projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 9df0ba5) | patch
[LoopVectorize] Change getRuntimeVFAsFloat to use unsigned int->FP conversion
authorDavid Sherwood <david.sherwood@arm.com>
Wed, 27 Oct 2021 10:40:01 +0000 (11:40 +0100)
committerDavid Sherwood <david.sherwood@arm.com>
Mon, 1 Nov 2021 09:58:14 +0000 (09:58 +0000)
commit87a294d5eb5a0638d9b5887f0a620e7435f71415
treeedc98117d79a6f67f8a435c38b80c0e6bb715a27tree | snapshot
parent9df0ba599c3c777a37557ddcc567c85dae4c33d3commit | diff
[LoopVectorize] Change getRuntimeVFAsFloat to use unsigned int->FP conversion

We never expect the runtime VF to be negative so we should use
the uitofp instruction instead of sitofp.

Differential revision: https://reviews.llvm.org/D112610
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp diff | blob | history
llvm/test/Transforms/LoopVectorize/scalable-inductions.ll diff | blob | history
Domain: System / Toolchain;