[llvm] [X86] Add test for VPBLENDW Tuning (PR #179906)

[Simon Pilgrim via llvm-commits]

================
@@ -0,0 +1,309 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
+# RUN: llc %s -mtriple=x86_64-unknown-unknown -mattr=+avxvnni -run-pass=x86-fixup-inst-tuning -o - | FileCheck %s --check-prefixes=AVX-COMMON,AVXVNNI
+# RUN: llc %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vnni -run-pass=x86-fixup-inst-tuning -o - | FileCheck %s --check-prefixes=AVX-COMMON,AVX512VNNI
+# RUN: llc %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vnni -mattr=+avx512vl -run-pass=x86-fixup-inst-tuning -o - | FileCheck %s --check-prefixes=AVX-COMMON,AVX512VNNI
+# RUN: llc %s -mtriple=x86_64-unknown-unknown -mattr=+avx  -run-pass=x86-fixup-inst-tuning -o - | FileCheck %s --check-prefixes=AVX-COMMON,AVX
+# RUN: llc %s -mtriple=x86_64-unknown-unknown -mattr=+avx2  -run-pass=x86-fixup-inst-tuning -o - | FileCheck %s --check-prefixes=AVX-COMMON,AVX2
+# RUN: llc %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fast-variable-crosslane-shuffle,+fast-variable-perlane-shuffle -run-pass=x86-fixup-inst-tuning  -o - | FileCheck %s --check-prefixes=AVX-COMMON,AVX2VCLS
----------------
RKSimon wrote:

Are the fast attributes having an effect? Fixup tuning is supposed to be scheduler model driven, IIRC the only subtarget tuning attributes that has an effect is domain crossing (and thats because scheduler can't model them).

https://github.com/llvm/llvm-project/pull/179906