SPCR

I looked at the client on my home PC yesterday and was shocked to see 40/2500 completed.

What's the deal with the gigantic WU?

It's worse than that. My 2500 took the better part of 2 days to chug through one of these for the grand total of 50 pts. Ick. I was fine with normal units for a day and now I have another one of these monsters...

According to KDFold, its really a 100 frame unit. Can't say much more though, I had to take my system down after I looked at that.

The frames are made up of a bunch of little frames. Those little frames take like seconds. Chill.

man i just got one of those WU. Man it sure looks intimidating even though it maybe small.

I've had some take 80 hours. I get about 25 points a day for that.

Have you guys heard of EMIII? You could download it here:

http://home.comcast.net/~wxdude1/emsite/download.html

If you want info on your folding projects. It is packed with information.
# of frames
Work unit completion timer showing you how much time the work unit will take.
Frame countdown timer.
Remote function allowing you to see your other computers progress on your home network.
Jeez I sound like an advert I like it a more than KDfold.
My apologies if this is all old news to you, just thought I could help.

Yeah, the points given for the two 2500 point WUs that I did this weekend completely sucked ass.

roughly 50 points or so for a completed one...

Still, it took a ton longer to finish than a normal 400-500 WU.

I'm new to this, and still working on my first WU, it seems huge too. It says it has 400 frames, so far i have completed 72, and each frames seems to take half an hour! I worked out it will take 7 and half days to finish this WU! is that normal?

I'm running it on my laptop. It has a pentium III 1GHz processor speedstepped (underclocked and undervolted) down to 733MHz for silent heatpiped cooling at 100% load.

And as a sidenote, have people with "noisy" hard drives looked into using a ramdisk? It lets your harddrive sleep properly and is a great help. Otherwise i've noticed every completed frame seems to wake the harddrive even with noflushd. (thats a linux daemon to spin down disks and keep them off (even with writes) till a read cannot be filled by the cache.)

And noflushd, that another matter, cannot rave enough about it.

yes i've gotten that wu and i;m on a 2.1 ghz barton and it still takes me a whole 24 hour to complete one.

it is worth 70 points though

What is the project number for this large WU?

A 400 frame WU is a Tinker. It is the older non-optimized version. On a 800 MHz machine, 7 hours a frame is not unusual.

i believe it's p639..

The one i'm folding now, the 400 frame, 7 DAY one is p638_L939_K12M_ext.

In the messages it says:

TINKER: Software Tools for Molecular Design
Version 3.8 October 2000

Look here for project info: http://folding.stanford.edu/psummary.html Your project, 638, is a huge Tinker at 70.90 points. That would take days on many machines.

guess my barton is not so bad.

Running a Tinker would make anyone without a RISC processor feel bad. A Barton, according to the AMD model number, should be outperforming my processor. Do you have -advmethods in your command line? This flag tells folding@home to send you Gromacs WUs, which have AMD optimizations.

actually the gromacs optimizations will help P4s a lot more than AMD.

And even with tinker the AMDs are very good.

When I look at newegg, the lowest Barton I see has a model number of 2800, which should be better than my 2.53 Intel. However, if we look at the EOC weekly listing, my output is ~550 this week while DryFire is 40% lower at ~330. If these totals are typical, this leaves three options: either A) you need to reevaluate your opinion of AMD Tinker performance, B) someone needs to kick csnow for being so stingy on points, or C) we need to recommend DryFire seek counseling for his computergame addiction

The -advmethods flag does not request gromacs instead of tinker.

-advmethods says you are willing to run beta code, even though it may fail and crash your machine and give you no points.

Back when Gromacs was in beta test the -advmethods flag did get you gromacs work units, but that is not quite so true anymore.

heros- basically, the AMDs get a huge advantage over P4s with 2 FPUs available instead of 1 on P4s. However, P4s have SSE2, a special instruction set designed to get more floating point operations without more FPUs, and a SSE2 optimized floating point intensive program can actually perform better on P4s than Athlons. Therefore, Athlons will win on Tinkers with their second FPU while Gromacs with SSE2 optimizations will go faster on P4s. Furthermore, CPU speed is hardly the only factor in the speed of WU; FSB, Memory, and other things will affect it too. And, most importantly, points for WUs can vary wildly, meaning a one week snapshot is hardly a indicator of performance, especially since dryfire may have had downtime, or used his computer for something other than folding (though that better not be true, dryfire!)

so, AMD people get nothing from using Gromacs?

hey heros i have a big gaming addiction. you guess it.


that's why when i;m asleep and at school i submit like 2 wu and when i;m home i submit like 0

or ont he weekends i submit severly less.

edit: also my comupter was down for repairs this week aswell.

Gromacs probably contains optimizations that benefit AMD users a lot too, but the P4s get a huge benefit from SSE2. Which, by the way, Athlon64s do have.

You seem to be missing what I am saying, Zhentar. I couldn't care less about your processor war debate. Whether AMD is better at Tinker doesn't matter, its history. I was talking about Tinker versus Gromacs. It is a huge difference for Intel, so I was wondering why you thought AMD was good with Tinker. I did not know that there was little difference between the two cores for AMD users, besides, DryFire said he was tapping the vein, so chill.

tapping the vein? what is that?