[PATCH] D64899: AMDGPU/GlobalISel: First pass at attempting to legalize load/stores
Nicolai Hähnle via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue Jul 23 02:32:58 PDT 2019
nhaehnle added a comment.
Just to clarify, how is selection of global_load_ubyte and friends going to work? I assume similar to today where the load returns an s32 value, but instruction selection does matching based on the MemOperand remembering the size?
Why are unaligned global loads split up on CI+? I see that you're trying to handle this in the code, but apparently it doesn't work correctly?
================
Comment at: lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:549-550
+ case AMDGPUAS::CONSTANT_ADDRESS:
+ case AMDGPUAS::GLOBAL_ADDRESS:
+ return 512;
+ default:
----------------
Shouldn't the max size for global be 128? It only goes up to dwordx4.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D64899/new/
https://reviews.llvm.org/D64899
More information about the llvm-commits
mailing list