Diable GPU usage

[So]

I know you can use tf.device to select preferred device for specific operations. However, this may result in undesired memory overhead. Is there an easy way to disable GPU usage entirely for an script? Can we set any flag for this purpose similar to other DL libraries?

I want to NOT use GPU at all for an experiment altough I have it available in the machine.


Thanks.

[josh11b]

If you specify the device as /cpu:0, it won't use the GPU.  See

https://www.tensorflow.org/versions/0.6.0/how_tos/using_gpu/index.html#manual-device-placement

Josh

[josh11b]

In the future, this sort of question is best asked on StackOverflow.

Josh

[So]

Thanks for your reply. As I mentioned, I am aware of this option. There are at least two problems with this approach:

1- It is not convenient to specify this for each operation, specially if the program is large.

2- It seems the GPU still remains active. See here:

https://groups.google.com/a/tensorflow.org/forum/#!searchin/discuss/cpu$20timing/discuss/jDQ00G0HnfQ/D6N-z_NNCAAJ

I have also noticed that even after using the flag, my GPU remains active.

Maybe this can be further improved in future releases. Theano,Caffe, Torch... are much more convenient in this regard for seemless switch between CPU and GPU.

[Vijay Vasudevan]

1) Try setting the environment variable CUDA_VISIBLE_DEVICES=-1 

2) https://github.com/tensorflow/tensorflow/blob/bca664cf253e2da345e005b0d90cf4ffa8b3c588/tensorflow/core/framework/config.proto#L79  set "GPU" to 0 when constructing your first Session.

--
You received this message because you are subscribed to the Google Groups "Discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss+u...@tensorflow.org.
To post to this group, send email to dis...@tensorflow.org.
To view this discussion on the web visit https://groups.google.com/a/tensorflow.org/d/msgid/discuss/1d8a37ed-6f86-4bd2-b174-ca10e4ef7cd9%40tensorflow.org.

[bahram...@gmail.com]

Thanks. The first instruction seems to not have any effect but the second one did the trick.
Here is what I did for reference:
config = tf.ConfigProto(
        device_count = {'GPU': 0}
    )
sess = tf.Session(config=config)