David Green 8675ef100f [LV] Logical and/or select costs ... D99674 stopped the folding of certain select operations into and/or, due to incorrect folding in the presence of poison. D97360 added some costs to attempt to account for the change, but only worked at the getUserCost level, not the getCmpSelInstrCost that the vectorizer will use directly. This adds similar logic into the vectorizer to handle these logical and/or selects, treating them like and/or directly. This fixes 60% performance regressions from code like the attached test case. Differential Revision: https://reviews.llvm.org/D99884		2021-04-08 10:39:47 +01:00
..
arm-ieee-vectorize.ll	…
arm-unroll.ll	…
gather-cost.ll	…
gcc-examples.ll	…
interleaved_cost.ll	…
lit.local.cfg	…
mve-gather-scatter-tailpred.ll	[LoopVectorize][ARM] Regenerate mve-gather-scatter-tailpred.ll test	2021-02-05 12:32:45 +00:00
mve-icmpcost.ll	[ARM] Tone down the MVE scalarization overhead	2021-03-19 18:30:11 +00:00
mve-interleaved-cost.ll	[ARM] Tone down the MVE scalarization overhead	2021-03-19 18:30:11 +00:00
mve-maskedldst.ll	…
mve-predstorecost.ll	[ARM] Test for predicated scalar memops. NFC	2021-03-09 21:57:18 +00:00
mve-qabs.ll	[SimplifyCFG] TryToSimplifyUncondBranchFromEmptyBlock() already knows how to preserve DomTree	2020-12-17 01:03:49 +03:00
mve-reduction-predselect.ll	[SLP,LV] Use poison constant vector for shufflevector/initial insertelement	2021-01-06 11:22:50 +09:00
mve-reduction-types.ll	[LV][ARM] Inloop reduction cost modelling	2021-01-21 21:03:41 +00:00
mve-reductions.ll	[LV][ARM] Inloop reduction cost modelling	2021-01-21 21:03:41 +00:00
mve-saddsatcost.ll	[ARM] Tone down the MVE scalarization overhead	2021-03-19 18:30:11 +00:00
mve-selectandorcost.ll	[LV] Logical and/or select costs	2021-04-08 10:39:47 +01:00
mve-shiftcost.ll	…
mve-vldn.ll	…
pointer_iv.ll	[ARM] Tone down the MVE scalarization overhead	2021-03-19 18:30:11 +00:00
prefer-tail-loop-folding.ll	[LoopVectorize] Remove unused check-prefixes	2020-11-09 12:18:20 +00:00
scalar-block-cost.ll	[LV] Fix scalar cost for tail predicated loops	2020-12-12 14:21:40 +00:00
sphinx.ll	[Vectorizers][TTI] remove option to bypass creation of vector reduction intrinsics	2021-02-12 08:13:50 -05:00
tail-fold-multiple-icmps.ll	[SimplifyCFG] TryToSimplifyUncondBranchFromEmptyBlock() already knows how to preserve DomTree	2020-12-17 01:03:49 +03:00
tail-folding-allowed.ll	…
tail-folding-counting-down.ll	[LV] Fallback strategies if tail-folding fails	2020-08-26 16:55:25 +01:00
tail-folding-loop-hint.ll	[LV] get.active.lane.mask consuming tripcount instead of backedge-taken count	2020-08-25 13:49:19 +01:00
tail-folding-not-allowed.ll	[ARM] Tone down the MVE scalarization overhead	2021-03-19 18:30:11 +00:00
tail-folding-prefer-flag.ll	[LV] Fallback strategies if tail-folding fails	2020-08-26 16:55:25 +01:00
tail-folding-reduces-vf.ll	[ARM][MVE] Enable tail-predication by default	2020-09-28 14:01:23 +01:00
tail-folding-reductions-allowed.ll	…
tail-folding-scalar-epilogue-fallback.ll	[LV] Fallback strategies if tail-folding fails	2020-08-26 16:55:25 +01:00
vector_cast.ll	…
width-detect.ll	…