HomePhabricator

[ARM] Avoid pointless vrev of element-wise vmov

Authored by john.brawn on Mar 17 2020, 10:58 AM.

Description

[ARM] Avoid pointless vrev of element-wise vmov

If we have an element-wise vmov immediate instruction then a subsequent vrev
with width greater or equal to the vmov element width, then that vrev won't do
anything. Add a DAG combine to convert bitcasts that would become such vrevs
into vector_reg_casts instead.

Differential Revision: https://reviews.llvm.org/D76514

Details

Committed
john.brawnApr 3 2020, 9:36 AM
Differential Revision
D76514: [ARM] Avoid pointless vrev of element-wise vmov
Parents
rG966ae762229f: Run update_llc_test on test/CodeGen/ARM/vmov.ll
Branches
Unknown
Tags
Unknown