Minor technical items for discussion. (Resolved)

Moderators: Site Moderators, PandeGroup

Minor technical items for discussion. (Resolved)

Postby larrym82 » Thu Nov 10, 2011 4:33 pm

Hello again,

Providing details is definitely good practice. I installed the v6.02 Uniprocessor client on a PentiumII 400MHz machine running Linux Ubuntu 10.04 Lucid Linux.

I understand that it is possible to operate more than one client on a machine. My questions are:

I like the graphic screensaver for its aesthetics, but wonder whether it's worth the cost to run the monitor and whether it's available for Linux?

Could I run another client under WINE and get the graphic screensaver?

Would it be worthwhile given the processor speed mentioned above?

Larry
Last edited by larrym82 on Wed Nov 16, 2011 5:07 am, edited 1 time in total.
larrym82
 
Posts: 30
Joined: Wed Nov 09, 2011 6:47 pm

Re: Minor technical items for discussion.

Postby 7im » Thu Nov 10, 2011 6:07 pm

You would need multiple processors to run multiple clients. As I recall, the P2-400 is a single processor. Sorry.
How to provide enough information to get helpful support
Tell me and I forget. Teach me and I remember. Involve me and I learn.
User avatar
7im
 
Posts: 14648
Joined: Thu Nov 29, 2007 4:30 pm
Location: Arizona

Re: Minor technical items for discussion.

Postby bruce » Thu Nov 10, 2011 8:52 pm

It might be possible to REPLACE your Linux client with a client running under WINE. I DO NOT recommend it, though.

The Windows sceensaver client was discontinued for several reasons. First, performance will suffer, compared to the console client. A 400 MHz machine can just barely meet the deadlines and you'll potentially just be wasting time on work that will be discarded anyway when it expires. Second, the screensaver was unreliable. Stopping and restarting processing carries risks and some percentage of those times when you wake up your machine, it will fail to resume work, restarting that assignment from the beginning.
bruce
 
Posts: 21686
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Minor technical items for discussion.

Postby larrym82 » Thu Nov 10, 2011 9:11 pm

7im wrote:You would need multiple processors to run multiple clients. As I recall, the P2-400 is a single processor. Sorry.


You are correct. Thank you. Once Windows 7 and any future consumer type OS make my machine obsolete, I will have a multi-core machine to play with. It may be the case that the older processor isn't suitable, but is happily operates under Ubuntu 10.04. I'll double check the requirements.

Also have cleared up a distinction between the screen saver and viewer. I'm working now to see whether the latter is available for Linux. No biggie, but it would be nice. First I have to resolve the apparent start-up issue. I thought it was resolved when I got my download, but still don't see any processor activity of note and the terminal isn't feeding back.
larrym82
 
Posts: 30
Joined: Wed Nov 09, 2011 6:47 pm

Re: Minor technical items for discussion.

Postby larrym82 » Thu Nov 10, 2011 9:18 pm

bruce wrote:It might be possible to REPLACE your Linux client with a client running under WINE. I DO NOT recommend it, though.

The Windows sceensaver client was discontinued for several reasons. First, performance will suffer, compared to the console client. A 400 MHz machine can just barely meet the deadlines and you'll potentially just be wasting time on work that will be discarded anyway when it expires. Second, the screensaver was unreliable. Stopping and restarting processing carries risks and some percentage of those times when you wake up your machine, it will fail to resume work, restarting that assignment from the beginning.


Bruce, you are also correct, and certainly one mistake I've made is to set my config file to accept very large WU's, so I'll correct that very soon. I'm a bit of a rabbit when it comes to this. I get all excited about something, skim read and end up doing some things over. What's that old saying? "If you don't have time to do it right, you must have time to do it again."

Such is life. Thanks.
larrym82
 
Posts: 30
Joined: Wed Nov 09, 2011 6:47 pm

Re: Minor technical items for discussion.

Postby bruce » Thu Nov 10, 2011 9:47 pm

larrym82 wrote:First I have to resolve the apparent start-up issue. I thought it was resolved when I got my download, but still don't see any processor activity of note and the terminal isn't feeding back.


Be sure you're looking at top or some other CPU monitor rather than terminal output. You'll probably see FahCore_78.exe using 99% of the processor for a long, long time before you see any change on the terminal. As I said earlier, 400 MHz will just barely meet the deadlines. Check the project number on the "Project Summary" link in the forum header. Divide that time by 100 and that's about when you can expect a message saying you've completed 1% of the WU.
bruce
 
Posts: 21686
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Minor technical items for discussion.

Postby larrym82 » Thu Nov 10, 2011 9:56 pm

Good advice, Thanks. It is important to us all that this work be done correctly!
larrym82
 
Posts: 30
Joined: Wed Nov 09, 2011 6:47 pm

Re: Minor technical items for discussion.

Postby larrym82 » Thu Nov 10, 2011 10:24 pm

Once again, thanks for all support. I have revisited my project and found that it performed 1% in just under 4 hours, and using one day less than the maximum deadline, I think my machine will do this. I will double check it on the 14 day deadline to make certain.

If the machine turns out not to be capable, is it worthwhile and/or possible to save the data gathered? I searched the forums but couldn't find anything on that topic.
larrym82
 
Posts: 30
Joined: Wed Nov 09, 2011 6:47 pm

Re: Minor technical items for discussion.

Postby Jesse_V » Thu Nov 10, 2011 11:01 pm

larrym82 wrote:If the machine turns out not to be capable, is it worthwhile and/or possible to save the data gathered? I searched the forums but couldn't find anything on that topic.


If your computer cannot make the Work Unit's deadline, then the servers at Stanford will reissue the WU to someone else. There are Preferred and Final deadlines, so if you cross the Preferred one you get credit but the WU is reissued, whereas if you cross the Final deadline the WU has already been reissued but even if you complete it you do not get credit. No, there is no need to save/email-off the data gathered. F@h takes care of itself, and its deadlines keep the overall simulation moving along.
Pen tester at Cigital/Synopsys
User avatar
Jesse_V
 
Posts: 2773
Joined: Mon Jul 18, 2011 4:44 am
Location: USA

Re: Minor technical items for discussion. (resolved)

Postby larrym82 » Fri Nov 11, 2011 3:02 am

Thank you Jesse. I've no idea what WU's are good for anyway, but as of now, I'm pretty sure that machine is going to have the job done by the first deadline, assuming a constant performance rate.
larrym82
 
Posts: 30
Joined: Wed Nov 09, 2011 6:47 pm

Re: Minor technical items for discussion.

Postby bruce » Fri Nov 11, 2011 5:06 am

Most classic WUs had a constant performance rate, with rare exceptions. (google the forum for "folding event" (with the quotes)).

The rate at which those exceptions occur increases for higher performance FahCores.
bruce
 
Posts: 21686
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Minor technical items for discussion. (resolved)

Postby Jesse_V » Fri Nov 11, 2011 6:13 am

larrym82 wrote:Thank you Jesse. I've no idea what WU's are good for anyway, but as of now, I'm pretty sure that machine is going to have the job done by the first deadline, assuming a constant performance rate.


Then let me see if I can explain. Basically, Work Units are the tiny fraction of the simulation your computer is being asked to perform. The WU is a packet of data, and when you get it your client uses the correct core to do all the calculations. The calculations are in the form of a loop, which is why you see things like "574/2000" or something like that. Every Work Unit has associated points and deadlines. So when you complete it, assuming you get it done before the deadline, you get credit for your efforts, and the cycle repeats. Each WU is unique. They are distinguished by its Project, Run, Clone, and Generation numbers. The Project is the particular protein being simulated, so it has a unique amino acid sequence. Here's how I currently see things:

You know the game of billiards? Here's a thought analogy to protein folding. Imagine taking some of the balls off of the table and placing them in the air, so that they are all touching each other. Imagine that they are all held together by rubber bands or something. You want to study how these balls interact with each other and whatnot, but your first step is to determine which set of balls you picked up. Say you happened to pick up the set {1,6,3,9,12,14,7}. You identify your choice with a Project number. Now that you have this set, how are you going to arrange them in the air? So you chose some particularly relevant layout, and this becomes your Run. Now you get all your friends together, give them all pool sticks, and have them each person get ready to hit some ball with it when you say "Go". But here you also have to decide how hard you want each person to hit a ball, and in what direction. Label your choice with a Clone number. And now you're just about ready to start your test. So you whip out your video camera to record the event, and say "Go!" Everyone hits their billiards ball, and everything starts wiggling and dancing around in space. You're recording this, so each second of your video you can think of as a Work Unit, and you number each second using a Generation number. Of course your not just doing one single test, you'd like to get a good feel for how all the variable interplay, so you do this again and again and again. You try a bunch of different choices of balls, a bunch of different spacial configurations, and your friends try all sorts of different striking positions and forces. You end up with a huge number of Work Units to do, but a large amount of knowledge in the end.

That was a simplified example, but I think of it like that. Of course we don't know how the protein will dance around in advance, that's what we're trying to figure out. So each unit of time is a Work Unit, and each much be computed before the one following it can start, which is why there are deadlines. Of course there are many optimization and special fancy things that the Pande Group does, so its not exactly like this. The "pool ball" represents atoms, which can be handled using a variety of techniques, not just like in my analogy. It should also be noted due to the way F@h divides up its simulations, it gets a near-linear speed increase for each person that joins. So if we have 50 people doing calculations science moves 50 times faster than if one single person was doing all the work. If you'd like more information there are plenty of places around to do so, but this should help you understand the idea of what's going on.
User avatar
Jesse_V
 
Posts: 2773
Joined: Mon Jul 18, 2011 4:44 am
Location: USA


Return to Linux CPU V6 Client

Who is online

Users browsing this forum: No registered users and 2 guests

cron