[PATCH] D18215: [CLANG][avx512][BUILTIN] Adding fixupimm{pd|ps|sd|ss} getexp{sd|ss} getmant{sd|ss} kunpck{di|si} loada{pd|ps} loaddqu{di|hi|qi|si} max{sd|ss} min{sd|ss} kmov16 builtins to clang

[Asaf Badouh via llvm-commits]

AsafBadouh added inline comments.

================
Comment at: lib/Headers/avx512fintrin.h:54
@@ -53,1 +53,3 @@
 
+typedef enum
+{
----------------
move the whole block to line 48, right after the rounding mode macros.

================
Comment at: lib/Headers/avx512fintrin.h:75
@@ -59,1 +74,3 @@
 
+static  __inline __m128i __DEFAULT_FN_ATTRS
+_mm512_setzero_sd128(void)
----------------
Not sure there is need for these two intrinsics.
you can use _mm_setzero_pd/ps.

================
Comment at: lib/Headers/avx512fintrin.h:78
@@ +77,3 @@
+{
+  return (__m128i)(__v2df){ 0.0, 0.0 };
+}
----------------
return type should be __m128d

================
Comment at: lib/Headers/avx512vlbwintrin.h:2968
@@ +2967,3 @@
+  return (__m128i) __builtin_ia32_loaddquhi128_mask ((__v8hi *) __P,
+                 (__v8hi)
+                 _mm_setzero_hi (),
----------------
can you use _mm_setzero_si128() instead?


================
Comment at: lib/Headers/avx512vlbwintrin.h:3003
@@ +3002,3 @@
+                 (__v16qi)
+                 _mm_setzero_hi (),
+                 (__mmask16) __U);
----------------
can you use _mm_setzero_si128() instead?


================
Comment at: lib/Headers/avx512vlintrin.h:6249
@@ +6248,3 @@
+                 (__v2di)
+                 _mm_setzero_di (),
+                 (__mmask8) __U);
----------------
please use _mm_setzero_si128()


http://reviews.llvm.org/D18215