devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merged in rmlarsen/eigen (pull request PR-643)

Browse Source 
Make Eigen build with cuda 10 and clang.

Approved-by: Justin Lebar <justin.lebar@gmail.com>
This commit is contained in:
Rasmus Larsen
2019-05-20 17:02:39 +00:00
parent cc7ecbb124 fd595d42a7
commit e92486b8c3
4 changed files with 33 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h
View File

@@ -674,10 +674,6 @@ struct InnerReductionLauncher<Self, Op, Eigen::half, true> {
if (num_blocks > 1) {
// We initialize the outputs outside the reduction kernel when we can't be sure that there
// won't be a race conditions between multiple thread blocks.
const int dyn_blocks = divup<int>(num_preserved_vals, 1024);
const int max_blocks = device.getNumGpuMultiProcessors() *
device.maxGpuThreadsPerMultiProcessor() / 1024;
const int num_blocks = numext::mini<int>(max_blocks, dyn_blocks);
LAUNCH_GPU_KERNEL((ReductionInitKernelHalfFloat<Self, Op, Index>),
1, 1, 0, device, reducer, self, num_preserved_vals, output);
}