[llvm] [AMDGPU] Ensure v_mfma_scale_f32_{16x16x128|32x32x64}_f8f6f4 instructions are convergent (PR #178627)

[Frederik Harwath via llvm-commits]

================
@@ -1,18 +1,52 @@
-# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx9-generic -run-pass=machine-sink %s -stats 2>&1 -o - | \
-# RUN:   not grep "Number of machine instructions sunk"
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
+# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx950 -run-pass=machine-sink -o - %s | FileCheck %s
 # machine-sink must not sink MFMA instructions.
 # Ensure that MFMA instructions are marked as convergent to prevent
 # machine-sink from sinking them.
 
+--- |
+  define void @test_V_MFMA_F32_32X32X64_F8F6F4_f4_f4_e64() { ret void }
----------------
frederik-h wrote:

Very good, thanks! I have updated the test.

https://github.com/llvm/llvm-project/pull/178627