Folding @ Home Berlin Overview

Folding@Home (also known as FAH or F@H) is a distributed computing project designed to perform computationally intensive simulations of protein folding and other molecular dynamics. It was launched on October 1, 2000, and is currently managed by the Pande Group, within Stanford University's chemistry department, under the supervision of Professor Vijay Pande. Folding@home is the most powerful distributed computing cluster in the world, according to Guinness, and one of the world's largest distributed computing projects. The goal of the project is "to understand protein folding, misfolding, and related diseases." (from Wikipedia)

this page is still under construction and right now a collection of things I was told about FaH...

General things to consider

Little mistakes can have a huge impact

This means, that each simulation, that is distributed to the community has to be thouroghly tested, to mak sure, that they do not crash. This is made sure by a 3-step testing procedure.

1. Step: Check with local computers, that are probably only placed in the CMB Group in Berlin. Once this clients report a stable simulation go to step 2

2. Step: There exists a small subcommunity, that agreed to run simulations still in an experimental stadium and will report any inconsistancies with them.

3. Step: Give it a go to the whole system

Also the server distributing the data has to be up ALL THE TIME! Since once a job has been run from a certain server the client waits until it can send the data back to the client. Thus a server is supposed to be offline not longer than a few hours!

Vijay Pande decides how much ressources each Server can use


The server need a user called "server", which does not need user rights, but must be capable to run a HTTP Server, so the ports and rights have to be set properly

All communication between the different servers is done via SSH

So the setup in general is the setting up of the Webserver

Trash.Projects are placed in subfolders of the server.

There are different possibilities to run long trajectories. Since all trajectories should be small, that is, they should be computable in about a day on a normal machine, So you can run all trajectories from a small set of initial conditions. If you want a long trajectory, you can assign the server to give trajectories, that start at the end of previously trajectories... That means, that one gets an exponential decreasing number of trajectories with the length. The details can be set in the server. I have to figure out, once we set up the server... This would be the point for the adaptive sampling to be used....

Internal Testing

I set up a team number for all computers of the FU Berlin Comp Mol Bio Group. And the internal testing is restricted to members of this group

The team number is


Make sure the testclients installed (files are on Gaia24) use this teamnumber, otherweise they will not get a project !

We should think about using all the desktop machines as clients. That might help or improve the speed of testing a lot...


Internal FAH Wiki at the Pande Group

There is a big homepage about folding at home, which I will hopefully get an account for. Otherwise I will ask to copy the essential parts... Actually FAH is not too good documented. Since is has grown very complex.


Topic revision: r4 - 15 Jun 2012, AntoniaMey
  • Printable version of this topic (p) Printable version of this topic (p)