David Sherwood d581d94385 [SVE] Fix the FP arithmetic instruction costs for SVE ...

Several FP instructions (fadd, fsub, etc.) were incorrectly assigned
a higher cost for SVE because they have custom lowering, however we
know they are legal. This patch explicitly assigns a cost of 2 to
these opcodes.

Tests added here:

  Analysis/CostModel/AArch64/arith-fp-sve.ll

Differential Revision: https://reviews.llvm.org/D108993

2021-09-02 09:55:13 +01:00

..

abs.ll

…

aggregates.ll

…

arith-fp-sve.ll

[SVE] Fix the FP arithmetic instruction costs for SVE

2021-09-02 09:55:13 +01:00

arith-fp.ll

[AArch64] Use custom lowering for fp16 vector copysign.

2021-07-02 11:15:30 +01:00

arith-ssat.ll

…

arith-usat.ll

…

bitreverse.ll

[AArch64] Add cost tests for bitreverse

2021-06-10 14:51:33 +01:00

bswap.ll

…

cast.ll

[AArch64] Cost-model i8 vector loads/stores

2021-07-05 11:25:10 +01:00

cmp.ll

…

cost-scalable-vector-gep.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

ctlz.ll

[CostModel][AArch64] Add tests for ctlz, ctpop and cttz. NFC.

2021-06-03 17:12:22 +01:00

ctpop.ll

[CostModel][AArch64] Improve the cost estimate of CTPOP intrinsic

2021-06-11 11:15:46 +01:00

cttz.ll

[CostModel][AArch64] Add tests for ctlz, ctpop and cttz. NFC.

2021-06-03 17:12:22 +01:00

div_cte.ll

…

div.ll

…

free-widening-casts.ll

…

gep.ll

…

getIntrinsicInstrCost-vector-reverse.ll

[COST][AARCH64] Improve cost of reverse shuffles for AArch64.

2021-04-20 13:47:56 -07:00

kryo-inseltpoison.ll

…

kryo.ll

…

lit.local.cfg

…

logicalop.ll

…

masked_ldst_vls.ll

[AArch64][SVE] Teach cost model that masked loads/stores are cheap

2021-08-19 13:01:33 +01:00

masked_ldst.ll

[CostModel][AArch64] Make loads/stores of <vscale x 1 x eltty> invalid.

2021-07-14 16:44:22 +01:00

mem-op-cost-model.ll

[AArch64][SVE] Teach cost model masked gathers/scatters are cheap

2021-08-26 11:17:47 +01:00

min-max.ll

[AArch64] Optimise min/max lowering in ISel

2021-08-02 13:40:21 +01:00

mul.ll

[CostModel][AArch64] Add floating point arithmetic tests. NFC.

2021-05-26 20:26:20 +01:00

neon-stepvector.ll

…

reduce-add.ll

[AArch64] Adjust the cost of integer sum reductions

2021-07-22 18:19:54 +01:00

reduce-and.ll

[AArch64] Add worst case shuffle costs

2021-07-23 09:01:58 +01:00

reduce-fadd.ll

[Analysis][AArch64] Make fixed-width ordered reductions slightly more expensive

2021-08-18 17:01:56 +01:00

reduce-minmax.ll

[AArch64] Expand the SVE min/max reduction costs to NEON

2021-08-05 23:23:24 +01:00

reduce-or.ll

[AArch64] Add worst case shuffle costs

2021-07-23 09:01:58 +01:00

reduce-xor.ll

[AArch64] Add worst case shuffle costs

2021-07-23 09:01:58 +01:00

select.ll

…

shuffle-broadcast.ll

…

shuffle-other.ll

[CostModel][AArch64] Add some shuffle concat tests. NFC.

2021-08-02 12:11:00 +01:00

shuffle-reverse.ll

[COST][AARCH64] Improve cost of reverse shuffles for AArch64.

2021-04-20 13:47:56 -07:00

shuffle-select.ll

…

shuffle-transpose.ll

…

splice.ll

[AArch64][CostModel] Add cost model for experimental.vector.splice

2021-07-05 14:30:24 +01:00

store.ll

[AArch64] Cost-model i8 vector loads/stores

2021-07-05 11:25:10 +01:00

sve-bitcast.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

sve-cmpsel.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

sve-fixed-length.ll

…

sve-fpext.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

sve-fptoi.ll

[BasicTTI] Set scalarization cost of scalable vector casts to Invalid.

2021-07-24 14:13:21 +01:00

sve-fptrunc.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

sve-gather.ll

[SVE] Remove usage of getMaxVScale for AArch64, in favour of IR Attribute

2021-08-17 14:42:47 +01:00

sve-illegal-types.ll

[CostModel] Return an invalid cost for memory ops with unsupported types

2021-06-08 12:07:36 +01:00

sve-intrinsics.ll

[SVE] Fix the FP arithmetic instruction costs for SVE

2021-09-02 09:55:13 +01:00

sve-ldst.ll

[CostModel][AArch64] Make loads/stores of <vscale x 1 x eltty> invalid.

2021-07-14 16:44:22 +01:00

sve-math.ll

[SVE] Fix the FP arithmetic instruction costs for SVE

2021-09-02 09:55:13 +01:00

sve-remainder.ll

[CostModel] Express cost(urem) as cost(div+mul+sub) when set to Expand.

2021-07-07 14:40:28 +01:00

sve-scatter.ll

[SVE] Remove usage of getMaxVScale for AArch64, in favour of IR Attribute

2021-08-17 14:42:47 +01:00

sve-shuffle-broadcast.ll

[CostModel][AArch64] Add missing costs for getShuffleCost with scalable vectors

2021-05-20 09:08:31 +01:00

sve-stepvector.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

sve-trunc.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

sve-widening-instruction.ll

[SVE] Remove checks for warnings in scalable-vector tests.

2021-04-07 15:59:32 +01:00

vector-reduce.ll

[AArch64] Expand the SVE min/max reduction costs to NEON

2021-08-05 23:23:24 +01:00

vector-select.ll

…