Using mxnet on Gradient with GPU support

performance

#1

Hello,

I created a Gradient Notebook with GPU+ hardware and ufoym/deepo:all-py36-jupyter image.

I would like to install mxnet with GPU support.

I tried various commands such as:

import sys
!{sys.executable} -m pip install mxnet-cu80==1.1.0

or

import sys
!{sys.executable} -m pip install mxnet-cu90==1.1.0

But I was not able to import and run mxnet with GPU properly.

With cu80, I get the following error on import:

OSError: libcudart.so.8.0: cannot open shared object file: No such file or directory

with cu90 I get on import:

---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
RuntimeError: module compiled against API version 0xc but this version of numpy is 0xb

---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
RuntimeError: module compiled against API version 0xc but this version of numpy is 0xb

Any suggestion to help me use Mxnet with GPU?

Thanks!


#2

Hey @mremond, mxnet is already installed on this container! Here’s the list of available frameworks:

https://hub.docker.com/r/ufoym/deepo/


#3

Indeed, thanks.
Now, I need to find a way to install TuriCreate and have it use the local mxnet and not install a new one breaking the local one.

Thanks !


#4

@mremond This should work:

import sys
!{sys.executable} -m pip install -U turicreate