Commit Graph

13 Commits (69b44f2f198509e29b8ab50edab9ab34f56fd1af)

Author SHA1 Message Date
tensor-tang 69b44f2f19 unify MKL macro definition
8 years ago
QI JUN 3a0a4586a3
refine GPU memory allocation policy (#6373)
8 years ago
dangqingqing a328ae3b9b Use posix_memalign to allocate aligned memory, since some SIMD instructions require the alignment of memory accesses.
8 years ago
Luo Tao 871a3f6e76 remove unused PADDLE_ONLY_CPU comment
8 years ago
Yi Wang 4558807c48 Use PADDLE_WITH_CUDA instead of PADDLE_WITH_GPU
8 years ago
Yu Yang 84500f9487 Change `PADDLE_ONLY_CPU` to `PADDLE_WITH_GPU`
8 years ago
liaogang 2f7489fbd5 change use_pinned_memory to true for cpu
8 years ago
liaogang f812de2cce ENH: unify PADDLE_ENFORCE
8 years ago
liaogang 199b5fcb45 ENH: refine code comments
8 years ago
liaogang 89110fd266 ENH: Add useGpu in system allocator
8 years ago
liaogang 26cd0bb5a5 ENH: count allocated fallback size for performance
8 years ago
liaogang 169022d014 FIX: Improve fallback gpu allocator
8 years ago
Yi Wang 3e087f763e Add buddy_allocator.cc and system_allocator.cc
8 years ago