projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: cd73ce4) | patch
[AMDGPU] Tweak VOP3_INTERP16 profile
authorJay Foad <jay.foad@amd.com>
Fri, 14 May 2021 12:38:47 +0000 (13:38 +0100)
committerJay Foad <jay.foad@amd.com>
Mon, 17 May 2021 14:28:00 +0000 (15:28 +0100)
commit472f856714fb1687b0727b3013f5d071d8fa86ae
tree73711af2403c5e3c217f204669d64126cc785bb5tree | snapshot
parentcd73ce4b5e622bb4d71b41f9dab157200501d352commit | diff
[AMDGPU] Tweak VOP3_INTERP16 profile

Set the output register class based on the output type, instead of
hard-coding VGPR_32. I think this is more correct. It doesn't make any
difference at the moment because we use the same class for 16- and
32-bit results, but it might in future if we make more use of true
16-bit register classes.

Differential Revision: https://reviews.llvm.org/D102622
llvm/lib/Target/AMDGPU/VOP3Instructions.td diff | blob | history
Domain: System / Toolchain;