simd.h: address compiler warnings (#4724)

* We passed a bad flag to _mm512_cvtps_ph intrinsics because of what I
can only imagine was a bad cut-and-paste years ago. (Harmless because it
set bits never used, but a warning on Windows.)

* Remove clauses for AVX512-EL, I'm not sure why they were there,
because those instructions were only for "Knights Landing" which was
never made available to the general public and has been discontinued. It
must be that years ago, I got a little overzealous with the SIMD
intrinsics.

Signed-off-by: Larry Gritz <[email protected]>

Author: lgritz

src/include/OpenImageIO/platform.h

@@ -602,7 +602,7 @@ inline bool cpu_has_avx512f() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<16)) !
 inline bool cpu_has_avx512dq() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<17)) != 0; }
 inline bool cpu_has_avx512ifma() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<21)) != 0; }
 inline bool cpu_has_avx512pf() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<26)) != 0; }
-inline bool cpu_has_avx512er() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<27)) != 0; }
+inline bool cpu_has_avx512er() { return false; /* knights landing only */ }
 inline bool cpu_has_avx512cd() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<28)) != 0; }
 inline bool cpu_has_avx512bw() {int i[4]; cpuid(i,7,0); return (i[1] & (1<<30)) != 0; }
 inline bool cpu_has_avx512vl() {int i[4]; cpuid(i,7,0); return (i[1] & (0x80000000 /*1<<31*/)) != 0; }

src/include/OpenImageIO/simd.h

@@ -190,11 +190,6 @@
 #  else
 #    define OIIO_AVX512PF_ENABLED 0
 #  endif
-#  if defined(__AVX512ER__)
-#    define OIIO_AVX512ER_ENABLED 1   /* Exponential & reciprocal */
-#  else
-#    define OIIO_AVX512ER_ENABLED 0
-#  endif
 #  if defined(__AVX512CD__)
 #    define OIIO_AVX512CD_ENABLED 1   /* Conflict detection */
 #  else
@@ -215,7 +210,6 @@
 #  define OIIO_AVX512VL_ENABLED 0
 #  define OIIO_AVX512DQ_ENABLED 0
 #  define OIIO_AVX512PF_ENABLED 0
-#  define OIIO_AVX512ER_ENABLED 0
 #  define OIIO_AVX512CD_ENABLED 0
 #  define OIIO_AVX512BW_ENABLED 0
 #endif
@@ -6903,7 +6897,7 @@ OIIO_FORCEINLINE void vfloat4::store (float *values, int n) const {
 #if defined(_HALF_H_) || defined(IMATH_HALF_H_)
 OIIO_FORCEINLINE void vfloat4::store (half *values) const {
 #if OIIO_F16C_ENABLED && OIIO_SIMD_SSE
-    __m128i h = _mm_cvtps_ph (m_simd, (_MM_FROUND_TO_NEAREST_INT |_MM_FROUND_NO_EXC));
+    __m128i h = _mm_cvtps_ph (m_simd, _MM_FROUND_TO_NEAREST_INT);
     _mm_store_sd ((double *)values, _mm_castsi128_pd(h));
 #elif OIIO_SIMD_NEON
     float16x4_t f16 = vcvt_f16_f32(m_simd);
@@ -7627,10 +7621,7 @@ OIIO_FORCEINLINE vfloat4 rsqrt (const vfloat4 &a)
 
 OIIO_FORCEINLINE vfloat4 rsqrt_fast (const vfloat4 &a)
 {
-#if OIIO_SIMD_AVX >= 512 && OIIO_AVX512ER_ENABLED
-    // Trickery: in and out of the 512 bit registers to use fast approx rsqrt
-    return _mm512_castps512_ps128(_mm512_rsqrt28_round_ps(_mm512_castps128_ps512(a), _MM_FROUND_NO_EXC));
-#elif OIIO_SIMD_AVX >= 512 && OIIO_AVX512VL_ENABLED
+#if OIIO_SIMD_AVX >= 512 && OIIO_AVX512VL_ENABLED
     // Trickery: in and out of the 512 bit registers to use fast approx rsqrt
     return _mm512_castps512_ps128(_mm512_rsqrt14_ps(_mm512_castps128_ps512(a)));
 #elif OIIO_SIMD_SSE
@@ -8794,7 +8785,7 @@ OIIO_FORCEINLINE void vfloat8::store (float *values, int n) const {
 #if defined(_HALF_H_) || defined(IMATH_HALF_H_)
 OIIO_FORCEINLINE void vfloat8::store (half *values) const {
 #if OIIO_SIMD_AVX && OIIO_F16C_ENABLED
-    __m128i h = _mm256_cvtps_ph (m_simd, (_MM_FROUND_TO_NEAREST_INT |_MM_FROUND_NO_EXC));
+    __m128i h = _mm256_cvtps_ph (m_simd, _MM_FROUND_TO_NEAREST_INT);
     _mm_storeu_si128 ((__m128i *)values, h);
 #elif OIIO_SIMD_SSE || OIIO_SIMD_NEON
     m_4[0].store(values);
@@ -9285,10 +9276,7 @@ OIIO_FORCEINLINE vfloat8 rsqrt (const vfloat8 &a)
 
 OIIO_FORCEINLINE vfloat8 rsqrt_fast (const vfloat8 &a)
 {
-#if OIIO_SIMD_AVX >= 512 && OIIO_AVX512ER_ENABLED
-    // Trickery: in and out of the 512 bit registers to use fast approx rsqrt
-    return _mm512_castps512_ps256(_mm512_rsqrt28_round_ps(_mm512_castps256_ps512(a), _MM_FROUND_NO_EXC));
-#elif OIIO_SIMD_AVX >= 512 && OIIO_AVX512VL_ENABLED
+#if OIIO_SIMD_AVX >= 512 && OIIO_AVX512VL_ENABLED
     // Trickery: in and out of the 512 bit registers to use fast approx rsqrt
     return _mm512_castps512_ps256(_mm512_rsqrt14_ps(_mm512_castps256_ps512(a)));
 #elif OIIO_SIMD_AVX
@@ -9673,7 +9661,7 @@ OIIO_FORCEINLINE void vfloat16::store (float *values, int n) const {
 #if defined(_HALF_H_) || defined(IMATH_HALF_H_)
 OIIO_FORCEINLINE void vfloat16::store (half *values) const {
 #if OIIO_SIMD_AVX >= 512
-    __m256i h = _mm512_cvtps_ph (m_simd, (_MM_FROUND_TO_NEAREST_INT |_MM_FROUND_NO_EXC));
+    __m256i h = _mm512_cvtps_ph (m_simd, _MM_FROUND_TO_NEAREST_INT);
     _mm256_storeu_si256 ((__m256i *)values, h);
 #else
     m_8[0].store (values);
@@ -10113,9 +10101,7 @@ OIIO_FORCEINLINE vint16 rint (const vfloat16& a)
 
 OIIO_FORCEINLINE vfloat16 rcp_fast (const vfloat16 &a)
 {
-#if OIIO_SIMD_AVX >= 512 && OIIO_AVX512ER_ENABLED
-    return _mm512_rcp28_ps(a);
-#elif OIIO_SIMD_AVX >= 512
+#if OIIO_SIMD_AVX >= 512
     vfloat16 r = _mm512_rcp14_ps(a);
     return r * nmadd (r, a, vfloat16(2.0f));
 #else
@@ -10146,9 +10132,7 @@ OIIO_FORCEINLINE vfloat16 rsqrt (const vfloat16 &a)
 
 OIIO_FORCEINLINE vfloat16 rsqrt_fast (const vfloat16 &a)
 {
-#if OIIO_SIMD_AVX >= 512 && OIIO_AVX512ER_ENABLED
-    return _mm512_rsqrt28_round_ps(a, _MM_FROUND_NO_EXC);
-#elif OIIO_SIMD_AVX >= 512
+#if OIIO_SIMD_AVX >= 512
     return _mm512_rsqrt14_ps (a);
 #else
     return vfloat16(rsqrt_fast(a.lo()), rsqrt_fast(a.hi()));

src/libOpenImageIO/imageio.cpp

@@ -170,7 +170,6 @@ hw_simd_caps()
     if (cpu_has_avx512dq())    caps.emplace_back ("avx512dq");
     if (cpu_has_avx512ifma())  caps.emplace_back ("avx512ifma");
     if (cpu_has_avx512pf())    caps.emplace_back ("avx512pf");
-    if (cpu_has_avx512er())    caps.emplace_back ("avx512er");
     if (cpu_has_avx512cd())    caps.emplace_back ("avx512cd");
     if (cpu_has_avx512bw())    caps.emplace_back ("avx512bw");
     if (cpu_has_avx512vl())    caps.emplace_back ("avx512vl");
@@ -202,7 +201,6 @@ oiio_simd_caps()
     if (OIIO_AVX512DQ_ENABLED)   caps.emplace_back ("avx512dq");
     if (OIIO_AVX512IFMA_ENABLED) caps.emplace_back ("avx512ifma");
     if (OIIO_AVX512PF_ENABLED)   caps.emplace_back ("avx512pf");
-    if (OIIO_AVX512ER_ENABLED)   caps.emplace_back ("avx512er");
     if (OIIO_AVX512CD_ENABLED)   caps.emplace_back ("avx512cd");
     if (OIIO_AVX512BW_ENABLED)   caps.emplace_back ("avx512bw");
     if (OIIO_AVX512VL_ENABLED)   caps.emplace_back ("avx512vl");