Tensorflow on GPU

[Anthony Wertz]

Hanqi,

I'm not sure if you solved this yet.

The problem I see is the version of the CUDA runtime your package is
looking for: 7.0 is not installed on either GPU1 or GPU2, it's 7.5. This
means you either need to use a newer version of the tensorflow package or
(my recommendation) build your own package in a python virtualenv and use
that.

Predrag

It might be a good idea for us to build the tensorflow package globally on
the GPU machines.



2016-06-30 17:17 GMT-04:00 Hanqi Sun <hanqis at andrew.cmu.edu>:

> I just set the environment path on gpu2 and the problem persists.
>
> Thanks,
> Hanqi
>
>
>
> On Thu, Jun 30, 2016 at 5:10 PM, Hanqi Sun <hanqis at andrew.cmu.edu> wrote:
>
>> Hi all,
>>
>> Thanks for all the replies! Actually, the error message was generated on
>> the server gpu2.int.autonlab.org. And when I tried
>>
>> printenv LD_LIBRARY_PATH
>>
>>
>> I got
>>
>>  /usr/local/cuda/lib64:/usr/lib64/mpich/lib
>>
>>
>> I do not know whether I should change the environment variable on the
>> server. Could someone have a look at it?
>>
>> Thanks,
>> Hanqi
>>
>> On Thu, Jun 30, 2016 at 4:44 PM, Benedikt Boecking <
>> boecking at andrew.cmu.edu> wrote:
>>
>>> Have you tried setting the cuda environment path?
>>>
>>> export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
>>>
>>> On Jun 30, 2016, at 4:31 PM, Hanqi Sun <hanqis at andrew.cmu.edu> wrote:
>>>
>>> Hi all,
>>>
>>> I am a research assistant for Prof. Poczos and would like to use
>>> Tensorflow for my project.
>>>
>>> When I try to execute
>>>
>>> import tensorflow
>>>
>>>
>>> on the GPU node, I get the following error message
>>>
>>> Traceback (most recent call last):
>>>
>>>   File "<stdin>", line 1, in <module>
>>>
>>>   File "/usr/lib/python2.7/site-packages/tensorflow/__init__.py", line
>>>> 23, in <module>
>>>
>>>     from tensorflow.python import *
>>>
>>>   File "/usr/lib/python2.7/site-packages/tensorflow/python/__init__.py",
>>>> line 50, in <module>
>>>
>>>     from tensorflow.python.framework.framework_lib import *
>>>
>>>   File
>>>> "/usr/lib/python2.7/site-packages/tensorflow/python/framework/framework_lib.py",
>>>> line 62, in <module>
>>>
>>>     from tensorflow.python.framework.ops import Graph
>>>
>>>   File
>>>> "/usr/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line
>>>> 40, in <module>
>>>
>>>     from tensorflow.python.framework import versions
>>>
>>>   File
>>>> "/usr/lib/python2.7/site-packages/tensorflow/python/framework/versions.py",
>>>> line 24, in <module>
>>>
>>>     from tensorflow.python import pywrap_tensorflow
>>>
>>>   File
>>>> "/usr/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py",
>>>> line 28, in <module>
>>>
>>>     _pywrap_tensorflow = swig_import_helper()
>>>
>>>   File
>>>> "/usr/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py",
>>>> line 24, in swig_import_helper
>>>
>>>     _mod = imp.load_module('_pywrap_tensorflow', fp, pathname,
>>>> description)
>>>
>>> ImportError: libcudart.so.7.0: cannot open shared object file: No such
>>>> file or directory
>>>
>>>
>>>
>>>
>>> Could anyone check this out and try to fix it? Thanks!
>>>
>>> Also, I will greatly appreciate it if any of you could share your
>>> experience with Tensorflow (especially about image processing) with me as I
>>> am new to this framework. If you are interested, please send a private
>>> email to me and I am happy to stop by at your office at any time.
>>>
>>> Best,
>>> Hanqi
>>>
>>>
>>>
>>
>


-- 
*Anthony Wertz*
Research Programmer and Analyst
Robotics Institute - Auton Lab
Carnegie Mellon University
awertz at cmu.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20160701/be34a933/attachment-0001.html>