8th Annual Holiday Season Folding@Home Race! The Race is on!!

[geecee]

To the new racers with Nvidia cards: The latest nvidia drivers show good OpenCL folding performance for the GTX780 and higher cards. However, the older cards, GTX6XX and earlier take a big hit in performance with the latest drivers. If you go back to dirver version 327.23 drivers or earlier, you will maximize your performance for the "older" cards.

Thanks biodoc. This is good info. Unfortunately, I updated my nvidia drivers right at the start of the race, and I have a GTS450.

Maybe when I get home tonight, I'll see if I can find the older drivers on the nvidia website.

 

[biodoc]

Thanks biodoc. This is good info. Unfortunately, I updated my nvidia drivers right at the start of the race, and I have a GTS450.

Maybe when I get home tonight, I'll see if I can find the older drivers on the nvidia website.

My GTX460 only gets the core 15 WUs. Let me know if you get any core 17 WUs on your GTS450.

My GTX570 can crunch zeta WUs at about 30K PPD and my GTX660Ti will do zetas at about 65K ppd (p8900 WUs).

 

[deerslayer]

I should show up in the stats today.

http://fah-web2.stanford.edu/cgi-bin/main.py?qtype=userpage&username=deerslayer&teamnum=198

 

[Pokey]

Maybe when I get home tonight, I'll see if I can find the older drivers on the nvidia website.

Nvidia download page has a link to "Beta and older drivers". Go there and look for the 327.23 driver. That is the commonly recommended version to get full production.

 

[Markfw]

Nvidia download page has a link to "Beta and older drivers". Go there and look for the 327.23 driver. That is the commonly recommended version to get full production.

Here:
http://www.nvidia.com/download/driverResults.aspx/66700/en-us

 

[biodoc]

December 23 stats @ ~21:00 UTC

The Scientists are now 1,683,701 points ahead of the Thinkers.

 

[deerslayer]

I'm not impressed with my output.

 

[Markfw]

If they would give me back my ZETA units to non-linux, we would have a chance. As it is, I am loosing almost 200k ppd per day.

 

[Pokey]

If they would give me back my ZETA units to non-linux, we would have a chance. As it is, I am loosing almost 200k ppd per day.

I'm surprised. I am getting 8900s and I am all Windows here.

You must not have been a good boy this year.....................()

 

[deerslayer]

Is there a link for average PPD for different CPUs and GPUs? I set it up on my old computer in the basement and the WU I got says it's going to take over 7 days to complete!

 

[Markfw]

Is there a link for average PPD for different CPUs and GPUs? I set it up on my old computer in the basement and the WU I got says it's going to take over 7 days to complete!

Install HFM.net... It will calculate that for you.

 

[Markfw]

I'm surprised. I am getting 8900s and I am all Windows here.

You must not have been a good boy this year.....................()

I just got ONE (of 4 usual) for the first time in 2 or 3 weeks.

 

[deerslayer]

Install HFM.net... It will calculate that for you.

Is that a replacement for the F@H client? I've reverted back to DC noob status.

 

[biodoc]

Is there a link for average PPD for different CPUs and GPUs? I set it up on my old computer in the basement and the WU I got says it's going to take over 7 days to complete!

You'll need to finish 10 WUs before you qualify for bonus points.

Here's a GPU database for the old core 15 WUs:

http://compdewddevelopment.com/projects/fah-gpu-statistics-database/

And one for the core 17 WUs:

https://docs.google.com/spreadsheet/pub?key=0AmQ6YZGMTUpFdHhxNkptVncyR2k4dVV6WDBmQXRJYVE&output=html

And here's the bonus point calculator for all WUs;

http://www.linuxforge.net/bonuscalc2.php

 

[deerslayer]

You'll need to finish 10 WUs before you qualify for bonus points.

That's probably why my numbers seem so low. Apparently I did F@H at another time too because I already had a user name but didn't realize it. Oops.

http://fah-web2.stanford.edu/cgi-bin/main.py?qtype=userpage&username=deerslayer_(TeAm_Anandtech)

 

[biodoc]

Is that a replacement for the F@H client? I've reverted back to DC noob status.

No, HFM.net just monitors your clients and gives you ppd plus a bunch of other stuff.

 

[Markfw]

Is that a replacement for the F@H client? I've reverted back to DC noob status.

That OK, we LOVE noobs of any kind here....

 

[biodoc]

The scientists are 1,683,701 points ahead of the thinkers. There are 9 days left in the race so the thinkers would have to produce on average 187,077 more ppd than the scientists to tie the score at the end of the race. This seems unlikely.

I'm thinking of moving VirtualLarry to the Thinkers and moving FullMetalChocobo (Chocobo Navy) to the Scientists. The lead that the Scientists have would be cut from 1,683,701 to about 120,000. Also based on ppd, the thinkers would gain 50,000 ppd.

What does everyone think about this change of team members? VirtualLarry and FullMetalChocobo, would both of you be ok with this change?

Peter, what do you think?

We could go one more day with the current teams but I'd like to make the change soon.

If you have any other suggestions, let me know. Remember, if a racer is moved to another team, ALL his production points go with him to the other team.


EDIT: Here's what the rebalanced teams would look like

 

[Sunny129]

The scientists are 1,683,701 points ahead of the thinkers. There are 9 days left in the race so the thinkers would have to produce on average 187,077 more ppd than the scientists to tie the score at the end of the race. This seems unlikely.

well first consider this - the TeAm member borging my dual 7970 host for the FAH race is folding for the underdogs, the Thinkers. since yesterday afternoon, the estimated PPD of my 7970s has been fluctuating between 185K-187K PPD. i know that may be just barely enough to make it a photo finish...or it may just be enough to fall short of winning, in which case i could bump up the core clock on one or both of my GPUs...again, my cards have been folding since yesterday afternoon, so i'm not sure if your projection calculations included any of my production or not...if they did, then i'm afraid some reshuffling of the teams may still be in order.

 

[petrusbroder]

Biodoc, first a cordial wish for a Merry Christmas.
As I see it I would have done the same change to balance the teams. There is a point in having the teams balances at the onset of the final week: if one team is to far ahead, some racers in the past have left the race. And that is a loss for the whole TeAm AnandTech (#198). I like your proposal!

 

[biodoc]

Thanks for your analysis and recommendation Peter! Merry Christmas to you too!

Sunny129, I see your point but the Thinkers, on a daily average, were already running at a significant deficit so I think your contribution helps solves that problem. I'm also seeing significant increases in production by at least one other member of the Scientists too.

Ok, I'll make the change to the teams in the Dec 24th stats and thus team composition is finalized.

Good luck all!

 

[biodoc]

For those of you new to F@H, here's the stats site that Peter and I have been using:

http://folding.extremeoverclocking.com/team_summary.php?s=&t=198

Here's another one you may find useful.

http://kakaostats.com/t.php?t=198

 

[VirtualLarry]

What does everyone think about this change of team members? VirtualLarry and FullMetalChocobo, would both of you be ok with this change?

Fine by me...

 

[Pokey]

Ok, so we lose one and pick one up.
I can't seem to find "FullMetalChocobo" or "Chocobo Navy" on the EOC stats. Crunching under another name?? or am I spelling it wrong?

Answered my own question............. did not catch the underscore in the second name.............. darned underscores..........who does that anyway?? :whiste:

 

[Pokey]

Merry Christmas everyone.