search

Seanlinx / mtcnn

593 stars 264 forks source link

When training P-net, the utilization rate of GPU was 0.Train is so slow. #68

Closed crz1336768325 closed 5 years ago

cayla12 commented 6 years ago

[17:03:21] src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:109: Running performance tests to find the best convolution algorithm, this can take a while... (setting env variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to disable)INFO:root:Epoch[0] Batch [0-200] Speed: 91.87 samples/sec Accuracy=0.616449 LogLoss=0.762187 BBOX_MSE=0.266136 INFO:root:Epoch[0] Batch [200-400] Speed: 101.24 samples/sec Accuracy=0.688269 LogLoss=0.621007 BBOX_MSE=0.086487 INFO:root:Epoch[0] Batch [400-600] Speed: 111.14 samples/sec Accuracy=0.729527 LogLoss=0.566472 BBOX_MSE=0.064344 I have installed mxnet-gpu. Can you tell me where is the mistake？thx