[sse2neon+compile] Re-apply sse2neon changes, and add support for Win…

…dows ARM64

CMakeLists.txt

@@ -35,7 +35,7 @@ SET(OPENPGL_ARM OFF)
 IF (APPLE AND CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND (CMAKE_SYSTEM_PROCESSOR STREQUAL "arm64" OR CMAKE_OSX_ARCHITECTURES MATCHES "arm64"))
   MESSAGE(STATUS "Building for Apple silicon")
   SET(OPENPGL_ARM ON)
-ELSEIF(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
+ELSEIF(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64" OR CMAKE_SYSTEM_PROCESSOR STREQUAL "ARM64")
   MESSAGE(STATUS "Building for AArch64")
   SET(OPENPGL_ARM ON)
 ENDIF()

openpgl/CMakeLists.txt

@@ -78,7 +78,8 @@ message(STATUS "Compiler: ${CMAKE_CXX_COMPILER_ID}")
 message(STATUS "Arch: ${CMAKE_SYSTEM_PROCESSOR}")
 
 if(WIN32)
-  if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
+  # Here we chack for MSVC, or Clang pretending to be MSVC via Clang-CL
+  if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC" OR (CMAKE_CXX_COMPILER_ID STREQUAL "Clang" AND CMAKE_CXX_SIMULATE_ID STREQUAL "MSVC"))
     set(OPENPGL_RELEASE_OPTIONS /Ox /Oi)  
     set(OPENPGL_COMMON_OPTIONS /fp:precise)
     #set(OPENPGL_RELEASE_OPTIONS ${OPENPGL_RELEASE_OPTIONS} -ftree-vectorize -mfpmath=sse -funsafe-math-optimizations -fno-rounding-math -fno-signaling-nans -fno-math-errno -fomit-frame-pointer )
@@ -95,6 +96,14 @@ if(WIN32)
     if(OPENPGL_ISA_AVX512)
       set_source_files_properties(api/deviceCPU16.cpp PROPERTIES COMPILE_FLAGS "/D__SSE__ /D__SSE2__ /D__SSE3__ /D__SSE4_1__ /D__SSE4_2__ /arch:AVX /arch:AVX2 /arch:AVX512")
     endif()
+    if(CMAKE_CXX_COMPILER_ID STREQUAL "Clang")
+      if(OPENPGL_ISA_NEON)
+       set_source_files_properties(api/deviceCPU4.cpp PROPERTIES COMPILE_FLAGS "/D__SSE4_2__  /D__SSE4_1__")
+      endif()
+      if(OPENPGL_ISA_NEON2X)
+        set_source_files_properties(api/deviceCPU8.cpp PROPERTIES COMPILE_FLAGS "/D__AVX2__ /D__AVX__ /D__SSE4_2__  /D__SSE4_1__  /D__BMI__ /D__BMI2__ /D__LZCNT__")
+      endif()
+    endif()
   elseif(CMAKE_CXX_COMPILER_ID STREQUAL "IntelLLVM" OR CMAKE_CXX_COMPILER_ID STREQUAL "dpcpp")
     set(OPENPGL_RELEASE_OPTIONS -O3)
     set(OPENPGL_COMMON_OPTIONS -Wall) 

third-party/embreeSrc/common/simd/arm/sse2neon.h

@@ -336,6 +336,14 @@ FORCE_INLINE void _sse2neon_smp_mb(void)
  * argument "a" of mm_shuffle_ps that will be places in fp1 of result.
  * fp0 is the same for fp0 of result.
  */
+#if defined(__aarch64__)
+#define _MN_SHUFFLE(fp3,fp2,fp1,fp0) ( (uint8x16_t){ (((fp3)*4)+0), (((fp3)*4)+1), (((fp3)*4)+2), (((fp3)*4)+3),  (((fp2)*4)+0), (((fp2)*4)+1), (((fp2)*4)+\
+2), (((fp2)*4)+3),  (((fp1)*4)+0), (((fp1)*4)+1), (((fp1)*4)+2), (((fp1)*4)+3),  (((fp0)*4)+0), (((fp0)*4)+1), (((fp0)*4)+2), (((fp0)*4)+3) } )
+#define _MF_SHUFFLE(fp3,fp2,fp1,fp0) ( (uint8x16_t){ (((fp3)*4)+0), (((fp3)*4)+1), (((fp3)*4)+2), (((fp3)*4)+3),  (((fp2)*4)+0), (((fp2)*4)+1), (((fp2)*4)+\
+2), (((fp2)*4)+3),  (((fp1)*4)+16+0), (((fp1)*4)+16+1), (((fp1)*4)+16+2), (((fp1)*4)+16+3),  (((fp0)*4)+16+0), (((fp0)*4)+16+1), (((fp0)*4)+16+2), (((fp0)*\
+4)+16+3) } )
+#endif
+
 #define _MM_SHUFFLE(fp3, fp2, fp1, fp0) \
     (((fp3) << 6) | ((fp2) << 4) | ((fp1) << 2) | ((fp0)))
 
@@ -2822,7 +2830,7 @@ FORCE_INLINE void _mm_stream_pi(__m64 *p, __m64 a)
 FORCE_INLINE void _mm_stream_ps(float *p, __m128 a)
 {
 #if __has_builtin(__builtin_nontemporal_store)
-    __builtin_nontemporal_store(a, (float32x4_t *) p);
+    __builtin_nontemporal_store(reinterpret_cast<float32x4_t>(a), (float32x4_t *) p);
 #else
     vst1q_f32(p, vreinterpretq_f32_m128(a));
 #endif
@@ -5660,7 +5668,7 @@ FORCE_INLINE void _mm_storeu_si32(void *p, __m128i a)
 FORCE_INLINE void _mm_stream_pd(double *p, __m128d a)
 {
 #if __has_builtin(__builtin_nontemporal_store)
-    __builtin_nontemporal_store(a, (__m128d *) p);
+    __builtin_nontemporal_store(reinterpret_cast<float32x4_t>(a), (float32x4_t *) p);
 #elif defined(__aarch64__) || defined(_M_ARM64)
     vst1q_f64(p, vreinterpretq_f64_m128d(a));
 #else
@@ -6809,14 +6817,14 @@ FORCE_INLINE __m64 _mm_sign_pi8(__m64 _a, __m64 _b)
     _sse2neon_define2(                                                  \
         __m128i, a, b,                                                  \
         const uint16_t _mask[8] =                                       \
-            _sse2neon_init(((imm) & (1 << 0)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 1)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 2)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 3)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 4)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 5)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 6)) ? (uint16_t) -1 : 0x0,    \
-                           ((imm) & (1 << 7)) ? (uint16_t) -1 : 0x0);   \
+            _sse2neon_init(((imm) & (1 << 0)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 1)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 2)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 3)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 4)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 5)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 6)) ? (uint16_t)0xffff : (uint16_t)0x0000,    \
+                           ((imm) & (1 << 7)) ? (uint16_t)0xffff : (uint16_t)0x0000);   \
         uint16x8_t _mask_vec = vld1q_u16(_mask);                        \
         uint16x8_t __a = vreinterpretq_u16_m128i(_a);                   \
         uint16x8_t __b = vreinterpretq_u16_m128i(_b); _sse2neon_return( \

third-party/embreeSrc/common/sys/intrinsics.h

@@ -91,7 +91,7 @@ namespace embree
 
 #if defined(__X86_64__) || defined (__aarch64__)
   __forceinline size_t bsf(size_t v) {
-#if defined(__AVX2__) 
+#if defined(__AVX2__) && !defined(__aarch64__)
     return _tzcnt_u64(v);
 #else
     unsigned long r = 0; _BitScanForward64(&r,v); return r;
@@ -140,7 +140,7 @@ namespace embree
 
 #if defined(__X86_64__) || defined (__aarch64__)
   __forceinline size_t bsr(size_t v) {
-#if defined(__AVX2__) 
+#if defined(__AVX2__) && !defined(__aarch64__)
     return 63 -_lzcnt_u64(v);
 #else
     unsigned long r = 0; _BitScanReverse64(&r, v); return r;