This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Test disassembling of some basic True16 VOP2 instructions.
ClosedPublic

Authored by kosarev on Aug 1 2023, 2:46 AM.

Diff Detail

Event Timeline

kosarev created this revision.Aug 1 2023, 2:46 AM
Herald added a project: Restricted Project. · View Herald TranscriptAug 1 2023, 2:46 AM
kosarev requested review of this revision.Aug 1 2023, 2:46 AM
Herald added a project: Restricted Project. · View Herald TranscriptAug 1 2023, 2:46 AM
Joe_Nash added inline comments.Aug 1 2023, 7:43 AM
llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop2.txt
2

I think we should use the same naming paradigm and workflow for CHECK names as in the codegen tests. To match codegen, it would make -mattr=+real-true16 tests are called 'GFX11'. There is an argument to do the naming the other way, where '-mattr=-real-true16' continues to be GFX11. Both are fine from my PoV; perhaps the choice should me made to minimize merge pain.

kosarev added inline comments.Aug 1 2023, 8:15 AM
llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop2.txt
2

To match codegen, it would make -mattr=+real-true16 tests are called 'GFX11'.

That was my aim, but in this particular test GFX11 is used as a common prefix.

Is there any test that verifies that -mattr=-real-true16 is the default? If so, LGTM

Is there any test that verifies that -mattr=-real-true16 is the default? If so, LGTM

All tests that don't use this flag rely on that be the default and wouldn't pass otherwise, so I assume no special tests needed.

Joe_Nash accepted this revision.Aug 22 2023, 6:20 AM
This revision is now accepted and ready to land.Aug 22 2023, 6:20 AM
This revision was landed with ongoing or failed builds.Sep 25 2023, 11:53 AM
This revision was automatically updated to reflect the committed changes.