Alexey Bataev ac254fc055 [SLP]Improve tryToGatherExtractElements by using per-register analysis. ...

Currently tryToGatherExtractElements function analyzes the whole vector,
regrdless number of actual registers, used in this vector. It may
prevent some optimizations, because per-register analysis may allow to
simplify the final code by reusing more already emitted vectors and
better shuffles.

Differential Revision: https://reviews.llvm.org/D148855

2023-11-06 07:29:27 -08:00

..

32-bit.ll

[AARCH64] Enable STORE of v4i8 to help more vectorization opportunities

2023-03-31 17:03:06 +00:00

64-bit-vector.ll

…

accelerate-vector-functions-inseltpoison.ll

[IR] Change shufflevector undef mask to poison

2023-04-27 14:41:10 +01:00

accelerate-vector-functions.ll

[IR] Change shufflevector undef mask to poison

2023-04-27 14:41:10 +01:00

buildvector-reduce.ll

…

buildvector-vectorize.ll

…

commute.ll

…

ext-trunc.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

external-non-inst-use.ll

…

extractelements-to-shuffle.ll

[SLP]Improve tryToGatherExtractElements by using per-register analysis.

2023-11-06 07:29:27 -08:00

extracts-from-scalarizable-vector.ll

[SLP] Use vector types for cmp alt instructions costs

2023-06-28 21:02:29 +01:00

fmulladd.ll

…

fpsat.ll

[test] Replace aarch64-*-eabi with aarch64

2023-06-27 20:02:52 -07:00

fshl.ll

[SLP]Fix cost estimation for buildvectors with extracts and/or constants.

2023-04-19 05:54:58 -07:00

gather-cost.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

gather-load-128.ll

[AArch64][TTI] Improve LegalVF when gather loads are scalarized

2023-10-27 20:22:54 +02:00

gather-load-min-required-vf-2.ll

…

gather-reduce.ll

…

gather-root.ll

…

getelementptr.ll

[InstCombine] Infer nneg on zext when forming from non-negative sext (#70706)

2023-10-30 12:09:43 -07:00

horizontal.ll

…

insertelement-inseltpoison.ll

…

insertelement.ll

…

invalid_type.ll

…

landing_pad.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

lit.local.cfg

[NFC][Py Reformat] Reformat lit.local.cfg python files in llvm

2023-05-17 17:03:15 +02:00

load-store-q.ll

…

loadi8.ll

…

loadorder.ll

[SLP]Check scalars before trying scheduling.

2023-07-24 09:25:19 -07:00

matmul.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

memory-runtime-checks-in-loops.ll

…

memory-runtime-checks.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

minimum-sizes.ll

…

mismatched-intrinsics.ll

…

mul.ll

…

multiple_reduction.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

nontemporal.ll

…

phi-use-order-scalable.ll

[test] Replace aarch64-arm-none-eabi with aarch64

2023-06-27 19:36:27 -07:00

PR38339.ll

…

reduce-add-i64.ll

[opt] Infer DataLayout from triple if not specified

2023-10-26 12:07:37 -07:00

remarks.ll

Recommit "[AArch64][TTI] Cost model FADD/FSUB/FNEG""

2023-04-11 12:48:15 +01:00

reorder-fmuladd-crash.ll

…

scalable-vector.ll

…

scalarization-overhead.ll

…

sdiv-pow2.ll

…

shuffle-vectors-mask-size.ll

[COST][AARCH64]Fix crash in cost calculation for shuffles.

2023-10-02 07:49:03 -07:00

slp-abs.ll

…

slp-and-reduction.ll

…

slp-fma-loss.ll

[AArch64] Change the cost of vector insert/extract to 2

2023-07-28 21:26:50 +01:00

slp-or-reduction.ll

…

slp-xor-reduction.ll

…

spillcost-di.ll

…

spillcost-order.ll

…

splat-loads.ll

[opt] Infer DataLayout from triple if not specified

2023-10-26 12:07:37 -07:00

store-ptr.ll

…

transpose-inseltpoison.ll

[SLP]Do not early exit if the number of unique elements is non-power-of-2. (#65476)

2023-09-08 10:00:46 -04:00

transpose.ll

[SLP]Do not early exit if the number of unique elements is non-power-of-2. (#65476)

2023-09-08 10:00:46 -04:00

trunc-insertion.ll

…

tsc-s116.ll

[IR] Change shufflevector undef mask to poison

2023-04-27 14:41:10 +01:00

tsc-s352.ll

…

vector-getelementptr.ll

…

vectorizable-selects-min-max.ll

…

vectorizable-selects-uniform-cmps.ll

[SLP]Do not early exit if the number of unique elements is non-power-of-2. (#65476)

2023-09-08 10:00:46 -04:00

vectorize-free-extracts-inserts.ll

[SLP]Improve canreuseExtracts for reordering analysis.

2023-09-15 12:09:45 -07:00

widen.ll

…