cuDNN error on Surface Book 2 when running CNN example.

7 次查看(过去 30 天)
Hello,
I have run into an error when trying to train my convolutional neural network. The error is
Error using trainNetwork (line 154)
Unexpected error calling cuDNN: CUDNN_STATUS_EXECUTION_FAILED.
Error in Analysis4 (line 65)
net = trainNetwork(image_datasetTrain,layers,options);
Caused by:
Error using nnet.internal.cnngpu.convolveForward2D
Unexpected error calling cuDNN: CUDNN_STATUS_EXECUTION_FAILED.
I am defining the layers of my network the same as in the CNN deep learning example. The total size of all images in 20MBs.
I am running R2018a on Windows 10 64-bit.
Please provide some guidance on how to fix this issue.

回答(2 个)

Aydin Sümer
Aydin Sümer 2018-12-3
I have same error too. I have no idea how can i fix this error. Someone help me to fix this error ?

FSB
FSB 2019-1-19
I have tried putting my variables in gpu first in the following manner:
ile=gpuArray(0.0001);
This was done for all numeric values and it worked for me.

类别

Help CenterFile Exchange 中查找有关 Image Data Workflows 的更多信息

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by