[Openmp-commits] [PATCH] D51875: [OPENMP][NVPTX] Add support for lastprivates/reductions handling in SPMD constructs with lightweight runtime.

[Alexey Bataev via Phabricator via Openmp-commits]

ABataev added inline comments.


================
Comment at: libomptarget/deviceRTLs/nvptx/src/option.h:37
 // memory.
-#if __CUDA_ARCH__ >= 600
+#if __CUDA_ARCH__ >= 900
+#define OMP_STATE_COUNT 32
----------------
Hahnfeld wrote:
> Hahnfeld wrote:
> > ABataev wrote:
> > > Hahnfeld wrote:
> > > > This doesn't exist unless you have information that are not public yet. Volta is `720` at most.
> > > According to this https://docs.nvidia.com/cuda/volta-tuning-guide/index.html, it is 84
> > I'm not commenting on `MAX_SM`, rather on the value of `__CUDA_ARCH__`. As such these defines are never active.
> That's now 1 GiB of global memory that can't be used by the user application.
Just like I said, I can reduce the size of the preallocated buffers.


Repository:
  rOMP OpenMP

https://reviews.llvm.org/D51875