aaron/ispc
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

add cast_sext plus shl/ashr function fix

Browse Source
This commit is contained in:
Anton Mitrokhin
2014-12-26 01:12:14 +03:00
parent 300ff7be75
commit 3f607ade14
1 changed files with 22 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
examples/intrinsics/knc.h
View File

@@ -963,8 +963,9 @@ static FORCEINLINE __vec16_i64 __shl(__vec16_i64 a, __vec16_i64 b) {
} }
static FORCEINLINE __vec16_i64 __shl(__vec16_i64 a, unsigned long long b) { static FORCEINLINE __vec16_i64 __shl(__vec16_i64 a, unsigned long long b) {
__vec16_i32 hi = _mm512_or_epi32(_mm512_slli_epi32(a.v_hi, b), __vec16_i32 hi;
_mm512_srli_epi32(a.v_lo, 32-b)); if (b <= 32) hi = _mm512_or_epi32(_mm512_slli_epi32(a.v_hi, b), _mm512_srli_epi32(a.v_lo, 32-b));
else hi = _mm512_slli_epi32(a.v_lo, b - 32);
__vec16_i32 lo = _mm512_slli_epi32(a.v_lo, b); __vec16_i32 lo = _mm512_slli_epi32(a.v_lo, b);
return __vec16_i64(lo, hi); return __vec16_i64(lo, hi);
} }
@@ -1006,10 +1007,9 @@ static FORCEINLINE __vec16_i64 __ashr(__vec16_i64 a, __vec16_i64 b) {
} }
static FORCEINLINE __vec16_i64 __ashr(__vec16_i64 a, unsigned long long b) { static FORCEINLINE __vec16_i64 __ashr(__vec16_i64 a, unsigned long long b) {
__vec16_i32 xfer __vec16_i32 xfer;
= _mm512_slli_epi32(_mm512_and_epi32(a.v_hi, if (b <= 32) xfer = _mm512_slli_epi32(_mm512_and_epi32(a.v_hi, _mm512_set1_epi32((1<<b)-1)), 32-b);
_mm512_set1_epi32((1<<b)-1)), else xfer = _mm512_srai_epi32(a.v_hi, b-32);
32-b);
__vec16_i32 hi = _mm512_srai_epi32(a.v_hi, b); __vec16_i32 hi = _mm512_srai_epi32(a.v_hi, b);
__vec16_i32 lo = _mm512_or_epi32(xfer, _mm512_srli_epi32(a.v_lo, b)); __vec16_i32 lo = _mm512_or_epi32(xfer, _mm512_srli_epi32(a.v_lo, b));
return __vec16_i64(lo, hi); return __vec16_i64(lo, hi);
@@ -1867,11 +1867,27 @@ static FORCEINLINE __vec16_i32 __cast_sext(const __vec16_i32 &, const __vec16_i1
return _mm512_extload_epi32(&val, _MM_UPCONV_EPI32_SINT16, _MM_BROADCAST32_NONE, _MM_HINT_NONE); return _mm512_extload_epi32(&val, _MM_UPCONV_EPI32_SINT16, _MM_BROADCAST32_NONE, _MM_HINT_NONE);
} }
static FORCEINLINE __vec16_i64 __cast_sext(const __vec16_i64 &, const __vec16_i1 &val)
{
__vec16_i32 ret = _mm512_mask_mov_epi32(_mm512_setzero_epi32(), val, _mm512_set1_epi32(-1));
return __vec16_i64(ret, ret);
}
static FORCEINLINE __vec16_i64 __cast_sext(const __vec16_i64 &, const __vec16_i8 &val)
{
__vec16_i32 a = _mm512_extload_epi32(&val, _MM_UPCONV_EPI32_SINT8, _MM_BROADCAST32_NONE, _MM_HINT_NONE);
return __vec16_i64(a.v, _mm512_srai_epi32(a.v, 31));
}
static FORCEINLINE __vec16_i64 __cast_sext(const __vec16_i64 &, const __vec16_i32 &val) static FORCEINLINE __vec16_i64 __cast_sext(const __vec16_i64 &, const __vec16_i32 &val)
{ {
return __vec16_i64(val.v, _mm512_srai_epi32(val.v, 31)); return __vec16_i64(val.v, _mm512_srai_epi32(val.v, 31));
} }
static FORCEINLINE __vec16_i8 __cast_zext(const __vec16_i8 &, const __vec16_i1 &val) static FORCEINLINE __vec16_i8 __cast_zext(const __vec16_i8 &, const __vec16_i1 &val)
{ {
return __vec16_i8(val[0], val[1], val[2], val[3], val[4], val[5], val[6], val[7], return __vec16_i8(val[0], val[1], val[2], val[3], val[4], val[5], val[6], val[7],