0

OS: Ubuntu 18.04/ cuda 10.0/ tensorflow-gpu 1.13.1/ cudnn 7.3.0

When I want to verify the installation of cudnn through ./mnistCUDNN, it runs about 20 mins and failed like this:


cudnnGetVersion() : 7300 , CUDNN_VERSION from cudnn.h : 7300 (7.3.0)
Host compiler version : GCC 7.5.0
There are 1 CUDA capable devices on your machine :
device 0 : sms 20  Capabilities 8.6, SmClock 1485.0 Mhz, MemSize (Mb) 3910, MemClock 6001.0 Mhz, Ecc=0, boardGroupID=0
Using device 0

Testing single precision
Loading image data/one_28x28.pgm
Performing forward propagation ...
Testing cudnnGetConvolutionForwardAlgorithm ...
Fastest algorithm is Algo 0
Testing cudnnFindConvolutionForwardAlgorithm ...
^^^^ CUDNN_STATUS_SUCCESS for Algo 1: 0.001824 time requiring 100 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 0: 0.010240 time requiring 0 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 2: 0.018432 time requiring 57600 memory
^^^^ CUDNN_STATUS_SUCCESS for Algo 4: 0.052224 time requiring 207360 memory
^^^^ CUDNN_STATUS_NOT_SUPPORTED for Algo 3: -1.000000 time requiring 0 memory
Resulting weights from Softmax:
0.0000000 0.9999399 0.0000000 0.0000000 0.0000561 0.0000000 0.0000012 0.0000017 0.0000010 0.0000000 
Loading image data/three_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0000000 0.0000000 0.0000000 0.9999288 0.0000000 0.0000711 0.0000000 0.0000000 0.0000000 0.0000000 
Loading image data/five_28x28.pgm
Performing forward propagation ...
Resulting weights from Softmax:
0.0997550 0.0892810 0.1054468 0.1071846 0.0902156 0.1043498 0.0953814 0.0938049 0.1155469 0.0990339 

Result of classification: 1 3 8

Test failed!
Prediction mismatch
mnistCUDNN.cpp:876
Aborting...

But when I use cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2, it showed:

#define CUDNN_MAJOR 7
#define CUDNN_MINOR 3
#define CUDNN_PATCHLEVEL 0
--
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

#include "driver_types.h"

Also, I tried apt-get like sudo apt-get install libcudnn7-doc=7.3.0.29-1+cuda10.0, they all showed libcudnn7 libcudnn-dev libcudnn-doc are the the newest version (7.3.0.29-1+cuda10.0).

I don’t know where goes wrong.

talonmies
  • 70,661
  • 34
  • 192
  • 269
Lei Shi
  • 11
  • 2

1 Answers1

1

Fixed. Due to GPU memory. 4G isn't enough for my package.

| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      1050      G   /usr/lib/xorg/Xorg                166MiB |
|    0   N/A  N/A      1321      G   /usr/bin/gnome-shell               55MiB |
|    0   N/A  N/A      4739      G   /usr/lib/firefox/firefox          136MiB
Lei Shi
  • 11
  • 2