All - Activity - cygnus_x1 - DeepTalk - Deep Learning Community

Intel for Display, NVIDIA for ML

For those interested in trying this out, this is the xorg.conf file I used (in /etc/X11/)

Section “ServerLayout”
    Identifier “Layout0”
    Screen 0 “BuiltIn”
EndSection
Section “Screen”
    Identifier “BuiltIn”
    Device “intel”
EndSection
Section “Device”
    Identifier “intel”
…

Hi, I’ve implemented a simplified version of the procedure described here (Intel for display, Nvidia for computing · GitHub) to ensure that the NVIDIA GPU of my tensorbook is dedicated only to tf/keras/python processes while using the Intel GPU to drive the ubuntu/gnome GUI. Prior to this, gnome/x…

TensorBook: was LambdaStack version of tf compiled with XLA support?

I am curious because I’ve just managed to compile tensorflow with XLA support to run on a dual-boot Microsoft Book 2 (SB2), with Ubuntu 18.04 achieving dramatic (-50%) improvement in runtimes.

Tensorbook suspend

I ran into the same problems… looking around there was a suggestion to add “nouveau.modeset=0” to the grub to fix this. This is how my entry in the grub looks like:

grub_CMDLINE_LINUX_DEFAULT=“quiet splash acpi_os_name=Linux acpi_osi=! acpi_osi=Linux acpi_backlight=vendor modprobe.blacklist=nouveau…

Import numba fails

Hi, import numba in a python 3.6 session gives: ModuleNotFoundError: No module named ‘numba’ Lambdastack apparently comes with a version of numba pre-installed (running sudo apt list we can find python3-numba/bionic 0.34.0-3 amd64 installed). Would I need to install numba manually (pip or apt?…