Page 1 of 1
Query on feasibility
Posted: Thu Apr 30, 2020 1:56 pm
by moos3h
Hey all,
I am new to folding@home and I will most likely start folding at home. However the company I work for has between 200 - 400 systems which are idle for up to 24 hours (all varying specifications). Would it be feasible to turn their idle GPU and CPU power to folding? Or would it be a giant headache to sort out.
Re: Query on feasibility
Posted: Thu Apr 30, 2020 2:07 pm
by Neil-B
There are a number of hurdles … first of which is usually permission from the "CEO" or whoever owns the kit … then the IT Security Teams my well have Policies that forbid this type of thing … and the matter of increased electricity for running kit and cooling and the inevitable kit failures that occur when running on larger pools of kit - not as such FAHs fault but that some kit will just fail when worked hard for many reasons (and FAAH works kit hard) … So in some ways it can seem to be a massive migraine.
Having said that there are many large/medium/small corporate contributors and numerous people have succeed in working through the layers of things to be sorted and are making great contribution - having "CEO" and IT Teams support tends to be pretty essential though.
Re: Query on feasibility
Posted: Thu Apr 30, 2020 3:01 pm
by moos3h
Thanks I am just looking at the feasibility of it before we approach the Director on this, a few top level managers are onboard so that is the first hurdle!
Re: Query on feasibility
Posted: Thu Apr 30, 2020 5:43 pm
by HaloJones
The client needs to be installed, not started, a configuration file created and set up, then the whole thing packaged up and deployed to all the machines. You need to start the machines up in small waves to avoid them all trying to get work simultaneously and also to be sure they're working properly.
It massively helps if all the machines are identical. If they aren't you might need individual configurations for each type.
You then need to be able to monitor them all so setting up a central machine that can stop and start the clients. All good fun if you have nothing else to do.
Re: Query on feasibility
Posted: Thu Apr 30, 2020 8:28 pm
by Joe_H
There have been a few topics here on deploying to large groups of machines over the last 4-6 weeks. Most have not been active the last 2-3 weeks, but a search for this topics should turn up some information from others who have done this.
Re: Query on feasibility
Posted: Fri May 01, 2020 5:11 am
by PantherX
Welcome to the F@H Forum moos3h,
I know that this application was a substitute for FAHControl to monitor clients:
https://gitlab.com/nilslennart.bruns/fahcluster