[PATCH] D39036: AMDGPU: Fix default range in non-kernel functions

Konstantin Zhuravlyov via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Oct 23 09:24:58 PDT 2017


kzhuravl accepted this revision.
kzhuravl added a comment.
This revision is now accepted and ready to land.

LGTM.



================
Comment at: lib/Target/AMDGPU/AMDGPUSubtarget.cpp:199
+  case CallingConv::SPIR_KERNEL:
+    return std::make_pair(getWavefrontSize() * 2, getWavefrontSize() * 4);
+  case CallingConv::AMDGPU_VS:
----------------
arsenm wrote:
> t-tye wrote:
> > Should there be a target property that returns the maximum size supported? 4 * wavefront size may not match every target.
> It's the same for every target now, so there's no real point in adding one.
I think, if need arises, this can be done in separate change.


https://reviews.llvm.org/D39036





More information about the llvm-commits mailing list