[llvm] [AArch64][NEON][SVE] Lower mixed sign/zero extended partial reductions to usdot (PR #107566)

Paul Walker via llvm-commits llvm-commits at lists.llvm.org
Wed Sep 18 04:17:55 PDT 2024


================
@@ -1,6 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
-; RUN: llc -mtriple aarch64 -mattr=+neon,+dotprod < %s | FileCheck %s --check-prefixes=CHECK,CHECK-DOT
-; RUN: llc -mtriple aarch64 -mattr=+neon < %s | FileCheck %s --check-prefixes=CHECK,CHECK-NODOT
+; RUN: llc -mtriple aarch64 -mattr=+neon,+dotprod,+i8mm < %s | FileCheck %s --check-prefixes=CHECK,CHECK-DOT,CHECK-I8MM
+; RUN: llc -mtriple aarch64 -mattr=+neon,+i8mm < %s | FileCheck %s --check-prefixes=CHECK,CHECK-NODOT,CHECK-I8MM
+; RUN: llc -mtriple aarch64 -mattr=+neon,+dotprod < %s | FileCheck %s --check-prefixes=CHECK,CHECK-DOT,CHECK-NOI8MM
----------------
paulwalker-arm wrote:

I think I see what has happened.  I had assumed `+i8mm` would enable `usdot` instructions but the DAG combine requires both `+dotprod` and `+i8mm` for that to happen.

https://github.com/llvm/llvm-project/pull/107566


More information about the llvm-commits mailing list