[llvm] [X86][APX] Enable NDD tunings (PR #186049)

[Phoebe Wang via llvm-commits]

================
@@ -1,8 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
 ; RUN: llc -mtriple=x86_64-linux -mcpu=atom < %s | FileCheck %s
 ; RUN: llc -mtriple=x86_64-linux -mcpu=slm < %s | FileCheck %s -check-prefix=SLM
-; RUN: llc -mtriple=x86_64-linux -mattr=+egpr,+ndd,+movbe --show-mc-encoding < %s | FileCheck %s -check-prefix=EGPR
-; RUN: llc -mtriple=x86_64-linux -mattr=+egpr,+ndd --show-mc-encoding < %s | FileCheck %s -check-prefix=NOMOVBE
+; RUN: llc -mtriple=x86_64-linux -mattr=+egpr,+ndd,+enable-ndd-mem,+movbe --show-mc-encoding < %s | FileCheck %s -check-prefix=EGPR
+; RUN: llc -mtriple=x86_64-linux -mattr=+egpr,+ndd,+enable-ndd-mem --show-mc-encoding < %s | FileCheck %s -check-prefix=NOMOVBE
----------------
phoebewang wrote:

Added one for line 4, they only have difference in `EVEX TO LEGACY Compression ...`. So we don't need new prefix after removing `--show-mc-encoding`.
For line 5, there are two tests have true differences, but they are `rolw` instructions, which is already covered in rol.ll

https://github.com/llvm/llvm-project/pull/186049