Simon Pilgrim 9bb1d0369c

[X86] getShuffleCost - when splitting shuffles, if a whole vector source is just copied we should treat this as free. (#120561) ...

If the shuffle split results in referencing a single legalised whole vector (i.e. no permutation), then this can be treated as free.

We already do something similar for broadcasts / whole subvector insertion + extraction - its purely an issue for register allocation.

2024-12-19 12:55:44 +00:00

..

AArch64

[AArch64] Tweak truncate costs for some scalable vector types (#119542)

2024-12-19 10:07:41 +00:00

AMDGPU

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548)

2024-11-06 11:53:33 +00:00

ARM

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548)

2024-11-06 11:53:33 +00:00

PowerPC

[PowerPC] Update data layout aligment of i128 to 16 (#118004)

2024-12-09 18:02:24 -05:00

RISCV

[RISCV][TTI] Model the cost of insert/extractelt when the vector split into multiple register group and idx exceed single group. (#118401)

2024-12-12 09:26:19 +08:00

SystemZ

[SystemZ] Add realistic cost estimates for vector reduction intrinsics (#118319)

2024-12-03 17:08:51 +01:00

X86

[X86] getShuffleCost - when splitting shuffles, if a whole vector source is just copied we should treat this as free. (#120561)

2024-12-19 12:55:44 +00:00

free-intrinsics-datalayout.ll

[CostModel] No cost for llvm.allow.{runtime,ubsan}.check() (#86064)

2024-03-31 22:27:57 -07:00

free-intrinsics-no_info.ll

[CostModel] No cost for llvm.allow.{runtime,ubsan}.check() (#86064)

2024-03-31 22:27:57 -07:00

no_info.ll

[CostModel] Convert some tests to opaque pointers (NFC)

2022-12-15 09:50:34 +01:00