 Teresa Johnson
|
ef512ca8e6
Change TargetLibraryInfo analysis passes to always require Function
|
6 years ago |
 Sjoerd Meijer
|
ba9e6a642f
[LV] Tail-folding, runtime scev checks
|
6 years ago |
|
Sjoerd Meijer
|
8103f963ba
[LV] Tail-folding with runtime memory checks
|
6 years ago |
 Ayal Zaks
|
cbb4ef7bbf
[LV] Fold tail by masking - handle reductions
|
6 years ago |
 Jonas Devlieghere
|
114087caa6
[llvm] Migrate llvm::make_unique to std::make_unique
|
6 years ago |
 Dorit Nuzman
|
ecfc353229
[LV] fold-tail predication should be respected even with assume_safety
|
6 years ago |
|
Dorit Nuzman
|
6c9a51ee41
[LV] Fold-tail flag
|
6 years ago |
 Craig Topper
|
5a370c1ad8
[LoopVectorize][X86] Clamp interleave factor if we have a known constant trip count that is less than VF*interleave
|
6 years ago |
 Mitch Phillips
|
7e7625f30a
Revert "[X86] Add more extract subvector cost model tests for smaller element sizes and smaller than 128-bit vectors."
|
6 years ago |
|
Craig Topper
|
49b912b85e
[X86] Add more extract subvector cost model tests for smaller element sizes and smaller than 128-bit vectors.
|
6 years ago |
 Hideki Saito
|
057a373e1c
[LV][NFC] Share the LV illegality reporting with LoopVectorize.
|
6 years ago |
|
Sjoerd Meijer
|
ad60f4fbf4
Follow up of rL367592, fix the build
|
6 years ago |
|
Sjoerd Meijer
|
d527e64640
[LV] Tail-Loop Folding
|
6 years ago |
|
Sjoerd Meijer
|
72855fc09a
[LV] Scalar Epilogue Lowering. NFC.
|
6 years ago |
 Florian Hahn
|
bbc7752ec4
[LoopVectorize] Pass unfiltered list of arguments to getIntrinsicInstCost.
|
6 years ago |
|
Florian Hahn
|
3748d783e0
[LV] Exclude loop-invariant inputs from scalar cost computation.
|
6 years ago |
 Orlando Cazalet-Hyams
|
b1ba8a93bc
[DebugInfo@O2][LoopVectorize] pr39024: Vectorized code linenos step through loop even after completion
|
6 years ago |
 Warren Ristow
|
31868b92df
[LV] Suppress vectorization in some nontemporal cases
|
6 years ago |
 Whitney Tsang
|
bdd7b78551
PHINode: introduce setIncomingValueForBlock() function, and use it.
|
6 years ago |
 Bjorn Pettersson
|
784000b4c2
[LV] Deny irregular types in interleavedAccessCanBeWidened
|
6 years ago |
|
Orlando Cazalet-Hyams
|
f36c1c031e
Revert "[DebugInfo@O2][LoopVectorize] pr39024: Vectorized code linenos step through loop even after completion"
|
6 years ago |
|
Orlando Cazalet-Hyams
|
6c60c71a4e
[DebugInfo@O2][LoopVectorize] pr39024: Vectorized code linenos step through loop even after completion
|
6 years ago |
|
Florian Hahn
|
6a170699ae
[LV] Remove the redundant using LoopVectorizationPlanner:VPlanPtr
|
6 years ago |
|
Craig Topper
|
8c0b3da9d4
[LoopVectorize] Add FNeg instruction support
|
6 years ago |
 Simon Pilgrim
|
79e92fbf63
LoopVectorizationCostModel::selectInterleaveCount - assert we have a non-zero loop cost. NFCI.
|
6 years ago |
|
Florian Hahn
|
75b2eadb06
[LV] Move getScalarizationOverhead and vector call cost computations to CM. (NFC)
|
6 years ago |
 Alina Sbirlea
|
c419d3f938
[MemorySSA] Teach LoopSimplify to preserve MemorySSA.
|
6 years ago |
 Kostya Serebryany
|
53dcce4331
revert r360162 as it breaks most of the buildbots
|
6 years ago |
|
Orlando Cazalet-Hyams
|
8c101025dc
[DebugInfo@O2][LoopVectorize] pr39024: Vectorized code linenos step through loop even after completion
|
6 years ago |
|
Alina Sbirlea
|
6ac87016ff
Enable LoopVectorization by default.
|
6 years ago |
