FAH Middleware

[Thalla]

Hi ,
my English isn't very well so I didn't know in which forum this question fits. I hope I'm not too wrong here.
I'm interested in the middleware used by Folding@Home, but I couldn't find any information about it.
I hope that someone of you can tell me, which middleware is used in this project and why.
Thanks for your help!
Thalla

[Jesse_V]

FAHClient might be what you are looking for. FAHClient is described briefly in several places, but it is middleware that manages the Folding@home processing. It's one of several components used client-side by F@h, the others are FAHCores, FAHControl, and FAHViewer. The calculations for the protein simulations are performed by FAHCores, and they are started, stopped, and given work by FAHClient. FAHClient also uploads the results and downloads fresh work units for processing. FAHControl is a front-end control GUI for FAHClient, and FAHViewer renders the proteins that are being processed by the FAHCores. Is this what you were looking for?

[Thalla]

Thank you very much! I wouldn't have thought that something called client is a middleware ^^.
Do you know which middleware ist used server-side?
FAHCores isn't a middleware, isn't it?
Does there exist a better middleware than FAHClient for this job? Are there some disadvantages?
It is really fascinating to see which parts work together on which tasks.

[ChristianVirtual]

What disadvantages do you see with the current infrastructure ?

There is no valid alternative and the main components used on the client and server are proprietary in their binary form. This is mainly to ensure the integrity of assigned/return WU and avoid possible manipulation can take place.

You could get the source code of some components like viewer, FAHconfig and the basic cores (https://simtk.org/xml/index.xml).
The later one you could use for your own setup of something similar to FAH; but not to integrate. The backend server will only accept specific signed WU. Viewer and FAHconfig you could change at free will as changes here not influence the integrity.

This said: there are 3rd party tools and methodologies to optimize local donor setup. Like: to link CPU cores for the technical processes and avoid many context switches for CPU cores or manage download/upload timings (The Kraken or Langouste comes in mind;Google it). Support for those tools are in the respective 3rd forums.

[bruce]

Are you a scientist trying to do research on proteins you are studying or are you a donor who wants to contribute computer time to Stanford's research?

Jesse_V's question aims at understanding what kind of middleware, if any, you might need, but the answer MIGHT be that you dont need any.

Two suppliers of the underlying analysis software are Simtk.org and Gromacs.org, but there's nothing in the middle except the proprietary code that Stanford has developed to prevent cheating.

[mdk777]

I think the OP is just asking out of curiosity.

I don't think PG has published many(any) papers on the IT infrastructure at Stanford as it does not touch on the issues of biology and numerical methods applied to molecular forces.