[MergeICmps] MergeICmps is a new optimization pass that turns chains of integer

comparisons into memcmp.

Thanks to recent improvements in the LLVM codegen, the memcmp is typically
inlined as a chain of efficient hardware comparisons.
This typically benefits C++ member or nonmember operator==().

For now this is disabled by default until:
 - https://bugs.llvm.org/show_bug.cgi?id=33329 is complete
 - Benchmarks show that this is always useful.

Differential Revision:
https://reviews.llvm.org/D33987

llvm-svn: 312315

1 files changed, 1 insertions, 0 deletions

diff --git a/llvm/lib/Transforms/Scalar/Scalar.cpp b/llvm/lib/Transforms/Scalar/Scalar.cpp
index ce6f93eb0c1..d41fe6a3ba8 100644
--- a/llvm/lib/Transforms/Scalar/Scalar.cpp
+++ b/llvm/lib/Transforms/Scalar/Scalar.cpp
@@ -73,6 +73,7 @@ void llvm::initializeScalarOpts(PassRegistry &Registry) {
   initializeLowerExpectIntrinsicPass(Registry);
   initializeLowerGuardIntrinsicLegacyPassPass(Registry);
   initializeMemCpyOptLegacyPassPass(Registry);
+  initializeMergeICmpsPass(Registry);
   initializeMergedLoadStoreMotionLegacyPassPass(Registry);
   initializeNaryReassociateLegacyPassPass(Registry);
   initializePartiallyInlineLibCallsLegacyPassPass(Registry);