[PATCH] D113778: [AMDGPU] Use shift for b64 mov
Stanislav Mekhanoshin via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Sun Nov 14 23:00:27 PST 2021
rampitec added a comment.
In D113778#3130490 <https://reviews.llvm.org/D113778#3130490>, @critson wrote:
> For GFX10, I don't think this is worth doing unless V_LSHLREV_B64 is full rate.
> 2x V_MOV_B32 in VOP1 takes the same space as V_LSHLREV_B64 in VOP3.
I think this is right. It can also be scheduled apart leaving room for something else to be scheduled in between. A 64 bit shift is rarely beneficial in general if you can get away without it.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D113778/new/
https://reviews.llvm.org/D113778
More information about the llvm-commits
mailing list