Instructions on the transcendental unit are executed in parallel to the
normal VALU, so add this as an extra resource.
This doesn't seem to have any effect, but it should be more correct.
Paths
| Differential D100123
[AMDGPU] Add TransVALU to gfx10 ClosedPublic Authored by sebastian-ne on Apr 8 2021, 9:09 AM.
Details
Diff Detail
Event TimelineComment Actions I have no objection to this as a first step in the right direction, but I think it should have no effect at all at the moment. The schedmodel already knows that we can only issue one instruction per cycle, and this patch does not affect that. It only becomes interesting when you model the fact that issuing a trans instruction consumes the trans ALU for (say) 4 cycles, so that on the following 3 cycles you could issue a normal VALU instruction but *not* another trans instruction. This revision is now accepted and ready to land.Apr 8 2021, 12:06 PM Closed by commit rG4897effb1415: [AMDGPU] Add TransVALU to gfx10 (authored by sebastian-ne). · Explain WhyApr 20 2021, 6:34 AM This revision was automatically updated to reflect the committed changes.
Revision Contents
Diff 338849 llvm/lib/Target/AMDGPU/SISchedule.td
llvm/test/tools/llvm-mca/AMDGPU/gfx10-add-sequence.s
llvm/test/tools/llvm-mca/AMDGPU/gfx10-double.s
llvm/test/tools/llvm-mca/AMDGPU/gfx10-trans.s
|