Here I am to make a recap, just in case someone encounter my same problems.
I've initially tried to follow the initial post to the letter, but I suppose that some updates happened somewhere and therefore something went wrong.
It could be also my ignorance, since my Linux/IT knowledge is limited.
I made deep use of chat GPT to try to troubleshoot some of the issues I encountered. For many things it was giving decent information. I am pasting here some of the results.
I was not able to use any GUI functions for the moment, so I followed the link provided in the first post only partially.
Therefore, with this guide you will only get the "base Ubuntu version". I am not sure I am using proper wording.
Clean WSL Install
First of all I recommend to clean the WSL installation that you could have had before.
To do so you can go in the Terminal as administrator
Step 1: Uninstall WSL Distributions
Open PowerShell as Administrator:
- Right-click on the Start button and select "Windows Terminal (Admin)" or "PowerShell (Admin)"
- Unregister/Uninstall Each Distribution:For each distribution listed, use the following command to unregister it:
wsl --unregister <DistroName>
Replace
<DistroName> with the name of the distribution you want to uninstall.
Step 2: Uninstall Windows Subsystem for Linux
- Open Settings:
- Press Win + I to open Settings.
- Navigate to Apps:
- Go to "Apps" > "Optional features".
- Find and Uninstall WSL:
- Scroll down and find "Windows Subsystem for Linux".
- Click on it and select "Uninstall".
Step 3: Restart Your Computer
Restart your computer to ensure all changes take effect.
Step 4: Follow the initial part of the second link from first post
https://kskroyal.com/run-ubuntu-24-04-on-windows-11-with-gui-using-wsl/
Follow until the end of Step 4 of the link above.
Step 5: Install WSL2
Open the power shell
Follow the base instructions, select your username and your passwords
At this point you should have installed the Ubuntu 22.04.
Step 6: Install stuff that are seemingly needed to avoid problems with the Cuda Toolkit installation
While in the WSL terminal
sudo add-apt-repository universe
sudo add-apt-repository multiverse
sudo apt install default-jre libnuma1 libxcb-xinerama0 libxcb-icccm4 libxcb-image0 libxcb-keysyms1 libxcb-render-util0 libxcb-shape0 libxkbcommon-x11-0 libxcb-xinput0 libxcb-cursor0 libopengl0
sudo apt install openjdk-8-jre
sudo apt install openjdk-11-jre
Step 7: Start the update from Ubuntu 22.04 to Ubuntu 24.04
sudo nano /etc/apt/sources.list
put
mantic instead of
jammy everywhere in the file that is open.
What does it mean? You are opening a file and there are several lines where the word jammy is. It's important to change it with mantic in the lines that are not commented (the ones that don't start with "#").
Basically from what chatGPT says jammy is referring to 22.04, mantic is referring to 24.04.
Of course save your file with CTRL+0.
Confirm the name as is.
Control+X to close the file.
During the steps above there will be several request to confirm to proceed and install stuff. I have accepted all blindly.
Step 8: Follow again the initial posts from rbotero
Here basically we are in a situation where we have Ubuntu 24.04 installed, which seems to have all the libraries or whatever they are that are prerequisites for rbotero's steps.
sudo apt install build-essential
sudo apt install nvidia-driver-550
sudo nano /etc/apt/sources.list.d/ubuntu.sources
within that file edit:
Types: deb
URIs:
http://cz.archive.ubuntu.com/ubuntu
Suites: jammy-updates
Components: main universe
Now install libtinfo5 just in case:
sudo apt-get install libtinfo5
Create a folder within your home directory to hold all your PxI, CUDA, Tensorflow files for ease of installation/deployment:
Step 9: Install the cuda toolkit 12.5 (slightly different approach)
You are now in your /PxI_Installation/ folder
Here I am using the points directly available on the Nvidia website.
Looking at the original post #3, there are the two lines under "Update the Cuda sources" that works properly, however the two down the "Download the latest Cuda and Install", are not. Most likely NVidia changed the link, as you could get some authentication error.
Therefore what i recommend to do is the visit the following link:
https://developer.nvidia.com/cuda-downloads
Your selection should be: Linux > x86_64 > WSL-Ubuntu > 2.0 > deb(local)
A small list of commands will be open, basically similar to what Rbotero postes in his post #3.
wget
https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin
sudo mv cuda-wsl-ubuntu.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.c...repo-wsl-ubuntu-12-5-local_12.5.1-1_amd64.deb
sudo dpkg -i cuda-repo-wsl-ubuntu-12-5-local_12.5.1-1_amd64.deb</span><span>sudo cp /var/cuda-repo-wsl-ubuntu-12-5-local/cuda-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get -y install cuda-toolkit-12-5
I want to highlight one thing. The
red line has a
"*" at some point. You shouldn't copy this line as is.
When you enter the
blue line, a feedback will be given in the terminal window. Basically it's going to give you the
red line but with actual numbers instead of the asterisk. Copy that and use that instead of the red line. Then go directly to the "sudo apt-get update".
Step 10: Install cuDNN v8.9.7 (December 5th, 2023)
Here basically I've followed to the letter rbotero's post.
You can't download the file via "wget", so you will be downloading it most likely in your "Download" folder.
Make sure you download the 8.9.7 from the 5th of December, when you read this it might be that you will see a more recent version. It may work, it may not.
The link is
here.
Install the Local Installer for Ubuntu22.04 x86_64(Deb), as rbotero said.
Now you need to move that from Download to your /PiX_Installation/ folder.
Here the example for my case:
mv /mnt/c/Users/USERNAMEWINDOWS/Downloads/cudnn-local-repo-ubuntu2204-8.9.7.29_1.0-1_amd64.deb ~/PxI_Installation/
Of course put your Windows11 username instead of
USERNAMEWINDOWS
sudo dpkg -i cudnn-local-repo-ubuntu2204-8.9.7.29_1.0-1_amd64.deb
sudo cp /var/cudnn-local-repo-ubuntu2204-8.9.7.29/cudnn-*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get -y install libcudnn8
I want to highlight again one thing. The
red line has a
"*" at some point. You shouldn't copy this line as is.
When you enter the
blue line, a feedback will be given in the terminal window. Basically it's going to give you the
red line but with actual numbers instead of the asterisk. Copy that and use that instead of the red line. Then go directly to the "sudo apt-get update".
Step 11: Installing tensor flow in the right folder
Here please make sure that you install your tensorflow folder within
/home/USERNAMEUBUNTU/ and not only
/home/.
Otherwise you will get some issue when you install some of the libraries (e.g. RusselClow Xsterminator suite).
cd /home/USERNAMEUBUNTU/
sudo mkdir tensorflow
cd tensorflow
Here again you must download the tensorflow libraries in your Donwload Windows11 folder and move it into the just created folder.
You can install the version 2.15.0. Mind that at the time I am writing this the 2.15.1 is already available. I didn't try it yet.
You can download it
here.
Otherwise visit the Nvidia website and find from the list of version this one. You can try a newer one but you might need to start all from beginning.
mv /mnt/c/Users/USERNAMEWINDOWS/Downloads/libtensorflow-gpu-linux-x86_64-2.15.0.tar .
sudo tar -xvf /home/tensorflow/libtensorflow-gpu-linux-x86_64-2.15.0.tar.gz
Step 12: Install Pixinsight
At the time that I am writing this, a newer version of 1.8.9-3 is available from the website. It's working. Hopefully it will work also whatever version you will be able to download.
From wherever you are you move back to the PxI_installation folder
For example, from where you should be:
Move Pixinsight from the Download folder to your PxI_Installation folder:
mv /mnt/c/Users/USERNAMEWIN11/Downloads/PI-linux-x64-1.8.9-3-20240625-c.tar.xz ~/PxI_Installation
sudo tar -xf PI-linux-x64-1.8.9-3-20240625-c.tar.xz
Please, note that the red part depends on the version you are downloading. It may be different when you do it.
Of course, put your Windows username to select the right source folder of the Pixinsight file you downloaded.
Step 13: Move the native libraries and set up the new ones
This is exactly the same as rbotero wrote. I explicited some passage.
sudo mkdir /opt/temp
cd /opt/PixInsight/bin/lib
sudo mv libtensorflow* /opt/temp
Update your bashrc at the bottom of the file:
add to bottom of file:
export TF_FORCE_GPU_ALLOW_GROWTH="true"
CTRL + O
Accept the file name as is
CTRL + X
Edit PixInsight.sh to choose the right tensorflow libraries:
cd /opt/PixInsight/bin/
sudo nano PixInsight.sh
And edit that file to point to the folder where you uncompressed Tensorflow v2.15:
#!/bin/bash
appname=`basename $0 | sed s,\.sh$,,`
dirname=`dirname $0`
if [ "${dirname:0:1}" != "/" ]; then
dirname=$PWD/$dirname
fi
LD_LIBRARY_PATH=$HOME/tensorflow/lib:$dirname/lib:$dirname <--- new line
#LD_LIBRARY_PATH=$dirname/lib:$dirname <--- old line now commented out with the addition of #
The
green is already existing,
you add the
blue without the yellow part
you add an "#" to the
orange line that is also already existing.
Step 14: Install your suite making use of cuda and enjoy
I've installed RC suite, tested it on one image.
If you open the "Task manager" you can check what's ramping up. If it's the CPU, not good, it's not working.
If it's the GPU... good for you.
Again I want to thank rbotero and also the other people mentioned/authors of the link in his post.
I have 2 x 3090 installed. Only one is ramping up.
I hope that I can find the way to use all of them.
) and probably i took some missteps when dealing with the keyring part.
