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

Prevent nvcc from miscompiling the cuda metakernel. Unfortunately this reintroduces some compulation warnings but it's much better than having to deal with random assertion failures.

Browse Source
This commit is contained in:
Benoit Steiner
2016-01-08 13:53:40 -08:00
parent 6639b7d6e8
commit 53749ff415
3 changed files with 7 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

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

@@ -115,11 +115,11 @@ struct FullReducer<Self, Op, GpuDevice, Vectorizable> {
internal::is_same<typename Self::CoeffReturnType, float>::value;
template <typename OutputType>
EIGEN_DEVICE_FUNC static void run(const Self& self, Op& reducer, const GpuDevice& device, OutputType* output) {
static void run(const Self& self, Op& reducer, const GpuDevice& device, OutputType* output) {
assert(false && "Should only be called on floats");
}
EIGEN_DEVICE_FUNC static void run(const Self& self, Op& reducer, const GpuDevice& device, float* output) {
static void run(const Self& self, Op& reducer, const GpuDevice& device, float* output) {
typedef typename Self::Index Index;
const Index num_coeffs = array_prod(self.m_impl.dimensions());