[PATCH] D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores

[Stanislav Mekhanoshin via Phabricator via llvm-commits]

rampitec added inline comments.


================
Comment at: lib/Target/AMDGPU/SILoadStoreOptimizer.cpp:1326
 
-  // TODO: Support Store.
-  if (!MI.mayLoad())
+  if (MI.mayLoad() && TII->getNamedOperand(MI, AMDGPU::OpName::vdata) != NULL)
     return false;
----------------
vpykhtin wrote:
> rampitec wrote:
> > rampitec wrote:
> > > nullptr or !TII->getNamedOperand()
> > Stores also have vdata, aren't they?
> this was done to save the original semantic for mayLoad checks, I'm not sure why it is used.
What will happen it you remove mayLoad() from this condition? I think it will still work as expected.


================
Comment at: test/CodeGen/AMDGPU/promote-constOffset-to-imm.mir:194
+# GFX9-LABEL: name: diffoporder_add_store
+# GFX9: GLOBAL_STORE_DWORD %{{[0-9]+}}, %0.sub0, 1000, 0, 0, 0
+# GFX9: GLOBAL_STORE_DWORD %{{[0-9]+}}, %0.sub1, 0, 0, 0, 0
----------------
vpykhtin wrote:
> rampitec wrote:
> > Needs test where stores are actually combined.
> I'm not sure what do you mean, this is about address offset promotion, in this sence it is combined.
I see, that is constant promotion, not the combining.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66958/new/

https://reviews.llvm.org/D66958