Hello,
Apologies if this is already addressed in one of the other threads, but I'm getting the following error whenever I try to use theano and the gpu (including if I just run the check1.py in the dm-theano example):
WARNING (theano.sandbox.cuda): CUDA is installed, but device gpu is not available (error: Unable to get the number of gpus available: initialization error)
I've tried this on my own machine with nvidia-docker and everything works fine, but can't get it working on the challenge nodes. If I do nvidia-smi, everything looks fine however. Do you know what the issue is?
Thanks,
Bill
Created by Bill Lotter bill_lotter Hi Bill,
I submitted the following job to "Digital Mammography Model Training". The [theano-gpu](https://hub.docker.com/r/tschaffter/theano-gpu/) base image that we provide seems to work fine.
`Dockerfile`:
```
FROM tschaffter/theano-gpu
WORKDIR /
COPY preprocess.sh .
```
`preprocess.sh`:
```
#!/bin/bash
/usr/bin/python /root/check1.py
```
Submission file:
```
preprocessing=docker.synapse.org//theano-test@sha256:44d438a95ae5d42733fa768708128a87fa1fe5780e41cb9e01017ecc6e188e5d
```
Output (log file):
```
STDERR: /usr/lib64/python2.7/site-packages/numpy/lib/utils.py:254: FutureWarning: Numpy has detected that you (may be) writing to an array returned
STDERR: by numpy.diagonal or by selecting multiple fields in a record
STDERR: array. This code will likely break in the next numpy release --
STDERR: see numpy.diagonal or arrays.indexing reference docs for details.
STDERR: The quick fix is to make an explicit copy (e.g., do
STDERR: arr.diagonal().copy() or arr[['f0','f1']].copy()).
STDERR: ai = a.__array_interface__
STDERR: Using gpu device 0: Tesla K80 (CNMeM is enabled with initial size: 10.0% of memory, cuDNN 5105)
STDOUT: [GpuElemwise{exp,no_inplace}(), HostFromGpu(GpuElemwise{exp,no_inplace}.0)]
STDOUT: Looping 1000 times took 0.292922 seconds
STDOUT: Result is [ 1.23178029 1.61879349 1.52278066 ..., 2.20771813 2.29967761
STDOUT: 1.62323296]
STDOUT: Used the gpu
``` Dear Bill,
Apologies for the delay in response. We are looking into this.
Best,
Thomas
Drop files to upload
Issue with Theano example using CUDA page is loading…