GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-06-04 18:54:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix CUDA compilation

Browse Source
This commit is contained in:
Gael Guennebaud 2016-07-21 18:47:07 +02:00
parent bf91a44f4a
commit 0f350a8b7e
2 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
Eigen/src/Core/arch/CUDA/MathFunctions.h
View File

@ -27,6 +27,7 @@ float4 plog<float4>(const float4& a)
template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
double2 plog<double2>(const double2& a) double2 plog<double2>(const double2& a)
{ {
using ::log;
return make_double2(log(a.x), log(a.y)); return make_double2(log(a.x), log(a.y));
} }
@ -39,6 +40,7 @@ float4 pexp<float4>(const float4& a)
template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
double2 pexp<double2>(const double2& a) double2 pexp<double2>(const double2& a)
{ {
using ::exp;
return make_double2(exp(a.x), exp(a.y)); return make_double2(exp(a.x), exp(a.y));
} }
@ -51,6 +53,7 @@ float4 psqrt<float4>(const float4& a)
template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE
double2 psqrt<double2>(const double2& a) double2 psqrt<double2>(const double2& a)
{ {
using ::sqrt;
return make_double2(sqrt(a.x), sqrt(a.y)); return make_double2(sqrt(a.x), sqrt(a.y));
} }

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

@ -461,8 +461,8 @@ EigenContractionKernelInternal(const LhsMapper lhs, const RhsMapper rhs,
#undef writeResultShmem #undef writeResultShmem
#undef writeRow #undef writeRow
const int max_i_write = (min)((int)((m_size - base_m - threadIdx.y + 7) / 8), 8); const int max_i_write = numext::mini((int)((m_size - base_m - threadIdx.y + 7) / 8), 8);
const int max_j_write = (min)((int)((n_size - base_n - threadIdx.z + 7) / 8), 8); const int max_j_write = numext::mini((int)((n_size - base_n - threadIdx.z + 7) / 8), 8);
if (threadIdx.x < max_i_write) { if (threadIdx.x < max_i_write) {
if (max_j_write == 8) { if (max_j_write == 8) {