[PATCH] [ARM] Prevent PerformVCVTCombine from combining a vmul/vcvt with 8 lanes
Hao Liu
Hao.Liu at arm.com
Tue Dec 16 01:22:57 PST 2014
Hi Bradley,
AArch64 backend also has such instructions (It even support double type), but it doesn't have such combine function to do such optimization. I think maybe we can also port this function to AArch64 backend.
Thanks,
-Hao
REPOSITORY
rL LLVM
http://reviews.llvm.org/D6657
EMAIL PREFERENCES
http://reviews.llvm.org/settings/panel/emailpreferences/
More information about the llvm-commits
mailing list