THGC Needs You -Team 40051

Page 32 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
My current 404 pointer is going to take about 50 hours.

What kind of system are you running? My trusty little venice is at 2.7Ghz (2.85 when i get aroudn to restoring the oc today) and it still takes me almost 70 hours for a 404 WU. Can't figure out why. Whats you benchies score from F@H?
 
My current 404 pointer is going to take about 50 hours.

What kind of system are you running? My trusty little venice is at 2.7Ghz (2.85 when i get aroudn to restoring the oc today) and it still takes me almost 70 hours for a 404 WU. Can't figure out why. Whats you benchies score from F@H?

I got a Winchester 3200+ in here at stock speeds. It's doing 10,000 frames at about 16s-17s per frame....supposedly.

I don't know what my benchies are but if you're doing it in 70 hours you should be even more pissed off than me. 😛

I don't monitor 2 of my machines but they're both obviously going really slow as well. My machine at work is a 2400+ Thoroughbred and it's getting lame WU's as well.
 
I'm getting 46mins and 8 seconds per frame on 404's. Which is like 78 hours on it. On 202's I get 26 mins 39 seconds or about 44 hours. Which is why I said I might go to deadlineless tinkers (4 min 30 seconds a frame / 28 hours) for 239 points.

For reference a 600 is 36 mins 1 second a frame which works out to 60.1 or so hours.
 
Everyone needs to keep their "frames" straight. As defined by Stanford, all WU's (except a few tinker cores that have 400) have 100 frames, which equates to a "step" in the log file. I understand where you're getting the numbers you are. Gromac log files look like:
[code:1:9929b99d0e][16:46:27] Completed 3400000 out of 10000000 steps (34)
[16:55:27] Timered checkpoint triggered.
[17:04:27] Timered checkpoint triggered.
[17:13:27] Timered checkpoint triggered.
[17:17:20] Writing local files
[17:17:20] Completed 3500000 out of 10000000 steps (35)
[17:26:20] Timered checkpoint triggered.
[17:35:20] Timered checkpoint triggered.
[17:44:20] Timered checkpoint triggered.
[17:46:29] Writing local files[/code:1:9929b99d0e]
For the sake of arguing, this is a 100 strep WU, that is currently on step 36.

@dluckle if your machines are on the same network, you should really download EMIII. It will let you monitor all of your clients, it gives relatively accurate times, let’s you read your benchmarks or log files from every machine it monitors, and it has a small footprint.
 
Everyone needs to keep their "frames" straight. As defined by Stanford, all WU's (except a few tinker cores that have 400) have 100 frames, which equates to a "step" in the log file. I understand where you're getting the numbers you are. Gromac log files look like:
[code:1:e9b71b7cba][16:46:27] Completed 3400000 out of 10000000 steps (34)
[16:55:27] Timered checkpoint triggered.
[17:04:27] Timered checkpoint triggered.
[17:13:27] Timered checkpoint triggered.
[17:17:20] Writing local files
[17:17:20] Completed 3500000 out of 10000000 steps (35)
[17:26:20] Timered checkpoint triggered.
[17:35:20] Timered checkpoint triggered.
[17:44:20] Timered checkpoint triggered.
[17:46:29] Writing local files[/code:1:e9b71b7cba]
For the sake of arguing, this is a 100 strep WU, that is currently on step 36.

@dluckle if your machines are on the same network, you should really download EMIII. It will let you monitor all of your clients, it gives relatively accurate times, let’s you read your benchmarks or log files from every machine it monitors, and it has a small footprint.

You are correct when I say frame I mean frame not steps.

EDIT: That looks like one of the first (sweet) 404's. On another note I use Fahmon to check my progress is EMIII better?
 
I actually run both. EMIII is great with a few clients 1-4; it lets you edit the .cfg file and what not. More than that and I prefer FAHmon, even though the ETA's are a bit less acurate, EM's display can't handle many clients (@4 it cuts off the bottom one, @13 it cuts off the bottom 3 and doesn't display anything but the basics anyways). Lately, I have found myself starting EMIII less and less.
Niether of the programs are very "refined", but they both beat manually browsing to the log file and using a calculator.
 
Anybody getting the Lambda 5-ways?

It's taking my 3ghz xeon about 5days to finish one...for a measely 400pts...And it's running on almost every machine i have. Meh! My 1.3 P4 is taking 7 days for the half size ones...worth 200pts. And i take that back, it isn't 5 days to finish...My Xeon 3.0, has 36hrs to go and is 70% done, for a 400 pointer. That's terrible, 10 days for 400pts...bah. No wonder my avg is down from 1500-1700 to 1000. ( >< )
 
Just added "2" more procs to my aresenal. I am home for spring break and stealthily slipped F@H onto my mothers computer. Its a HT enabled P4C so I had to install F@H as 2 services, otherwise I was just getting 50% core utilization, oh well. I'm just glad it seamlessly integrates w/o any windows as a service and doesn't degrade performanc just causes the CPU to run at 56C hehe. Thanks mom!
 
You've got to be kidding me.... i got a 20k frame Wu that is taking 1Min a frame, im so screwed lol. Does stanford hate us all of a sudden?

Edit: Oops that was the estimated time while i was playing Battlefield 2 LOL

Down to 13sec a frame :)
 
This may sound like a stupid question, but how do I get F@H to run on a linux system?

I just finished setting up another PC with Mandriva Linux on it to play around with, and would like to install Folding@Home on it but don't know how to make it run. I have downloaded the exe file from the Stanford website, but am lost from there. BTW, I am a total Linux n00b, if you couldn't tell already.

Thanks.

P.S. I updated the Toms Hardware Guide Community Folding@Home web page again here. Check it out.
 
This may sound like a stupid question, but how do I get F@H to run on a linux system?

I just finished setting up another PC with Mandriva Linux on it to play around with, and would like to install Folding@Home on it but don't know how to make it run. I have downloaded the exe file from the Stanford website, but am lost from there.

Check here:

http://fahwiki.net/index.php/Running_the_FAH_client_on_Linux

or check this thread I found (its about Ubuntu but its still linux)

http://forum.folding-community.org/viewtopic.php?t=14440

I really don't know though since I've never tried running FAH on Linux.
 
are you running all those computers as a farm? sweet


Btw, does sjonnie even come on these forums? i don't think i have seen him in all my time on the forums (less then half a year)
 
Hehe, Seems like Sparky and Cat and Me Are pretty Eager Canadians wanting to help out with the folding and Team TOms Hardware. yay i just got upgraded. i'm a higher middle class folder now.
 
No, those aren't a farm. Some are computers here at home, some were computers at places I USED to work.. I'm sure by now some are non-functional.

As for sjonnie, I haven't seen him on these forums in a LONG time. Not sure if he still visits or not.
 
I seem to have added a golden CPU to my CPU pool. I'm not sure what I exactly did when i configured the F@H on my mom's comp but that thing is churning out 200pts a day almost. Its a P4C 3Ghz w/HT. I have on F@H on each "core" and its just terring apart these little 56 point WU's in about 14 hrs per core. So about 200 pts in 28 hours isnt bad from 1 CPU :)

Edit: Where is that guide to setting up a farm with a diskless client, network boot, and server?
 
Well if anyone is interested in a farm, I'm going to be eBaying a lot of ipaqs soon (shameless, advertising URLs to follow). I think I have way more than I want to run/refurbish/house/pay for.




PM if you guys are interested in a preconfigured farm. I'd give anyone here a really good price as long as you are folding for THG.


Get this they just sent my AMD64 an AMBER core Project 1808 for 153 points. Takes 25 hours, but WTF is Stanford doing these days?

@SuperFly03 For the 3rd time (at least) that I've posted this FOlding Farm Link

I swear to phuking god if they don't give us a section in this forum soon, I'm going to start my own team with their own forumS
 
It looks like conroe could turn out to be a folding monster.
I'm not sure how optimized Stanford’s Gromacs core is, but according to current info on Intel's "core" procs (article here) it will execute SSE instructions in 1 clock cycle instead of 2.
 
Hello Mr. Cunninglinguist (love the name :lol: )

Mind if I fold for 40051? I've got three machines I can throw into the mix. Are you gonna need machine stats or does that really matter?