java tensorflow GPU环境运行的问题

问题：在本地cpu环境下运行良好，放到服务器上运行时报如下错误

Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR

 is org.tensorflow.TensorFlowException: 2 root error(s) found.
  (0) Unknown: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
         [[{{node 0-convolutional}}]]
  (1) Unknown: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
         [[{{node 0-convolutional}}]]
         [[output/_3]]

原因：在服务器上运行时使用的GPU环境，会使用cudnn，cudnn版本与cuda版本不匹配

解决：安装匹配的cudnn版本