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

Add a EIGEN_NO_CUDA option, and introduce EIGEN_CUDACC and EIGEN_CUDA_ARCH aliases

Browse Source
This commit is contained in:
Gael Guennebaud
2017-07-17 01:02:51 +02:00
parent f0b154a4b0
commit bbd97b4095
33 changed files with 134 additions and 126 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
unsupported/Eigen/CXX11/src/Tensor/TensorReduction.h
View File

@@ -334,7 +334,7 @@ struct OuterReducer {
};
#if defined(EIGEN_USE_GPU) && defined(__CUDACC__)
#if defined(EIGEN_USE_GPU) && defined(EIGEN_CUDACC)
template <int B, int N, typename S, typename R, typename I>
__global__ void FullReductionKernel(R, const S, I, typename S::CoeffReturnType*, unsigned int*);
@@ -694,7 +694,7 @@ struct TensorEvaluator<const TensorReductionOp<Op, Dims, ArgType, MakePointer_>,
#ifdef EIGEN_USE_THREADS
template <typename S, typename O, bool V> friend struct internal::FullReducerShard;
#endif
#if defined(EIGEN_USE_GPU) && defined(__CUDACC__)
#if defined(EIGEN_USE_GPU) && defined(EIGEN_CUDACC)
template <int B, int N, typename S, typename R, typename I> KERNEL_FRIEND void internal::FullReductionKernel(R, const S, I, typename S::CoeffReturnType*, unsigned int*);
#ifdef EIGEN_HAS_CUDA_FP16
template <typename S, typename R, typename I> KERNEL_FRIEND void internal::ReductionInitFullReduxKernelHalfFloat(R, const S, I, half2*);
@@ -781,7 +781,7 @@ struct TensorEvaluator<const TensorReductionOp<Op, Dims, ArgType, MakePointer_>,
Op m_reducer;
// For full reductions
#if defined(EIGEN_USE_GPU) && defined(__CUDACC__)
#if defined(EIGEN_USE_GPU) && defined(EIGEN_CUDACC)
static const bool RunningOnGPU = internal::is_same<Device, Eigen::GpuDevice>::value;
static const bool RunningOnSycl = false;
#elif defined(EIGEN_USE_SYCL)