CuDNN library incompatibility error

[barry]

I am getting the following error when running tensorflow:

Loaded runtime CuDNN library: 4007 (compatibility version 4000) but source was compiled with 5103 (compatibility version 5100).  If using a binary install, upgrade your CuDNN library to match.  If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
F tensorflow/core/kernels/conv_ops.cc:457] Check failed: stream->parent()->GetConvolveAlgorithms(&algorithms)

Based on tensorflow issue #3225, I think perhaps cudnn v.5.1 is required, rather than cudnn v.4.0.

[mGalarnyk]

You could be right about that! i will change it and get back to you. Have you changed it and made it work properly?

[barry]

Not yet.

By the way, the standard checks of the tensorflow install work fine:

>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
>>> sess.run(hello)
Hello, TensorFlow!
>>> a = tf.constant(10)
>>> b = tf.constant(32)
>>> sess.run(a+b)
42

The problem arises when I try to run other network examples:

git clone https://github.com/tensorflow/tensorflow
cd tensorflow/tensorflow/models/image/mnist
python convolutional.py
[snip]
E tensorflow/stream_executor/cuda/cuda_dnn.cc:347] Loaded runtime CuDNN library: 4007 (compatibility version 4000) but source was compiled with 5103 (compatibility version 5100).  If using a binary install, upgrade your CuDNN library to match.  If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
F tensorflow/core/kernels/conv_ops.cc:457] Check failed: stream->parent()->GetConvolveAlgorithms(&algorithms) 
Aborted (core dumped)

[zdx3578]

use tensorflow gpu 0.9 i am ok;