Slo-Tech is doing really good! They are hanging in there. They seem to lose ground on weekends.

Germany is a different story. I hope they can get more support, otherwise they will not be able to fend off Slo-Tech.

SPCR has 545 "active" cpu's, very active I might add!

We conquer Germany tomorrow!
A 0-value post this. But I couldn't stop myself - it is something I always wanted to say

That's funny.. but it gets weirder though - come Dec 22nd we'll be conquering a bit of the USA.

EDIT: Sorry, that should be ABIT USA

Sniff... You guys have already conquered my heart.

Germany has, as of this hour, completed 99,271 WU to SilentPCReview's 36,631. 2.7 times more work units completed than SPCR. We have 92 active users to Germany's 125. They have us outnumbered. Their average points per WU is 13.6. But looking at yesterday's totals, their average was 47.6 points per WU to our 46.6, so virtually identical. It doesn't add up. How the heck do they have such a low average? Early on they must have done a lot of tiny WUs worth next to nothing. Now they must be tackling the same WU's we are, but it's taking them a lot longer to complete them. Their 125 users completed 77 WUs yesterday, whereas SPCR's 92 users completed 211, worth virtually the same amount per WU. Their average cpu must be way smaller than SPCR's, or maybe they run them fewer hours per day.

These statistics remain interesting only until we pass them, which should be about 12 hours from now.

Then it's on to Abit and AOA, not necessarily in that order, all the while trying to stave off Slo-Tech. Abit will be a little tougher since their trend line is up, whereas Germany was merely holding steady, so with Germany it was simply a matter of time. Likewise AOA is simply a matter of time; their production has fallen off dramatically over the past month, from nearly 8000 per day to less than 5500.

Darn, the stats are updating so I guess I'm done!

David

About the low points per WU average of Germany I have a theory.

For example in our team you can look at Siili (#99) he has completed 225 for 2133, that's less than 10 points per WU. He has a lot of WUs "finished" of 1.5 , 2.5 etc. And the reason for that I think is due to instability, it happened to me sometime that I had my memory with a little more agressive timing and a bit of overclocking. It happens mostly with 3xx proteins and 10xx I think. In a log file it appears as some failure, so the program just send what it has completed and Stanfords gives you points according to the percentage done.That's why it appears that you completed WUs for less than 5 points or so, which doesn't exist. So maybe Germany have a lot of members which have the same problem.

What I don't know is if this uncompleted WU are useless or not, I guess that if Stanford gives you points for them, they worth it. However, I don't like the idea of not finishing my WU, so I went back to my original timings.

It's just a theory, if you like to discuss about it, you're welcomed.

Until recently (about September) Stanford gave zero credit for WU that were not complete. They give credit now as a favour, not because partial results are useful. IIRC in the old days your WU counter moved up, but your score stayed the same.

Another contribution is many of the old teams/folk didn't pick up on the potential gains in points per day offered by Gromacs if you have 3DNow or SSE. If you look at the Stanford projects pageyou will see many low value Tinker projects.

Although in Siili's case I think you are right: just the 62 WU 'completed' for project 638 should total nearly 4400 points!

If your system is failing to complete Folding units then IMHO it is desperately unstable and like mormakil says you should fix it. 'Folding stable' is less rigourous than 'Prime stable'.

Germany has fallen. SilentPCReview has overtaken the #27 spot in the rankings according to extremeoverclocking.

We are back down to green. Arg! We just need a wee bit more production to stay in the yellow, and keep climbing in the rankings.

David

Yeah yeah... I'm feeling part of that is my loss of over 1000pts/week. <sigh>

Yep, noticed that we slipped below 10k pts per day. Great that we crushed our BMW friends tho! Too much beer drinking and not enough folding!

On a more serious matter. My average is about 44.9 pts per wu and I did notice, by accident, that I got credit for 10 pts only on 1 wu - I think today. Anyone able to hazard a guess if that's within the area of acceptable average?

How would I go about figuring out which is the problem computer?

Congrats btw on breaking 10k in the last week.

If you are running Electron Microscope it is easier: on the Options page if you click on the Queue.dat icon (left of the Frame Progress button) and scroll through the last 10 WU. Normal is either "Current" or "Finished and sent".

Some 'duff' units need you to analyse the log files - in EM the middle icon, any other way you are running just open FAHlog.txt in Notepad. You should see a nice pattern climbing up from step 1 to step 100. You can also look at FAHlog-Prev.txt (manually) if you want to go back further in time!

But you may also have just had a 10 pointer WU If you look at the current projects there is P618 that is 10 points at the mo.

My hats off to the team! Obviously another milestone reached. Victories are far tougher and slow to come now compared to early days.

Don't think my contribution has been that significant recently, but I am afraid it will drop even further over the next week. Unplugged all my systems while I am away to away any disasters. Unless I can borg some unspecting machines while "abroad"...

My production has slipped a bit the past few days due to my power supply fan issues (i.e., me busting it then having to replace it), so I've been doing about half my normal points.

That should get back to normal soon. And I need a new project, so might just build myself one of those multiple-board folding farms...