[llvm] [X86] Prefer 128-bit vectors for Bulldozer CPUs (PR #225838)

[Simon Pilgrim via llvm-commits]

================
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt -mcpu=bdver1 -passes=loop-vectorize -S %s | FileCheck %s --check-prefix=CHECK
-; RUN: opt -mcpu=bdver1 -passes=loop-vectorize -debug-only=loop-vectorize -disable-output %s 2>&1 | FileCheck %s --check-prefix=CHECK-COST
+; RUN: opt -mcpu=bdver1 -mattr=-prefer-128-bit -passes=loop-vectorize -S %s | FileCheck %s --check-prefix=CHECK
----------------
RKSimon wrote:

unacceptable to add the mattr hack - these need to test what the cpu will actually generate - some of these might benefit from moving to sandybridge instead though (assuming the test doesn't touch XOP/FMA4/TBM)?

https://github.com/llvm/llvm-project/pull/225838