I create a virtual machine in Azure for training a machine learning model using tensor flow(python) with the following specification:
Image: NVIDIA GPU-Optimized VMI with vGPU driver
size: D4s_v3
System Operative: Ubuntu 20.04.5 LTS
I'm trying accelerate the training process but don't recognizer the Nvidia/cuda.
I download the Nvidia toolkit manually with the follow command
apt install nvidia-cuda-toolkitlsb_release -a
but when I try to see the controllers with inxi show this.
inxi -Gx
Graphics: Message: No Device data found.
Display: server: No display server data found. Headless machine? tty: 204x63
Message: Advanced graphics data unavailable in console for root.
Related
I've been developing a Nodejs app using face-api.js that depends on #tensorflow/tfjs-node. So far everything was working well on my windows local machine and on Heroku server. But after I pushed the app to a Ubuntu VPS server the app won't run, whenever I write either
var tensor = require('#tensorflow/tfjs-node');
or
var faceapi = require('#vladmandic/face-api');
The console prints "Illegal instruction" and that's it, no trace or anything. I figured maybe tensorflow is not compatible with the server system/hardware. I read that I can build tensorflow using bazel, which I tried but the server runs out of memory before the build completes. Someone told me that I can build tensorflow on my windows local machine. I tried and failed... I don't know what configurations I should use. Is there is a way to make tensorflow work on the VPS? Or successfully build one?
VPS information (I got it from NameCheap, Pulsar VPS):
Processor information: QEMU Virtual CPU version 2.5+, 2 cores
RAM: 2 GB
Operating system: Ubuntu Linux 20.04
Kernel and CPU: Linux 5.4.0-81-generic on x86_64
Storage: 40 GB
I tried setting up Nannou following the instructions here.
I am running Debian Buster on a MacBook with an NVIDIA graphics card (GK107M [GeForce GT 750M Mac Edition]).
I tried running the example and get the error
thread 'main' panicked at 'could not build default app window: NoAvailableAdapter'
It seems this is an error when Nannou tries to open a window -- something about its communication with the window manager (Gnome on xOrg), or something about vulcan-tools.
Any ideas for how to debug this?
The issue is that I was using the open-source Nouveau driver for my graphics card (the default on Debian) which does not currently support Vulkan, which is required for Nannou.
By installing the proprietary nvidia graphics driver, the problem was resolved.
I was able to install the nvidia driver by running
apt install nvidia-driver,
and then rebooting my computer,
as described in this tutorial https://linuxconfig.org/how-to-install-nvidia-driver-on-debian-10-buster-linux
Now nannou works :)
I am trying to run the XV6 OS on ubuntu image in a docker container (on windows).
I succeed in building the image, running it and accessing the docker container.
But inside the container cmd when I try to use the make qemu-nox command it crashes and gives the following error:
SeaBIOS (version 1.13.0-1ubuntu1)
iPXE (http://ipxe.org) 00:03.0 CA00 PCI2.10 PnP PMM+1FF8CA10+1FECCA10 CA00
Booting from Hard Disk...
Boot failed: not a bootable disk
No bootable device.
I am following the instructions here: https://github.com/anton-christensen/xv6-docker
I am not sure what causes this.
The other commands such as make, make clean qemu does work without error.
I installed docker CE on an Ubuntu 16.04, and when I try:
docker pull microsoft/windowsservercore
It fails:
Using default tag: latest
latest: Pulling from microsoft/windowsservercore
3889bb8d808b: Pulling fs layer
da87b55a9b63: Pulling fs layer
image operating system "windows" cannot be used on this platform
Is it possible to run a windows image on a linux host with docker? If yes, how should I do that?
Unlike other Virtual computing systems like Virtual Box or VMWare, Docker images are small and not fully complete stand alone entities.
The Docker image does not include the OS kernel, instead it uses the kernel of the Docker host
This makes the images very light weight, but it also means you won't be able to run a Windows image on a Linux host.
I am trying to boot a cloned image on KVM ,which is in the raw format, the image is a clone of aws ubuntu 14.04 LTS hvm instance. It gives me an error saying no bootable device found. The same image boots up when I specify the kernel path explicitly while creating the VM.
I am using virt-manager to create the VM and the qemu version is 2.0.0
I have tried changing the disk bus but nothing helped.
Can anyone help ?
It was because of the partitions on the Ubuntu 14.04 LTS. Cloned the entire disk. Boots up fine now.