diff options
author | Sjoerd Meijer <sjoerd.meijer@arm.com> | 2021-04-06 08:53:42 +0100 |
---|---|---|
committer | Sjoerd Meijer <sjoerd.meijer@arm.com> | 2021-04-06 09:47:50 +0100 |
commit | d5f1131c812df57560c7563475cb0d674a101636 (patch) | |
tree | 971e55903cc9d24c31d5f118e01fba20dfe10f44 /clang/lib/Frontend/CompilerInstance.cpp | |
parent | 2935737da32dd21bf02e1a3c912a340de660d557 (diff) | |
download | llvm-d5f1131c812df57560c7563475cb0d674a101636.zip llvm-d5f1131c812df57560c7563475cb0d674a101636.tar.gz llvm-d5f1131c812df57560c7563475cb0d674a101636.tar.bz2 |
[AArch64] Default to zero-cycle-zeroing FP registers
It is generally beneficial to prefer "movi d0, #0" over "fmov s0, wzr" as this
is most efficient across all cores; it is recognised as a zeroing idiom. For
newer cores, fmov instructions can also be eliminated early and there is no
difference with movi, but some implementations lack this so is not true for
other/older cores. Thus this standardises on using movi as this should always
gives the same or better performance than the fmov with wzr.
Differential Revision: https://reviews.llvm.org/D99586
Diffstat (limited to 'clang/lib/Frontend/CompilerInstance.cpp')
0 files changed, 0 insertions, 0 deletions