Difference between revisions of "Login to SciComp GPUs"

From epsciwiki
Jump to navigation Jump to search
Line 1: Line 1:
−
The following is how to use one of the ML scicomp machines that has 4 Titan RTX GPU cards installed. </br>
+
<p>The following is how to use one of the ML scicomp machines that has 4 Titan RTX GPU cards installed.
−
A couple of notes: </br>
+
<br>
−
Setting up the software environment seems to be more easily done using conda. Will Phelps sent me the magic commands.
+
A couple of notes: <br>
−
It seems at least part of the environment is maintained when the srun command is called. At least I was still in the same working directory and python3 was pointing to the venv. However, the virtual environment was not completely set up so I had to source the activate script again
+
Setting up the software environment seems to be more easily done using conda. We need to first log into jlab common environment with ssh as <br>
−
I was able to pip install from the sciml190X node whereas it failed to do it from the regular ifarm node due to missing SSL modules.
+
<br>
 +
ssh login.jlab.org

Revision as of 01:56, 26 August 2020

The following is how to use one of the ML scicomp machines that has 4 Titan RTX GPU cards installed.
A couple of notes:
Setting up the software environment seems to be more easily done using conda. We need to first log into jlab common environment with ssh as

ssh login.jlab.org

Retrieved from "https://wiki.jlab.org/epsciwiki/index.php?title=Login_to_SciComp_GPUs&oldid=285"