projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 4989779) | patch
[InstCombine] Extend SVEVectorFuseMulAddSub to support newly added "undef" intrinsics.
authorPaul Walker <paul.walker@arm.com>
Mon, 20 Feb 2023 17:52:54 +0000 (17:52 +0000)
committerPaul Walker <paul.walker@arm.com>
Sun, 12 Mar 2023 10:39:24 +0000 (10:39 +0000)
commit2f887c9a760dfdffa584ce84361912fe122ad79f
tree843f31797c8941b0f8f668d1f5de361c53f85bb5tree | snapshot
parent4989779d7a412a19bab168a1281f6f33edc40870commit | diff
[InstCombine] Extend SVEVectorFuseMulAddSub to support newly added "undef" intrinsics.

D143767 will change the intrinsics used to lower floating-point
svadd_x, svmul_x and svsub_x builtins. This will result in the
combines added as part of D140200 to no longer fire in all cases.
This patch extends the existing combines for contraction to cover
fadd_u, fmul_u and fsub_u intrinsics.

Differential Revision: https://reviews.llvm.org/D144413
llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp diff | blob | history
llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-muladdsub.ll diff | blob | history
Domain: System / Toolchain;