[PATCH] D31030: [AMDGPU] Fix bug in S_MOVREL check (PR32248)

Matt Arsenault via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Mar 16 14:22:24 PDT 2017


arsenm added inline comments.


================
Comment at: test/CodeGen/AMDGPU/inserted-wait-states.mir:456-471
-# GFX9: S_NOP
 # GCN-NEXT: V_INTERP_P1_F32
 
 # GCN-LABEL: bb.1:
 # GCN: S_MOV_B32
-# GFX9: S_NOP
 # GCN-NEXT: V_INTERP_P2_F32
----------------
RKSimon wrote:
> arsenm wrote:
> > tstellar wrote:
> > > Aren't these nops are still required.  Have you checked the GFX9 ISA guide?
> > I don't think it's been published yet. I think these are required though
> In which case the correct codegen is relying on the always true result from isSMovRel.....
I think I found the problem, there is another place that is supposed to check the same condition. I'll commit this soon


Repository:
  rL LLVM

https://reviews.llvm.org/D31030





More information about the llvm-commits mailing list