SPCR

Rusty075 wrote:After about 30 seconds I was reduced to nodding and saying, "uh huh, yup" just so she wouldn't realise she had completely lost me.

Don't worry, Rusty. That's a pretty standard male reaction to anything girls talk about. I quote the Immortal Homer Simpson, "Sure they will." If thou knowest the Simpsons, thou knowest to which I refer.

It is similar to Finite Element Analysis of a glob. Developing an equation to solve for the physical properties of the entire glob is a nearly impossible task. However, if you look at an tiny section of the glob, the equations boil down and become simple and easy to solve. Unfortunately, those simple calculations need to be solved for every single tiny piece of the glob. So instead of a huge nasty problem, you now have millions of tiny problems. The trick in all of this is finding a bunch of suckers willing to work on all those millions of problems for you. That is where we come in.

herosformula wrote:The trick in all of this is finding a bunch of suckers willing to work on all those millions of problems for you. That is where we come in.

SIB
SUCKERS IN BLACK
Protecting Humanity
From The Glob Of Life

alright.

its been about 24 hours since i started using the CLI client with gromacs, and i've churned through 3 WU's. two 250k step WUs and one 500k step WU. (the 250k step WU takes about 7-8 hours. while the 500k step WU takes about 15 hours.)

i think that's impressive, don't you think? i'll heed to the fact the CLI client is especially faster since you can specify your preference for gromacs. but when it comes to tinker units, its just about the same as the graphical version.

i'm finally in the top 20!!

The work units are no where near the same sizes, look at this page http://folding.stanford.edu/psummary.html My first Tinker core I was able to process a frame in about 4 minutes, it was worth 23 points. Then I had a couple of Gromacs, at 3.5 minutes a frame, each worth 18 points. My current unit is a huge Tinker, 7.5 minutes a frame, but worth 60 points.

doing a quick and dirty calculation, i ended up churning out 69 points in less than 24 hours with gromacs... is that good or bad?

It seems about right, riff--I mentioned elsewhere that it's averaging about 12 hours for my P4 2.4 to do a 33-unit Gromacs WU. I'm not using particularly fast memory and my mobo is an Asus P4PE. I'm not sure if any other factors affect folding speed.

Assuming the computer isn't occupied by doing other things, of course.

in my enthusiasm, i miscalculated. those 69 points were for 32 hours. i average about 51 points per day. (one 18point WU and one 33point WU) the 33point WU takes me about 15 hours while the 18point is just under 8 hours.

How's everyone getting their points / WU totals?

riff wrote:i'll heed to the fact the CLI client is especially faster since you can specify your preference for gromacs. but when it comes to tinker units, its just about the same as the graphical version.

Actually, you can make the graphical version do gromacs too. Just create a shortcut of winfagh.exe and add the -advmethods -forceasm in the shortcut.

i'm finally in the top 20!!

Im in pursuit.

no fair.. you've got three more processors than me!

Erm, of which one p2 and one p3-450... not exactly state of the art. The p2 needs over a week to complete 1 WU.
And 2 of em refuse to run Gromacs for some strange obscure reason.

fair enough... BRING IT ON

EDIT: thanks for the tip on the graphical client

/me runs off to search for more PC's to hijack at work.

Wrah wrote:/me runs off to search for more PC's to hijack at work.

That's hilarious, you're like a guerrilla sniper or something.

Wrah wrote:/me runs off to search for more PC's to hijack at work.

I had the same thought while at class the other day. Here I am, sitting in a room with 20 dual-Xeon 2.4Gig HP workstations, down the hall there are 80 more machines, the slowest ones beings P4 2.0's.

Hmmm..... Imagine the folding potential.

Now if only I could get admin rights on all of them.....Muwahahaha!!!!

Well....maybe. But so what if I install Folding and an Olson twins screensaver on all thir machines? I don't think anyone would mind, do you?

Wrah wrote:/me runs off to search for more PC's to hijack at work.

ACK!!! okay, okay, i give - you win... :bitter scowl:

Found 2 more unused PC's, sorry.

hehehe... its all good. i'll kick your butt some other time

I could increase my output by almost 60% if experiment #537 would leave me alone in favor of #542. #537 on my computer takes 3 1/2 minutes a frame and is worth 18 points. #542 takes 4 1/2 minutes and is worth 33 points. So, #542 gives me a 22% increase in workload, but results in 91% increase in points per WU completed.

Yes, I noticed the same. If you look on this page, you'll see that the 33 pointers connect to a differtent IP address then the others. This raises 2 questions.
1. What would happen if you block this address with your firewall? If it uses this IP to download the project and not solely to return results to, you could prevent getting these less profitable projects.
2. Would this be cheating or not?

arg, another #537 - I'm dooooommmmmed.

If the "dice" program just tells your program which IP number to get its next WU from but does not check to see if your computer can actually receive data from that port, then blocking that port would just lock up the program.

Or perhaps it might simply contact the assignment server again for a new project. Have you tried it yet?
It is possible that project servers are offline once in a while, I would find it strange if they haven't anticipated on that and the client would sit around doing nothing until the server is back up again.

You can also just delete the work directory when you see it has just downloaded a 537.

riff wrote:actually its your FPU, IPC and cache.

this is what i've found out so far:

the max operations per clock cycle for athlons & durons is 9, while for pentium 4s and celerons is 6. the older celeron's were 5. so clock per clock, amd cpu's are better for folding. higher the max IPC of the processor, the better. IPC=operations per clock cycle X speed of processor.

the FPU for athon xp's is also faster than the celeron's and pentium 4's. again, higher the FPU, the better.

more the cache, the better.

um.. and that's about all i know.

EDIT: had to check my numbers

EDIT: linkie http://www.cyberbreakcafe.com/jpj_compu ... uvscpu.htm

Sounds like the Bartons are pretty sweet for this.

Athlon Powers, EXACTLY!

Everyone, time to upgrade!

I hear over at hardforums that the bartons fold like crazy over and over again.

Now i wonder how opterons fold...