沧海一粟

More know, more less

Tensorflow 源码分析-  从GPU OOM开始说Tensorflow的BFC内存管理

前言在平台上跑GPU训练,结果CUDA OOM了,错误提示E Internal: failed initializing StreamExecutor for CUDA device ordinal 0: Internal: failed call to cuDevicePrimary CtxR...

2018-05-04 11:46:13

阅读数:2388

评论数:0

提示
确定要删除当前文章?
取消 删除