mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-01-30 17:40:05 +08:00
Code Issues Packages Projects Releases Wiki Activity

Added support for prefetching on cuda devices

Browse Source
This commit is contained in:
Benoit Steiner 2015-07-17 15:35:16 -07:00
parent 943035e5bd
commit 513e357b48
1 changed files with 9 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
Eigen/src/Core/GenericPacketMath.h
View File

@ -281,7 +281,15 @@ template<typename Scalar, typename Packet> EIGEN_DEVICE_FUNC inline void pstoreu
/** \internal tries to do cache prefetching of \a addr */
template<typename Scalar> inline void prefetch(const Scalar* addr)
{
#if !EIGEN_COMP_MSVC
#ifdef __CUDA_ARCH__
#if defined(__LP64__)
// 64-bit pointer operand constraint for inlined asm
asm(" prefetch.L1 [ %1 ];" : "=l"(addr) : "l"(addr));
#else
// 32-bit pointer operand constraint for inlined asm
asm(" prefetch.L1 [ %1 ];" : "=r"(addr) : "r"(addr));
#endif
#elif !EIGEN_COMP_MSVC
__builtin_prefetch(addr);
#endif
}