--temp-dir for hpc

[NicolasProvencher]

Hi,

I am looking to test you pipeline on a HPC of ours where the scratch is a lustre filesystem and they have node specific tmpdir ssd avalaible

I was looking in you script and there doesnt seem to be any implementation of the --temp-dir argument for the single gpu setup in your sh script even though you specify in the hpc doc to set this argument, is this an oversight?

Thankyou

[NicolasProvencher]

AH nevemind I found the info in the more advanced markdown, its an argument given to run_databases.py

[benjamin-perry-duke]

Great! Let me know if you run into any other issues running.

[NicolasProvencher]

I am trying to run on an HPC and I am realising that the --temp-dir doesnt seeem to be set up in the run_alphafast.sh scripts to correctly assign it,

I am not super familiar in complex bash but I am going to try and implement it but it would make like easier for people like to me to have this implemented.

Also, is --temp-dir also implemented for the multi gpu setup

Also I would like to know if you are aware of the amount of temp_dir memory needed, if its not too much it would be optimal for us to set a ramdisk location for this variable and increased the ram requirement