projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 002005e) | patch
[RISCV] Add fmin/fmax scalar instructions to isAssociativeAndCommutative
authorHsiangkai Wang <hsiangkai@google.com>
Fri, 23 Dec 2022 03:05:10 +0000 (03:05 +0000)
committerHsiangkai Wang <hsiangkai@google.com>
Thu, 29 Dec 2022 11:43:40 +0000 (11:43 +0000)
commitaf5dd2706c845b8bea6ba9fd3e68c1ee982b01fa
tree8ce298c75498306836d14b4ec8badf3200b7c86etree | snapshot
parent002005e6740e6de8f748ed234db08625cc533195commit | diff
[RISCV] Add fmin/fmax scalar instructions to isAssociativeAndCommutative

Follow-up patch of D140530.

We can add FMIN, FMAX to isAssociativeAndCommutative to
increase instruction-level parallelism by the existing MachineCombiner
pass.

Differential Revision: https://reviews.llvm.org/D140602
llvm/lib/Target/RISCV/RISCVInstrInfo.cpp diff | blob | history
llvm/test/CodeGen/RISCV/machine-combiner.ll diff | blob | history
Domain: System / Toolchain;