Razer Blade with Ubuntu 16.04 + CUDA + cuDNN

Discussion in 'The Linux Corner' started by mgarciaa, Jun 16, 2017.

Thread Status:
Not open for further replies.
  1. mgarciaa

    mgarciaa New Member

    Hello,

    I am new to this forum, and thinking in getting a new razer blade. I use Ubuntu regularly (double boot /windows) and wanted to verify if the New Razer Blade is capable of running Ubuntu 16.04 + CUDA for Machine Learning apps.

    I read a lot of posts with different experiences, but I think this guy pulled it off: https://asierarranz.github.io/Razer-Blade-1060GTX-CUDA-cuDNN-Ubuntu/ .

    Anyone with similar-successful experiences?

    Thanks for your help!!!!
    Martin
     
  2. Hi,
    i have succesfully installed that set of apps without any issue in Linux Mint 18.1 wich is based on Ubuntu 16.04. So, I think it should work for you.

    First, ensure you have enabled/installed the nVidia drivers.
    Second, I have followed basicaly this procedure: https://www.artofai.io/en/get-noticed/installing-tensorflow/
    The installation of cuDNN library didn't work for me but installing the corresponding .deb package from the page described in the tutorial was succesfull.

    JP
     
  3. linuxrazer2_no_id

    linuxrazer2_no_id New Member

    @mgarciaa
    That's what I bought it for.

    For stability I needed to update to Kernel 4.11
    I had also updated to Nvidia 383 prior

    (both are outside standard Ubuntu but only took 10-15mins each, plenty of tutorials).

    had to fiddle a bit to get external monitor behaving but it was solved in 20 mins.

    I haven't installed TensorFlow etc. yet, will probably take a crack at that in the next few days.
    ---

    Microsoft Store Sydney (Australia) has a 30 day return policy and that's where I bought it, knowing I would need to make sure it works. It was shaky until Kernal 4.11. It's not perfect feature wise, but fine stability wise and seems to be the most GPU you can pack in this weight/size class (there are some others with a 1060 but they will have tougher return policies/warranties). Also the Microsoft Store honors their generous student discount for the Udacity course I'm doing as well.
     
  4. @mgarciaa I also bought this computer for this purpose. I am successfully running Ubuntu Gnome 16.04 with CUDA + cuDNN.... Once you update the Kernel it works great.

    I initially tried using the CUDA script (found here (found here https://developer.nvidia.com/cuda-downloads) to install graphics drivers for me but it broke my machine. Use the CUDA script to install CUDA, but install your graphics card drivers from the nvidia ppa repositories (apt-get install nvidia-381 as these are compatible with the 4.11 kernel)

    webcam and suspend are super glitchy for me, and the battery life is very bad on ubuntu (2 hours-max just web browsing) (i haven't tried to optimize yet), but pretty happy overall. its a beautiful machine, happy i made the transition away from apple.
     
  5. mgarciaa

    mgarciaa New Member

    Thank you so much for your replies. So in summary, Blade runs well @Ubuntu for Machine Learning, but If i have a conference, will have to use webcam @Windows - dual boot.

    All the best!
    MG
     
  6. @mgarciaa very mysteriously, I tried out 'Zoom' conference software on ubuntu, and the webcam works perfectly.... No other program works.
     
  7. alexfridman

    alexfridman New Member

    Hi everyone! I have a few questions for Razer Blade holders (in DL/ML purposes). How do you deal with a low battery lifetime with enabled GPU? Whiles you are not running CUDA computations do you disable GPU? I read somewhere, disabling GPU requires a restart. Is it causes inconvenience in regular daily usage or maybe restart is not necessary?
     
Thread Status:
Not open for further replies.
Sign In with Razer ID >


Don't have a Razer ID yet?
Get Razer ID >