David Green f0e79d9152 [AArch64] Add a cost for identity shuffles. ...

These are mostly handled at a higher level when costing shuffles, but some
masks can end up being identity or concat masks which we can treat as free.

2024-04-09 17:16:14 +01:00

..

abs.ll

…

aggregates.ll

…

arith-fp-frem.ll

[AArch64][CostModel] Improve scalar frem cost (#80423)

2024-02-23 09:29:45 +00:00

arith-fp-sve.ll

…

arith-fp.ll

[AArch64][CostModel] Improve scalar frem cost (#80423)

2024-02-23 09:29:45 +00:00

arith-overflow.ll

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

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

arith-ssat.ll

…

arith-usat.ll

…

arith-widening.ll

…

arith.ll

…

bitreverse.ll

…

bswap.ll

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

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

cast.ll

[test] Replace aarch64-*-{eabi,gnueabi}{,hf} with aarch64

2024-02-12 18:29:55 -08:00

cmp.ll

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

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

cost-scalable-vector-gep.ll

[SVE] Fix crash when costing getelementptr with scalable target type.

2023-09-18 12:48:30 +00:00

ctlz.ll

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

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

ctpop.ll

…

cttz.ll

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

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

div_cte.ll

…

div.ll

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

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

ext-rhadd.ll

[LLVM][test] Convert remaining instances of ConstantExpr based splats to use splat().

2024-02-27 13:37:23 +00:00

fptoi_sat.ll

[test] Replace aarch64-*-{eabi,gnueabi}{,hf} with aarch64

2024-02-12 18:29:55 -08:00

free-widening-casts.ll

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

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

fshl.ll

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

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

fshr.ll

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

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

gep.ll

…

getIntrinsicInstrCost-vector-reverse.ll

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

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

insert-extract.ll

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

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

kryo-inseltpoison.ll

…

lit.local.cfg

…

load-to-trunc.ll

…

logicalop.ll

…

masked_ldst_vls.ll

…

masked_ldst.ll

[TTI] getCommonMaskedMemoryOpCost - consistently use getScalarizationOverhead instead of ExtractElement costs for address/mask extraction. (#87771)

2024-04-09 15:42:51 +01:00

mem-op-cost-model.ll

[TTI] getCommonMaskedMemoryOpCost - consistently use getScalarizationOverhead instead of ExtractElement costs for address/mask extraction. (#87771)

2024-04-09 15:42:51 +01:00

min-max.ll

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

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

mul.ll

…

neon-stepvector.ll

…

no-sve-no-neon.ll

[AArch64] Fix crash when neither Neon nor SVE are enabled

2023-08-17 16:07:27 +01:00

reduce-add.ll

…

reduce-and.ll

…

reduce-fadd.ll

[AArch64] Add more complete support for BF16

2024-03-03 22:39:50 +00:00

reduce-minmax.ll

[TTI]Fallback to SingleSrcPermute shuffle kind, if no direct estimation for

2024-02-12 07:09:49 -05:00

reduce-or.ll

…

reduce-xor.ll

…

rem.ll

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

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

select.ll

…

shuffle-broadcast.ll

…

shuffle-load.ll

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

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

shuffle-other.ll

[AArch64] Add a cost for identity shuffles.

2024-04-09 17:16:14 +01:00

shuffle-reverse.ll

…

shuffle-select.ll

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

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

shuffle-store.ll

[AArch64] Add costs for ST3 and ST4 instructions, modelled as store(shuffle). (#87934)

2024-04-09 16:36:08 +01:00

shuffle-transpose.ll

…

splice.ll

…

store-ptr.ll

…

store.ll

…

sve-arith.ll

…

sve-bitcast.ll

…

sve-cmpsel.ll

…

sve-ext.ll

[Analysis][SVE] Improve cost model for some extending masked loads (#65957)

2023-10-02 10:50:56 +01:00

sve-fixed-length.ll

…

sve-fpext.ll

…

sve-fptoi.ll

…

sve-fptrunc.ll

[opt] Infer DataLayout from triple if not specified

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

sve-gather.ll

[TTI] getCommonMaskedMemoryOpCost - consistently use getScalarizationOverhead instead of ExtractElement costs for address/mask extraction. (#87771)

2024-04-09 15:42:51 +01:00

sve-illegal-types.ll

…

sve-insert-extract.ll

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

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

sve-intrinsics.ll

[TTI] getCommonMaskedMemoryOpCost - consistently use getScalarizationOverhead instead of ExtractElement costs for address/mask extraction. (#87771)

2024-04-09 15:42:51 +01:00

sve-ldst.ll

[Analysis][SVE] Improve cost model for some extending masked loads (#65957)

2023-10-02 10:50:56 +01:00

sve-math.ll

…

sve-min-max.ll

…

sve-remainder.ll

…

sve-scatter.ll

…

sve-shuffle-broadcast.ll

…

sve-stepvector.ll

…

sve-trunc.ll

[opt] Infer DataLayout from triple if not specified

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

sve-vscale.ll

…

sve-widening-instruction.ll

…

vec3-ops.ll

[AArch64] Improve cost computations for odd vector mem ops. (#78181)

2024-01-17 21:32:06 +00:00

vector-reduce.ll

…

vector-select.ll

[DAG] narrowExtractedVectorBinOp - ensure we limit late node creation to LegalOperations only (#72130)

2023-11-20 10:56:41 +00:00