异步计算:同步各GPUs 的 loss 时出错 CUDA: an illegal memory access was encountered

请问有人遇到过这种情况吗? 使用list来存储N个GPUs的loss,同步时出错了

batch_loss = (sum(costs)/len(costs)).asscalar()
File "/home/miniconda3/envs/mxnet/lib/python3.6/site-packages/mxnet/base.py", line 149, in check_call
    raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: [14:54:29] /home/travis/build/dmlc/mxnet-distro/mxnet-build/3rdparty/mshadow/mshadow/./stream_gpu-inl.h:62: Check failed: e == cudaSuccess CUDA: an illegal memory access was encountered

用的是pip install mxnet-cu80 版本

pip install mxnet

会安装CPU版的mxnet吧?官网上给出的GPU版命令是:

pip install mxnet-cu90

我记错了,我是用cu80版本的

这个问题你后来解决了吗?

这个问题怎么解决呢?我也遇到了