Client folding other causes. [curious]
Moderators: Site Moderators, FAHC Science Team
Client folding other causes. [curious]
So, I've been folding for a little bit... but I don't quite understand the system yet. I've got a couple of identical twin systems running proxmox, and they needed a job to do in their spare time, cool, lets fight cancer (Mom's had it, Killed dad, grandpa's got it, killed Grandma. Time to do what I can.) Anyways... I noticed at least 2 of my machines that are folding are working on the Flu. Don't get me wrong, no problem there... Flu is a great cause, just curious, what made it decide to deviate from the "Cause Preference"?
-
Joe_H
- Site Admin
- Posts: 7871
- Joined: Tue Apr 21, 2009 4:41 pm
- Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp2 - Location: W. MA
Re: Client folding other causes. [curious]
Welcome to the folding support forum.
Well, as you have seen it is treated as a preference, not as a filter excluding any other cause. There can be many reasons for getting a WU from a different cause. The easiest to understand is just being temporarily out of WUs for the cause set in the client.
Each project may take hundreds of thousands WUs to complete, but at any given time WUs for a smaller number of trajectories are being processed or waiting to be assigned. That comes from projects have both parallel and serial aspects. Each Run of a project represents starting from different initial configurations of the system, then for each Run there are multiple Clones with different starting velocities for the atoms making up the protein system being investigated. So if for example there were 50 Runs, each having 100 clones, that would define 5000 trajectories through the system that can be done in parallel. Each of these Run, Clone trajectories start with Gen 0, and each Gen needs to be returned before the next Gen can be created and sent out for assignment, that is the serial part of the processing.
Other things that can result in assignment to a different cause are things like the server is temporarily offline, or it has reached its limit of simultaneous connections. The Assignment Servers get regular updates on availability and select a Work Server based on the cause preference, project priorities, and whether WUs and connections are available.
Well, as you have seen it is treated as a preference, not as a filter excluding any other cause. There can be many reasons for getting a WU from a different cause. The easiest to understand is just being temporarily out of WUs for the cause set in the client.
Each project may take hundreds of thousands WUs to complete, but at any given time WUs for a smaller number of trajectories are being processed or waiting to be assigned. That comes from projects have both parallel and serial aspects. Each Run of a project represents starting from different initial configurations of the system, then for each Run there are multiple Clones with different starting velocities for the atoms making up the protein system being investigated. So if for example there were 50 Runs, each having 100 clones, that would define 5000 trajectories through the system that can be done in parallel. Each of these Run, Clone trajectories start with Gen 0, and each Gen needs to be returned before the next Gen can be created and sent out for assignment, that is the serial part of the processing.
Other things that can result in assignment to a different cause are things like the server is temporarily offline, or it has reached its limit of simultaneous connections. The Assignment Servers get regular updates on availability and select a Work Server based on the cause preference, project priorities, and whether WUs and connections are available.
iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Re: Client folding other causes. [curious]
Ah, Cool. Well that explains it! Hey as long as long as it's helping with something... right? Thank you for the clean explanation! I was also going to ask about having trouble with remote access, but I figured it out. My control port was binding to localhost... fancy that.