[Mesa-dev] [PATCH 1/3] swr/rast: simdlib changes for clang/gcc
Tim Rowley
timothy.o.rowley at intel.com
Thu Jul 20 22:09:42 UTC 2017
Tested with clang-4.0 and gcc-6.3.
---
.../swr/rasterizer/common/simdlib_512_avx512.inl | 43 +++++++++++++++++-----
.../swr/rasterizer/common/simdlib_types.hpp | 2 +-
2 files changed, 35 insertions(+), 10 deletions(-)
diff --git a/src/gallium/drivers/swr/rasterizer/common/simdlib_512_avx512.inl b/src/gallium/drivers/swr/rasterizer/common/simdlib_512_avx512.inl
index 7d90b7d1b0..7447d35ee2 100644
--- a/src/gallium/drivers/swr/rasterizer/common/simdlib_512_avx512.inl
+++ b/src/gallium/drivers/swr/rasterizer/common/simdlib_512_avx512.inl
@@ -24,6 +24,21 @@
#error Do not include this file directly, use "simdlib.hpp" instead.
#endif
+#if defined(__GNUC__) && !defined( __clang__) && !defined(__INTEL_COMPILER)
+// gcc missing these intrinsics
+#ifndef _mm512_cmpneq_ps_mask
+#define _mm512_cmpneq_ps_mask(a,b) _mm512_cmp_ps_mask((a),(b),_CMP_NEQ_UQ)
+#endif
+
+#ifndef _mm512_cmplt_ps_mask
+#define _mm512_cmplt_ps_mask(a,b) _mm512_cmp_ps_mask((a),(b),_CMP_LT_OS)
+#endif
+
+#ifndef _mm512_cmplt_pd_mask
+#define _mm512_cmplt_pd_mask(a,b) _mm512_cmp_pd_mask((a),(b),_CMP_LT_OS)
+#endif
+#endif
+
//============================================================================
// SIMD16 AVX512 (F) implementation
//
@@ -138,6 +153,17 @@ using SIMD256T = SIMD256Impl::AVX2Impl;
}
#define SIMD_IWRAPPER_2I(op) SIMD_IWRAPPER_2I_(op, op)
+#define SIMD_EMU_IWRAPPER_2(op) \
+ static SIMDINLINE \
+ Integer SIMDCALL op(Integer a, Integer b)\
+ {\
+ return Integer\
+ {\
+ SIMD256T::op(a.v8[0], b.v8[0]),\
+ SIMD256T::op(a.v8[1], b.v8[1]),\
+ };\
+ }
+
private:
static SIMDINLINE Integer vmask(__mmask8 m)
{
@@ -234,14 +260,6 @@ SIMD_IWRAPPER_1I(slli_epi32); // return a << ImmT
SIMD_IWRAPPER_2(sllv_epi32);
SIMD_IWRAPPER_1I(srai_epi32); // return a >> ImmT (int32)
SIMD_IWRAPPER_1I(srli_epi32); // return a >> ImmT (uint32)
-SIMD_IWRAPPER_1I_(srli_si, srli_si512); // return a >> (ImmT*8) (uint)
-
-template<int ImmT> // same as srli_si, but with Float cast to int
-static SIMDINLINE Float SIMDCALL srlisi_ps(Float a)
-{
- return castsi_ps(srli_si<ImmT>(castps_si(a)));
-}
-
SIMD_IWRAPPER_2(srlv_epi32);
//-----------------------------------------------------------------------
@@ -443,10 +461,17 @@ static SIMDINLINE Integer SIMDCALL insert_si(Integer a, SIMD256Impl::Integer b)
return _mm512_inserti64x4(a, b, imm);
}
+#if !defined(AVX512F_STRICT)
SIMD_IWRAPPER_2(packs_epi16); // See documentation for _mm512_packs_epi16 and _mm512_packs_epi16
SIMD_IWRAPPER_2(packs_epi32); // See documentation for _mm512_packs_epi32 and _mm512_packs_epi32
SIMD_IWRAPPER_2(packus_epi16); // See documentation for _mm512_packus_epi16 and _mm512_packus_epi16
SIMD_IWRAPPER_2(packus_epi32); // See documentation for _mm512_packus_epi32 and _mm512_packus_epi32
+#else
+SIMD_EMU_IWRAPPER_2(packs_epi16)
+SIMD_EMU_IWRAPPER_2(packs_epi32)
+SIMD_EMU_IWRAPPER_2(packus_epi16)
+SIMD_EMU_IWRAPPER_2(packus_epi32)
+#endif
static SIMDINLINE Integer SIMDCALL permute_epi32(Integer a, Integer swiz) // return a[swiz[i]] for each 32-bit lane i (float)
{
@@ -679,4 +704,4 @@ static SIMDINLINE Float SIMDCALL vmask_ps(int32_t mask)
#undef SIMD_IWRAPPER_2
#undef SIMD_IWRAPPER_2_
#undef SIMD_IWRAPPER_2I
-
+#undef SIMD_EMU_IWRAPPER_2
diff --git a/src/gallium/drivers/swr/rasterizer/common/simdlib_types.hpp b/src/gallium/drivers/swr/rasterizer/common/simdlib_types.hpp
index 07775e7b83..bc23867c7b 100644
--- a/src/gallium/drivers/swr/rasterizer/common/simdlib_types.hpp
+++ b/src/gallium/drivers/swr/rasterizer/common/simdlib_types.hpp
@@ -262,7 +262,7 @@ namespace SIMDImpl
namespace SIMD512Impl
{
-#if !defined(_MM_K0_REG)
+#if !defined(__AVX512F__)
// Define AVX512 types if not included via immintrin.h.
// All data members of these types are ONLY to viewed
// in a debugger. Do NOT access them via code!
--
2.11.0
More information about the mesa-dev
mailing list