summaryrefslogtreecommitdiff
path: root/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'lib/Target/AArch64/AArch64TargetTransformInfo.cpp')
-rw-r--r--lib/Target/AArch64/AArch64TargetTransformInfo.cpp7
1 files changed, 5 insertions, 2 deletions
diff --git a/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index 88c98865bbc6..1a17691fc584 100644
--- a/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -417,14 +417,17 @@ int AArch64TTIImpl::getArithmeticInstrCost(
}
}
-int AArch64TTIImpl::getAddressComputationCost(Type *Ty, bool IsComplex) {
+int AArch64TTIImpl::getAddressComputationCost(Type *Ty, ScalarEvolution *SE,
+ const SCEV *Ptr) {
// Address computations in vectorized code with non-consecutive addresses will
// likely result in more instructions compared to scalar code where the
// computation can more often be merged into the index mode. The resulting
// extra micro-ops can significantly decrease throughput.
unsigned NumVectorInstToHideOverhead = 10;
+ int MaxMergeDistance = 64;
- if (Ty->isVectorTy() && IsComplex)
+ if (Ty->isVectorTy() && SE &&
+ !BaseT::isConstantStridedAccessLessThan(SE, Ptr, MaxMergeDistance + 1))
return NumVectorInstToHideOverhead;
// In many cases the address computation is not merged into the instruction