r2d3 commented on issue #17826: [quetions] Check failed: e == cudaSuccess: CUDA: initialization error URL: https://github.com/apache/incubator-mxnet/issues/17826#issuecomment-609056307 HI @rondogency and @Rainweic, even with an "import mxnet" in each process, I got similar issue. For example: https://github.com/aws-samples/parallelize-ml-inference is creating `global model` in `init_worker`. It does not do `import mxnet` in each process. I modified the code to do the import in `init_worker` but still gets the CUDA error ` File "/home/ubuntu/anaconda3/envs/mxnet_p36/lib/python3.6/site-packages/mxnet/symbol/symbol.py", line 1776, in simple_bind ctypes.byref(exe_handle))) File "/home/ubuntu/anaconda3/envs/mxnet_p36/lib/python3.6/site-packages/mxnet/base.py", line 255, in check_call raise MXNetError(py_str(_LIB.MXGetLastError())) mxnet.base.MXNetError: [16:40:00] src/engine/./../common/cuda_utils.h:379: Check failed: e == cudaSuccess || e == cudaErrorCudartUnloading: CUDA: initialization error` https://github.com/aws-samples/parallelize-ml-inference/issues/3 Regards David
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
