diff options
| author | Paul Walker <paul.walker@arm.com> | 2025-08-05 11:43:41 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-08-05 11:43:41 +0100 |
| commit | 94d374ab6cd559fa96e96bfb805b73a367bfda08 (patch) | |
| tree | 3f837d40c9af95120c069235e8ed36a3b2cde2a8 /llvm/lib/CodeGen/TargetLoweringBase.cpp | |
| parent | 11eeb4d133aea2c823231bc771bb58de6d8387e3 (diff) | |
| download | llvm-94d374ab6cd559fa96e96bfb805b73a367bfda08.zip llvm-94d374ab6cd559fa96e96bfb805b73a367bfda08.tar.gz llvm-94d374ab6cd559fa96e96bfb805b73a367bfda08.tar.bz2 | |
[LLVM][CGP] Allow finer control for sinking compares. (#151366)
Compare sinking is selectable based on the result of
hasMultipleConditionRegisters. This function is too coarse grained by
not taking into account the differences between scalar and vector
compares. This PR extends the interface to take an EVT to allow finer
control.
The new interface is used by AArch64 to disable sinking of scalable
vector compares, but with isProfitableToSinkOperands updated to maintain
the cases that are specifically tested.
Diffstat (limited to 'llvm/lib/CodeGen/TargetLoweringBase.cpp')
| -rw-r--r-- | llvm/lib/CodeGen/TargetLoweringBase.cpp | 1 |
1 files changed, 0 insertions, 1 deletions
diff --git a/llvm/lib/CodeGen/TargetLoweringBase.cpp b/llvm/lib/CodeGen/TargetLoweringBase.cpp index acc7fd2..9f525ea 100644 --- a/llvm/lib/CodeGen/TargetLoweringBase.cpp +++ b/llvm/lib/CodeGen/TargetLoweringBase.cpp @@ -697,7 +697,6 @@ TargetLoweringBase::TargetLoweringBase(const TargetMachine &tm) MaxGluedStoresPerMemcpy = 0; MaxStoresPerMemsetOptSize = MaxStoresPerMemcpyOptSize = MaxStoresPerMemmoveOptSize = MaxLoadsPerMemcmpOptSize = 4; - HasMultipleConditionRegisters = false; HasExtractBitsInsn = false; JumpIsExpensive = JumpIsExpensiveOverride; PredictableSelectIsExpensive = false; |