[PATCH] D72931: AMDGPU: Add a16 feature to gfx10

[Sebastian Neubauer via Phabricator via llvm-commits]

sebastian-ne marked an inline comment as done.
sebastian-ne added a comment.

The added CodeGen/AMDGPU/llvm.amdgcn.image.a16.dim.ll fails for gfx10, but a16 does not work anyway at the moment.



================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:5628
                                           : AMDGPU::MIMGEncGfx10Default,
-                                   NumVDataDwords, NumVAddrDwords);
+                                   NumVDataDwords, VAddrs.size());
   } else {
----------------
Are you sure that’s right?
The test CodeGen/AMDGPU/llvm.amdgcn.image.nsa.ll fails for me with this change.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D72931/new/

https://reviews.llvm.org/D72931