[Mlir-commits] [clang] [llvm] [mlir] [x86][AVX-VNNI] Fix VPDPBXXD Argument Type (PR #159222)

[llvmlistbot at llvm.org]

================
@@ -0,0 +1,291 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx10.2 --show-mc-encoding | FileCheck %s --check-prefixes=X86
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.2 --show-mc-encoding | FileCheck %s --check-prefixes=X64
+
+declare <4 x i32> @llvm.x86.avx2.vpdpbssd.128(<4 x i32>, <4 x i32>, <4 x i32>)
----------------
BaiXilin wrote:

`BuiltinsX86.td` lists all `vpdpbxxd_128/256` intrinsics under both avxvnniint8 and avx10.2, and `vpdpbxxd_512` is only under avx10.2. I chose avx10.2 so I can put all test cases into one file.

Should I move 128- and 256-bit wide cases to avxvnniint8?

https://github.com/llvm/llvm-project/pull/159222