Toggle sidebar Toggle sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

F@H: Production back on the rise

shhhhhhhh.

Don't jynx it! 😛

Also lets see if those Gromac core does anything better for the mac based teams!
 
Three straight days on the up-ity! and our projected line is about level now, that's another positive sign.
 
Re: F@H: Production back on the rise

Originally posted by Rower_CPU
Now that our PC lab is back up we're back on an upswing in daily production.

Let's see if we can get back into the yellow! 🙂

http://folding.extremeoverclocking.com
Click to expand...

Now, if someone can convince Virginia Tech to kindly donate some CPU from their new 1100 machine Dual-G5 cluster supercomputer. We could be in the red.
 
Re: Re: F@H: Production back on the rise

Originally posted by vannote
Now, if someone can convince Virginia Tech to kindly donate some CPU from their new 1100 machine Dual-G5 cluster supercomputer. We could be in the red.
Click to expand...

😉

We can always dream...
 
Put fully into folding, it seems like a computer of that magnitude would tear through even 70 point WU in a matter of seconds or minutes!

Is there anyway to find out how many flops my lil' toppy it cranking out??
 
I wanna pass Knights Who Say Ni!

Has anyone been tracking the various teams to see if there is any hope of passing those in front of us? I wanna pass SOMEONE!

BTW I am curious- what is a typical time for a pentium 4 say 2 GHz machine to complete a frame in a 72 point tinker? just curious.
z
 
If my DPG5 is a rough indication, the Big Mac cluster would produce about 165,000 points a day. If all the processors were working on the same WU (which might not be the case, as with current DP machines), you'd be finishing about two work units a minute.
 
2 weeks

I am in a training class for 2 weeks, so I have configured the machines here to all run folding. 😎
Instead of 100pts/week or less, I have gotten 3000 in the last 7 days! 😀

Not sure how long they will leave them running, but unless there is another class, it could be for a while...
 
Re: Re: Re: F@H: Production back on the rise

Originally posted by Rower_CPU
😉

We can always dream...
Click to expand...

We sure can. Seems to me that one 24/7 week would probably put us all but at the top of the charts.
 
Re: I wanna pass Knights Who Say Ni!

Originally posted by frankzeg
Has anyone been tracking the various teams to see if there is any hope of passing those in front of us? I wanna pass SOMEONE!

BTW I am curious- what is a typical time for a pentium 4 say 2 GHz machine to complete a frame in a 72 point tinker? just curious.
z
Click to expand...

The Knights passed us some time back and are spreading the gap even as we speak. Though if this new F@H client improves everyone else as much as it seems to be improving my output we'll be passing a lot of people in the next couple months. My output has nearly tripled with this new client though the long term affects will take a while to be realized.

A good place to go to see how we compare to others with graphs and number is.

http://folding.extremeoverclocking.com/
 
Combining has been discussed with a couple of the teams, but no-one ever seems to want to. I guess people feel more strongly about folding for their site than for promoting Macs.
 
i stopped using f@h back when in installed panther (some time ago). The folding at home client seemed incompatible with panther. seemed like its priorities were wrong so that it slowed everything else down. has this been resolved in the newest client?
 
Originally posted by AmbitiousLemon
i stopped using f@h back when in installed panther (some time ago). The folding at home client seemed incompatible with panther. seemed like its priorities were wrong so that it slowed everything else down. has this been resolved in the newest client?
Click to expand...

I think it was a problem with Panther and not the client. If you use mc68k's install script it will be installed with a +20 nice setting. As far as I can see there is no noticeable slow down from folding. At least not any more then there ever was with Jaguar. I still pause folding when I play UT2003 as it does make a 5fps difference or so which is important when you are just toeing the line with the large outdoor levels. As for everyday use no noticeable slow down what so ever.
 
i use the gui client (don't like messing in the terminal). I just run it headless so there is no window or dock icon or anything to slow it down.

I never noticed a slow down in jag (i too would stop it if i played a game usually) but in panther simple things like opening apps and moving windows felt sluggish. i was told panther changed the priority settings to a broader range and that that could explain the problem. I hoped that when panther was released we would see a new f@h client that mentioned better panther compatability but haven't seen that specifically stated.

Anyone use the gui client who could verify that it now works well in panther?
 
Originally posted by AmbitiousLemon
i use the gui client (don't like messing in the terminal). I just run it headless so there is no window or dock icon or anything to slow it down.

I never noticed a slow down in jag (i too would stop it if i played a game usually) but in panther simple things like opening apps and moving windows felt sluggish. i was told panther changed the priority settings to a broader range and that that could explain the problem. I hoped that when panther was released we would see a new f@h client that mentioned better panther compatability but haven't seen that specifically stated.

Anyone use the gui client who could verify that it now works well in panther?
Click to expand...

Why use the GUI client? Jut use mc68k's install script and fold and forget. It will just crank away on it's own. If you need to stop folding to play a game or such you just open the terminal and type pause. When you want it to start again open the terminal and type resume. If you want to see your progress you can either check the .txt output that you have in the F@H folder in your user folder or you can open the terminal and type work. Also the terminal client will run faster then the GUI client from what I have heard even if you are running it, "headless," as you put it.

As far as a new client there is a new F@H client for Gromacs (Altivec) units that is much much faster. None of us believe it was rewritten though simply recompiled with the latest GCC compiler. I don't know of any reason why the client would need rewritten at all for Panther. If you are comparing how it ran on a prerelease version of Panther then it was most definitely Panther and not the client that was the problem.
 
I prefer to use the GUI client. It is easier and I haven't found the terminal version to be any faster. Its just easier to click quit or launch to quit or launch then open up a terminal session and tell f@h to quit. Also its easier to just look at a progressbar to see how much it has done. I'm not much of a fan of the terminal. I use it when I need to(like to modify apache settings or something) but don't like running applications in it.

I'm using a store bought copy of panther now and still find the old client to slow down the big cat. I just want to know if someone can tell me if the new one works properly (or maybe if someone can give me detailed instructions on how to renice the f@h gui client)
 
Originally posted by MacBandit

...

As far as a new client there is a new F@H client for Gromacs (Altivec) units that is much much faster. None of us believe it was rewritten though simply recompiled with the latest GCC compiler. I don't know of any reason why the client would need rewritten at all for Panther. If you are comparing how it ran on a prerelease version of Panther then it was most definitely Panther and not the client that was the problem.
Click to expand...

None of us?

Well, I've read for myself from Vijay Pande and others in charge at Stanford that it had been re-written to an extent that there was more AltiVec code and the remainder of the code was better-placed and the whole thing was re-compiled with gcc 3.3.
 
Originally posted by bousozoku
None of us?

Well, I've read for myself from Vijay Pande and others in charge at Stanford that it had been re-written to an extent that there was more AltiVec code and the remainder of the code was better-placed and the whole thing was re-compiled with gcc 3.3.
Click to expand...

Well no one that had mentioned anything until now. So my post still stands though it is now outdated. 😛

Good to know they put more work into it then simply recompiling.
 
Lemon -- the command (script) uses less CPU overall around .4% up to 4% CPU less then the GUI.

All around this is nothing, but over time it amounts.

Basically if your running it headless and want to squeeze more power out of the machine the Script is the way to go.
 
I can tell there are a tremendous number of people doing this now... I used to be near the top of the stats, and now I've dropped to 167th place.

😎
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back