[PATCH] D99152: [AMX] Prototype for vector and amx bitcast.

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri added a comment.

I'm a little bit lost with all this AMX stuff.
Could you please explain in normal human words, what does `__tile_loadd()` do?
I.e. given

  void wrapper(__tile& dst, const void *base, int stride) {
    _tile_loadd(dst, const void *base, int stride);
  }

which bytes from `base` will be loaded?



================
Comment at: clang/test/CodeGen/X86/amx_cast.c:1
+// RUN: %clang_cc1 %s -O2 -ffreestanding -triple=x86_64-unknown-unknown  -target-feature +avx512f  -target-feature +amx-int8  \
+// RUN: -target-feature +amx-bf16 -emit-llvm -o - -Werror -pedantic | FileCheck %s --check-prefixes=CHECK
----------------
Please don't use llvm optimizations in clang tests.


================
Comment at: clang/test/CodeGen/X86/amx_cast.c:4
+
+#include <immintrin.h>
+
----------------
The tests should be hermetic.
Which `immintrin.h` is being used there?
Hopefully the one from clang, not the one from system?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D99152/new/

https://reviews.llvm.org/D99152