Results 1 to 3 of 3

Thread: tensorflow2 installation error

  1. #1

    Default tensorflow2 installation error

    Hi when I try to install tensorflow2 from official repository by this command:

    Code:
    zypper in tensorflow2
    The following error immediately returned:

    Code:
    Loading repository data...
    Reading installed packages...
    Resolving package dependencies...
    
    Problem: nothing provides libgrpc.so.8()(64bit) needed by tensorflow2-2.1.1-3.5.x86_64
     Solution 1: do not install tensorflow2-2.1.1-3.5.x86_64
     Solution 2: break tensorflow2-2.1.1-3.5.x86_64 by ignoring some of its dependencies
    
    Choose from above solutions by number or cancel [1/2/c/d/?] (c):
    I guess it's a bug and missing dependency. can you confirm this is packageing bug?

    more information:
    I already installed python3-numpy
    python3-pandas python3-matplotlib python3-scikit-learn python3-opencv python3-pylint python3-pip python3-dev cairo-dev
    and from pip: pygobject pycairo

    I don't have any dedicated gpu card such as nvidia gpu, my gpu is an old AMD APU (A4-6300)

  2. #2
    Join Date
    Sep 2012
    Posts
    5,973

    Default Re: tensorflow2 installation error

    tensorflow2 fails to build since 2 months, so binaries that you get are outdated and built against older library version. In the meantime grpc was updated, and old version (that was used to build tensorflow2 in the past) is no more available.

    The only solution is to fix tensorflow2 to build again.

  3. #3
    Join Date
    Jun 2008
    Location
    San Diego, Ca, USA
    Posts
    12,876
    Blog Entries
    2

    Default Re: tensorflow2 installation error

    I don't know what is the problem in the OBS build,
    But you can easily install from PyPi.
    Although the official Tensorflow documentation describes installing on Ubuntu, I don't see any dependencies that would prevent installing on an RPM system like openSUSE.
    The CUDA instructions are another matter though... Any viewer should look up and follow the instructions for your version of CUDA.

    From the Tensorflow website,

    https://www.tensorflow.org/install/pip

    For GPU computing, but use it only as a general setup guide. You should instead install CUDA using nVidia instructions for your version of CUDA.

    https://www.tensorflow.org/install/gpu

    For those who want to attempt to build from source instead of from PyPi packages (recommended). Skimming the instructions, no mention is made of dependencies that might exist on the system, only packages from PyPi. The procedure to build Tensorflow may have changed since the OBS project was created or something may be left out.

    https://www.tensorflow.org/install/source

    HTH,
    TSU
    Beginner Wiki Quickstart - https://en.opensuse.org/User:Tsu2/Quickstart_Wiki
    Solved a problem recently? Create a wiki page for future personal reference!
    Learn something new?
    Attended a computing event?
    Post and Share!

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •