14 @@ simde_vget_low_s16(simde_int16x8_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1, 2, 3);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_int16x4_from_private(r_);
Expand All @@ -142,10 +158,14 @@ simde_vget_low_s32(simde_int32x4_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_int32x2_from_private(r_);
Expand All @@ -168,10 +188,14 @@ simde_vget_low_s64(simde_int64x2_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_int64x1_from_private(r_);
Expand All @@ -194,10 +218,14 @@ simde_vget_low_u8(simde_uint8x16_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1, 2, 3, 4, 5, 6, 7);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_uint8x8_from_private(r_);
Expand All @@ -220,10 +248,14 @@ simde_vget_low_u16(simde_uint16x8_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1, 2, 3);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_uint16x4_from_private(r_);
Expand All @@ -246,10 +278,14 @@ simde_vget_low_u32(simde_uint32x4_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0, 1);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_uint32x2_from_private(r_);
Expand All @@ -272,10 +308,14 @@ simde_vget_low_u64(simde_uint64x2_t a) {
#if defined(SIMDE_X86_SSE2_NATIVE) && defined(SIMDE_X86_MMX_NATIVE)
r_.m64 = _mm_movepi64_pi64(a_.m128i);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#if HEDLEY_HAS_BUILTIN(__builtin_shufflevector)
r_.values = __builtin_shufflevector(a_.values, a_.values, 0);
#else
SIMDE_VECTORIZE
for (size_t i = 0 ; i < (sizeof(r_.values) / sizeof(r_.values[0])) ; i++) {
r_.values[i] = a_.values[i];
}
#endif
#endif

return simde_uint64x1_from_private(r_);
Expand Down

0 comments on commit ea3f75e

Please sign in to comment.