[PATCH] D135930: [X86] Add AVX-NE-CONVERT instructions.

[Phoebe Wang via Phabricator via cfe-commits]

pengfei added inline comments.


================
Comment at: clang/include/clang/Basic/BuiltinsX86.def:2106-2107
+TARGET_BUILTIN(__builtin_ia32_vcvtneoph2ps256, "V8fV16xC*", "nV:256:", "avxneconvert")
+TARGET_BUILTIN(__builtin_ia32_vcvtneps2bf16128, "V8sV4f", "nV:128:", "avxneconvert")
+TARGET_BUILTIN(__builtin_ia32_vcvtneps2bf16256, "V8sV8f", "nV:256:", "avxneconvert")
 TARGET_HEADER_BUILTIN(_InterlockedAnd64,         "WiWiD*Wi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
----------------
These should be shared with AVX512-BF16.


================
Comment at: clang/lib/Headers/avxneconvertintrin.h:86-94
+static __inline__ __m128bh __DEFAULT_FN_ATTRS128
+_mm_cvtneps_avx_pbh(__m128 __A) {
+  return (__m128bh)__builtin_ia32_vcvtneps2bf16128((__v4sf)__A);
+}
+
+static __inline__ __m128bh __DEFAULT_FN_ATTRS256
+_mm256_cvtneps_avx_pbh(__m256 __A) {
----------------
Add unified intrinsics like AVXVNNI.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D135930/new/

https://reviews.llvm.org/D135930