[X86] avx512fp16intrin.h - allow fp16 truncation casts to be used in constexpr (llvm#162250)

_mm256_castph256_ph128
_mm512_castph512_ph128
_mm512_castph512_ph256

Part of llvm#156348

Author: RKSimon

clang/lib/Headers/avx512fp16intrin.h

@@ -193,17 +193,17 @@ _mm512_castsi512_ph(__m512i __a) {
   return (__m512h)__a;
 }
 
-static __inline__ __m128h __DEFAULT_FN_ATTRS256
+static __inline__ __m128h __DEFAULT_FN_ATTRS256_CONSTEXPR
 _mm256_castph256_ph128(__m256h __a) {
   return __builtin_shufflevector(__a, __a, 0, 1, 2, 3, 4, 5, 6, 7);
 }
 
-static __inline__ __m128h __DEFAULT_FN_ATTRS512
+static __inline__ __m128h __DEFAULT_FN_ATTRS512_CONSTEXPR
 _mm512_castph512_ph128(__m512h __a) {
   return __builtin_shufflevector(__a, __a, 0, 1, 2, 3, 4, 5, 6, 7);
 }
 
-static __inline__ __m256h __DEFAULT_FN_ATTRS512
+static __inline__ __m256h __DEFAULT_FN_ATTRS512_CONSTEXPR
 _mm512_castph512_ph256(__m512h __a) {
   return __builtin_shufflevector(__a, __a, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11,
                                  12, 13, 14, 15);

clang/test/CodeGen/X86/avx512fp16-builtins.c

@@ -341,18 +341,21 @@ __m128h test_mm256_castph256_ph128(__m256h __a) {
   // CHECK: shufflevector <16 x half> %{{.*}}, <16 x half> %{{.*}}, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   return _mm256_castph256_ph128(__a);
 }
+TEST_CONSTEXPR(match_m128h(_mm256_castph256_ph128((__m256h){-1.0, 2.0, -3.0, 4.0, -5.0, 6.0, -7.0, 8.0, -9.0, 10.0, -11.0, 12.0, -13.0, 14.0, -15.0, -16.0}), -1.0, 2.0, -3.0, 4.0, -5.0, 6.0, -7.0, 8.0));
 
 __m128h test_mm512_castph512_ph128(__m512h __a) {
   // CHECK-LABEL: test_mm512_castph512_ph128
   // CHECK: shufflevector <32 x half> %{{.*}}, <32 x half> %{{.*}}, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   return _mm512_castph512_ph128(__a);
 }
+TEST_CONSTEXPR(match_m128h(_mm512_castph512_ph128((__m512h){0.0, -1.0, 2.0, -3.0, 4.0, -5.0, 6.0, -7.0, 8.0, -9.0, 10.0, -11.0, 12.0, -13.0, 14.0, -15.0, -16.0, -17.0, 18.0, -19.0, 20.0, -21.0, 22.0, -23.0, 24.0, -25.0, 26.0, -27.0, 28.0, -29.0, 30.0, -31.0}), 0.0, -1.0, 2.0, -3.0, 4.0, -5.0, 6.0, -7.0));
 
 __m256h test_mm512_castph512_ph256(__m512h __a) {
   // CHECK-LABEL: test_mm512_castph512_ph256
   // CHECK: shufflevector <32 x half> %{{.*}}, <32 x half> %{{.*}}, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
   return _mm512_castph512_ph256(__a);
 }
+TEST_CONSTEXPR(match_m256h(_mm512_castph512_ph256((__m512h){-1.0, 2.0, -3.0, 4.0, -5.0, 6.0, -7.0, 8.0, -9.0, 10.0, -11.0, 12.0, -13.0, 14.0, -15.0, -16.0, -17.0, 18.0, -19.0, 20.0, -21.0, 22.0, -23.0, 24.0, -25.0, 26.0, -27.0, 28.0, -29.0, 30.0, -31.0, 32.0}), -1.0, 2.0, -3.0, 4.0, -5.0, 6.0, -7.0, 8.0, -9.0, 10.0, -11.0, 12.0, -13.0, 14.0, -15.0, -16.0));
 
 __m256h test_mm256_castph128_ph256(__m128h __a) {
   // CHECK-LABEL: test_mm256_castph128_ph256