[PATCH] D140531: [X86] Add reduce_*_ep[i|u]8/16 series intrinsics.

[Kan Shengchen via Phabricator via cfe-commits]

skan added inline comments.


================
Comment at: clang/docs/ReleaseNotes.rst:806
 - Lift _BitInt() supported max width from 128 to 8388608.
+- Support *_reduce_*_ep[i|u]8/16 series intrinsics.
 
----------------
This is not clear to me, try sth like "mm*_reduce_(add|and|max)_epi" and put the name in a code-block like the previous lines.


================
Comment at: clang/test/CodeGen/X86/avx512vlbw-reduceIntrin.c:1
+// RUN: %clang_cc1 -ffreestanding %s -O0 -triple=x86_64-apple-darwin -target-feature +avx512bw -target-feature +avx512vl -emit-llvm -o - -Wall -Werror | FileCheck %s
+
----------------
drop "-apple-darwin"?


================
Comment at: clang/test/CodeGen/X86/avx512vlbw-reduceIntrin.c:115
+
+char test_mm_reduce_mul_epi8(__m128i __W){
+// CHECK-LABEL: @test_mm_reduce_mul_epi8(
----------------
char has different meaning on different platforms. Should we use "signed char" or "unsigned char" explicitly?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D140531/new/

https://reviews.llvm.org/D140531