As the Work Unit Crunches

Come join the... uh... er... fold.

Moderators: just brew it!, farmpuma

Re: As the Work Unit Crunches

Postposted on Tue Jun 15, 2010 5:23 pm

--- edit ---
Ragnar Dan wrote:All my machines have been getting those 921 point P6701 WU's lately, except my Opty 165 which got an A1 core P5102 (3340 points).

The problem: my old X2-3800 running at stock, now, won't make the deadline. If I'd ever had success transferring SMP WU's from one machine to another and back again, I might consider moving it to another machine temporarily and then moving it back for the final 1% and upload, but I've usually gotten a checkpoint error and a restart from 0% after 2 transfers. Bah. I could quickly buy a mobo for my old C2D 6400 and move it to that thing, and it might make it. But I'd have to get the thing within the next 2 days, and even then it might not make it (as of this typing under 55⅔ hours until bonus runs out and the WU is re-issued by Stanford, and after less than 21.25 hours it's already ~3⅓ hours behind).

Stanford has basically eliminated any possibility of sneaker netting on nearly all of the newer work units. I blame some paranoid IT twit for fixing a problem (duplicated WUs) that was already detectable, but maybe now they think they don't even need to check for them. A secondary downside for we the contributers is the server issuing duplicate WUs that might not be noticed by us and thus wasting a full work unit processing cycle. I've heard that sneaker netting is still possible as long as the Stanford issued system id numbers are identical (move the entire folding folder?), but I have not bothered to try it.

--- edit 2 --- "Bonus points" constantly decline and then finally expire completely at the Preferred deadline (WU reissued), but the basic (listed) points are still fully available until the Final deadline.
Last edited by farmpuma on Tue Jun 15, 2010 5:34 pm, edited 2 times in total.
Reason: added quote for continuity e1 and deadline def e2
Image Image
.* * M-51 * *. .The Whirlpool Galaxy.
farmpuma
Minister of Gerbil Affairs
Silver subscriber
 
 
Posts: 2311
Joined: Mon Mar 22, 2004 12:33 am
Location: Soybean field, IN, USA, Earth .. just a bit south of John .. err .... Fart Wayne, Indiana

Penalties of the A3 "Bonus System"

Postposted on Tue Jun 15, 2010 6:05 pm

Penalties of the A3 "Bonus System"

Current Penalty -
Since 8:15 AM ET 15 June 2010
With the 171.64.65.54 server (again) STILL down and the collection server repeatedly dropping the upload at 2.2MBs my 6052 work unit is rapidly losing bonus points. It has the potential of zero bonus points and eventually becoming a complete waste of resources.
BTW - This 6052 WU froze my mouse about halfway through. However I was able to work around the lack of mouse after the WU finished and did a soft restart.
Upload finished - 4:25 PM ET 15 June 2010 = Eight plus hours of bonus lost.

Worst Penalty -
"Bonus points will be awarded after the completion of ten work units." As I finished my tenth WU I was pretty excited. I was expecting a bonus in the 10,000 point neighborhood! What posted was a bonus for ONLY my tenth work unit. Net penalty of about 9,000 points.

Misc Penalties -
Electric power blinks and outages - mild to disastrous

Lightning Storms - mild to substantial - I will let my systems crunch through hell and high water, but an active phone modem can be killed by even an only marginally close strike, uploads are often delayed.

Bad work units hard locking the system - mild to substantial - Check your systems regularly, although the newer WUs seem more reliable.

My Personal Penalty -
There is no reasonably afford broadband available this far out in the soybean field. And the land-line infrastructure limits my dial-up connection to an average of 28.8kbps. It takes me twenty to twenty-five minutes to upload a 4MB A3 and almost exactly two hours for a 20MB one (WHICH VIOLATES MY CLIENT CONFIGURATION SETTINGS). These numbers are best case and don't allow for my "ISP" connection drops which happen occasionally. I also discovered the hard way that the Stanford servers reset at 3:00AM EDT and again at 4:00AM EDT.
Image Image
.* * M-51 * *. .The Whirlpool Galaxy.
farmpuma
Minister of Gerbil Affairs
Silver subscriber
 
 
Posts: 2311
Joined: Mon Mar 22, 2004 12:33 am
Location: Soybean field, IN, USA, Earth .. just a bit south of John .. err .... Fart Wayne, Indiana

Re: As the Work Unit Crunches

Postposted on Wed Jun 16, 2010 5:33 pm

I sheepishly admit that I shouldn't complain about my trifling difficulties. I'm surprised that DSL isn't available for you, but have a vague memory of your mentioning the problem in a thread somewhere before. I expected power companies to be rolling out power line ISP services by now, too, but having done no research, I suppose one problem may be its producing radio frequency interference, but I don't know what frequencies they chose, if any, yet. (I wonder if Western submarines still have long wave communications hardware on them. If not... then maybe that should be explored.)

I think your configuration settings violation problem is that all SMP WU's are in high performance or in testing or some other thing, so they ignore some client settings. For example -forceasm and -local are supposedly already part of every -smp client's operation.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Thu Jun 17, 2010 10:21 pm

Well, about 3 hours before the X2-3800's WU was due, I switched the flash drives in that machine and the Q8400, and restarted both machines. I had meant to kill the Q8400 before it finished the WU completely, but it uploaded it and I'm not sure what happened. It seems to have worked. Having set both machines to work on each other's WU, I forgot at first but then remembered to add -oneunit and -pause parameters to their invoking command lines, which is why I was less concerned about the Q8400 finishing. I also forgot at first to change the -smp to 4 instead of 2 on the quad, but that was fixed quickly.

I got a lot of points from my other machines on such WU's, but since I waited too long to try the switch, I was down to something under 40 minutes until the deadline when it uploaded, so the bonus points were minimal. Anyway, after that was done I switched them back and everything seems OK. I had to kill the X2-3800's backed up version of the WU it tried to finish the last part of, then delete the work directory files and the machinedependent.dat and queue.dat files just to make sure it didn't try to redo the WU again, and it appears to have gotten everything it needed from Stanford. My local stats may be a little off, but they always get screwed up every now and then anyway.

I'm surprised it worked. I'd like to have a network file system running so it wouldn't be as much of a PITA to do all that, but for the time being I'll be thankful for the success. :)
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Thu Jul 01, 2010 10:45 pm

Now I have to see which virtualization software works on Windows 7 with up to 8 cores, even if 4 of them are of the hyperthreading variety. And I'll probably have an interesting time searching to see how to overclock the GPU in Win7, too. I found some change you make in RivaTuner to make it work with newer drivers and maybe with different a Windows version when I ran the W7 RC, but I can't remember where I found it any longer.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Thu Jul 01, 2010 11:11 pm

Howdy folks! thought I'd drop in with a question. I'm in windows 7 pro 64 bit, with an SMP on my phenom II 965 and GPU2 on my 3870.
Recently I've been getting failed WU's, and declined points, even though I don't even use this computer for gaming anymore. Anyone got any ideas as to why I'm failing WU's (2 on my phenom and 1 on my 3870)
Image
Chun¢
Gerbil XP
 
Posts: 454
Joined: Mon Oct 20, 2008 7:46 pm

Re: As the Work Unit Crunches

Postposted on Fri Jul 02, 2010 9:22 am

Maybe your GPU is starting to get unstable? My first thought would be heat/power issues.
(this space intentionally left blank)
just brew it!
Administrator
Gold subscriber
 
 
Posts: 37522
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: As the Work Unit Crunches

Postposted on Fri Jul 02, 2010 4:29 pm

It sounds like it could be heat, but I would normally expect only one of the components to have a problem if heat only increased somewhat. GPU's often run too hot unless you're careful, especially because many of us overclock the living heck out of them. More information is needed to respond more usefully. Are you overclocking both the CPU and the GPU? If so, what are the stock and current frequencies? And can you give us tempreratures.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Fri Jul 02, 2010 4:34 pm

I posted in another thread about this, but it bears repeating here. Stanford says they're putting -bigadv on the Windows client, and shutting it off on the Linux client. Here's the link where I read it. And here's the useful bit:
Introduction:
Finally after a long wait, bigadv is now available on Windows Natively: (compiled from different posts in this forum)

kasson wrote:We've turned off bigadv on linux for the time being. We're switching bigadv over to A3, but as you may have noticed there's a bug in the linux A3 core that affects bigadv stability. We've been working on the bug, but it's still in hiding at this time. If you're running VM's under windows, I'd suggest trying the native windows client, which is still enabled for bigadv. OS/X is also still enabled. Run with the -bigadv command line option. Depending on the work unit availability, machines requesting bigadv may roll over to normal -advmethods.

More info here. Please note that it hasn't been updated for Windows yet, but it can give you a general idea about the bigadv WUs.


Running bigadv WUs:
1) A system with 8 or more Physical Cores. (MUST READ NOTE #1)
2) Download the latest SMP2 Client: <Click>
3) Install the SMP2 Client by following this guide or this one.
4) If you want to take advantage of the Bonus Points (highly recommended), make sure that your Passkey/Username is already qualified.
5) Run the Client with these flags before it downloads the WU:
-verbosity 9 -bigadv -smp
6) Windows OS 32/64 bit.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Sat Jul 03, 2010 12:43 pm

I haven't over clocked either my card or my phenom. I also have a 500W 80+ certified antec power supply.
Should I just give them some cool down time between work units?
Image
Chun¢
Gerbil XP
 
Posts: 454
Joined: Mon Oct 20, 2008 7:46 pm

Re: As the Work Unit Crunches

Postposted on Mon Jul 05, 2010 1:48 am

Stanford has had trouble with the WU's for ATI cards for some days now.

I turned OFF the option to allow >10MB WU's. I just picked up a P5733 and a P5745 and they didn't bomb. Or maybe Stanford just finally fixed it.

[Oh yeah, I also cleared out the work folder, as well as other files in the folding directory, such as unitinfo.txt.]
Image
ritsu
Gerbil In Training
 
Posts: 9
Joined: Sun Sep 27, 2009 11:07 am

Re: As the Work Unit Crunches

Postposted on Mon Jul 05, 2010 1:03 pm

Last weekend, I failed 9 / 14 WU's. I think I found out why. I had a Registry hack to allow me to control the fan speed from CCC, but I guess I updated it so I can't just set the speed at 100% all the time anymore.
Image
Chun¢
Gerbil XP
 
Posts: 454
Joined: Mon Oct 20, 2008 7:46 pm

Re: As the Work Unit Crunches

Postposted on Tue Jul 06, 2010 11:23 pm

It's odd your SMP client is having trouble, though. Unless your GPU overheated and in so doing the case warmed up enough to make the CPU too hot, crashing its WU's.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Fri Jul 09, 2010 10:28 am

It could be due to my internet connection, right? On bad days it gets pretty flakey.
Image
Chun¢
Gerbil XP
 
Posts: 454
Joined: Mon Oct 20, 2008 7:46 pm

Re: As the Work Unit Crunches

Postposted on Fri Jul 09, 2010 4:32 pm

I've never seen a WU fail because of a bad net connection. It can fail to upload or download a WU and wait a long time between attempts, but that shouldn't cause the WU itself to crash or anything. It only uses the network on uploads and downloads from what I've seen and read.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Fri Jul 09, 2010 5:27 pm

Ragnar Dan wrote:I've never seen a WU fail because of a bad net connection. It can fail to upload or download a WU and wait a long time between attempts, but that shouldn't cause the WU itself to crash or anything. It only uses the network on uploads and downloads from what I've seen and read.

It did for me with A1/A2 WinSMP when I last ran it. The mpiexec stuff went nuts after the network connection reported a disconnect. :o Basically it meant I could not run it on a wireless connection.
Image
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
Flying Fox
Gerbil God
 
Posts: 24297
Joined: Mon May 24, 2004 2:19 am

Re: As the Work Unit Crunches

Postposted on Sat Jul 10, 2010 6:46 am

Flying Fox wrote:
Ragnar Dan wrote:I've never seen a WU fail because of a bad net connection. It can fail to upload or download a WU and wait a long time between attempts, but that shouldn't cause the WU itself to crash or anything. It only uses the network on uploads and downloads from what I've seen and read.

It did for me with A1/A2 WinSMP when I last ran it. The mpiexec stuff went nuts after the network connection reported a disconnect. :o Basically it meant I could not run it on a wireless connection.

Sorry, when I said "net" I meant internet. I've seen problems when the network DHCP server restarts, of course, though it only really requires restarting Windows machines so they can reset their IP addresses and figure out how to connect to other machines in the Workgroup. But I've only recently been using the Windows SMP client after a brief test 2-3 months ago, so perhaps I'm in for an unpleasant experience at some point. After all this time, Windows networking still seems to be crappy.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Sat Jul 10, 2010 9:38 pm

Download of the WU and upload of the results from the Internet don't affect stability as they are checksummed. Local network connection can cause problems because mpiexec uses loopback networking to communicate between the multiple processes using the processor cores.
notfred
Grand Gerbil Poohbah
 
Posts: 3716
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: As the Work Unit Crunches

Postposted on Wed Jul 14, 2010 3:27 pm

So lately my GTS 250 has been acting strange. Normally it takes a minute or so to churn out 1% of the WU, but after each WU completed it takes longer and longer to complete a unit peaking at 12 minutes. Once I reset my computer it resumes where it left off and starts churning out that WU in a minute or so again. This just started happening about two days ago. I am also running an smp client on my 1055T. Any ideas why this is happening?
To Start Press Any Key'. Where's the ANY key?
If something's hard to do, then it's not worth doing
You know, boys, a nuclear reactor is a lot like a woman. You just have to read the manual and press the right buttons.
mmmmmdonuts21
Gerbil Elite
 
Posts: 590
Joined: Wed Jul 16, 2008 9:09 am

Re: As the Work Unit Crunches

Postposted on Wed Jul 14, 2010 10:48 pm

The only thing that comes to mind is that the SMP client and any other processes are dominating the CPU, not allowing it to service the GPU client (which it has to poll, supposedly, instead of merely being interrupted to grab data from and send data to) rapidly enough. But maybe I'm getting you mixed with someone else and you don't have the SMP client running. ?
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Thu Jul 15, 2010 7:48 am

No I definitely have the smp client running. Its churning out most of my ppd right now (10-12k). I have the smp client priority set to idle and the GPU's set to low. Should I try and set the smp flag to -smp 5 instead of letting it run on all the cores? I don't really have any other services running in the background for the most part. I do occasionally game on the system but then shut down steam and everything else using the task manager once I am done. I guess I am just trying to optimize my folding as best as I can and its bugging me that it just started doing this three days ago and I haven't done anything to my system.
To Start Press Any Key'. Where's the ANY key?
If something's hard to do, then it's not worth doing
You know, boys, a nuclear reactor is a lot like a woman. You just have to read the manual and press the right buttons.
mmmmmdonuts21
Gerbil Elite
 
Posts: 590
Joined: Wed Jul 16, 2008 9:09 am

Re: As the Work Unit Crunches

Postposted on Thu Jul 15, 2010 10:53 pm

mmmmmdonuts21 wrote:No I definitely have the smp client running. Its churning out most of my ppd right now (10-12k). I have the smp client priority set to idle and the GPU's set to low. Should I try and set the smp flag to -smp 5 instead of letting it run on all the cores? I don't really have any other services running in the background for the most part. I do occasionally game on the system but then shut down steam and everything else using the task manager once I am done. I guess I am just trying to optimize my folding as best as I can and its bugging me that it just started doing this three days ago and I haven't done anything to my system.

You may as well try the "-smp 5" thing just to see what it does. I can't recall what I set my GPU clients to, but generally I think I leave it at whatever it defaults to.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Fri Jul 16, 2010 7:49 pm

-smp 5 will stop -bigadv right? But then again we are running out of those big WUs now, right?
Image
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
Flying Fox
Gerbil God
 
Posts: 24297
Joined: Mon May 24, 2004 2:19 am

Re: As the Work Unit Crunches

Postposted on Sat Jul 17, 2010 4:57 pm

Somewhere I read that -bigadv will only work on CPU's that report 8+ cores. I don't know why hyperthreaded cores report as 2 instead of one, but since they do, the i7 gets to run the WU's, and a fair number of people seem to be making the deadlines. Anyway, I got my first -bigadv WU on Thursday and I'll see how it does. It may turn out not to be worth the time investment. One thing's for sure, a huge multiplier like this CPU has makes finding the fastest OC less precise, with more "error" in "trial and error".
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Mon Jul 19, 2010 7:44 am

Ragnar Dan is correct. I can not run the bigadv units because you need 8 cores as a minimum requirement.

Just an update for me. After some testing this weekend with smp 5 with the GPU overclocked (how I had it) & resetting the GPU back to stock clocks with smp 6 (I still don't know why this made a difference because I had it overclocked stable for a month before having issues) I decided to run the stock clock GPU with the full smp client. I suffer a little bit on the points for the GPU from when it was overclocked but it was not even close to the amount of points dropped by not running the SMP client full bore. So far no GPU slow down which makes me happy.
To Start Press Any Key'. Where's the ANY key?
If something's hard to do, then it's not worth doing
You know, boys, a nuclear reactor is a lot like a woman. You just have to read the manual and press the right buttons.
mmmmmdonuts21
Gerbil Elite
 
Posts: 590
Joined: Wed Jul 16, 2008 9:09 am

Re: As the Work Unit Crunches

Postposted on Mon Jul 26, 2010 2:29 pm

What name are you folding under, by the way, mmmmmdonuts21? I'd just like to get a look at how your machine is producing from http://folding.extremeoverclocking.com/.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Mon Jul 26, 2010 8:34 pm

What name are you folding under, by the way, mmmmmdonuts21? I'd just like to get a look at how your machine is producing from http://folding.extremeoverclocking.com/.


I am folding under the name rappr. I just added a GTX 260 yesterday (Thanks DanZMan) to my GTS 250 and Phenom 1055T@3.625 GHz. Should be around 22-24k ppd. Lately though I have had random BSOD every few days, probably from the heat every once in a while the AC goes off. I would like to know what a few of the higher ranked people are folding with on the team and ppd/w is. Mine should be close to 46-48 ppd/w as I draw about 500 watts from the wall.
To Start Press Any Key'. Where's the ANY key?
If something's hard to do, then it's not worth doing
You know, boys, a nuclear reactor is a lot like a woman. You just have to read the manual and press the right buttons.
mmmmmdonuts21
Gerbil Elite
 
Posts: 590
Joined: Wed Jul 16, 2008 9:09 am

Re: As the Work Unit Crunches

Postposted on Tue Jul 27, 2010 4:50 pm

Yeah, hopefully only until I finish transferring my machine to its new case and re-applying thermal goop, I've got heat problems. It BSOD's every now and then, and auto restarts most of those times. I have to find out what voltages that I can change in the BIOS mean what, because what the BIOS calls the MCH is running hot.

Anyway, I get about 40% of the output of this machine from my GTX 260 when the SMP client is folding something crappy. On better WU's, it's under 32%. Eventually I'll move to -bigadv units, but not until I get the heat, reliability and speed problems worked out.

I'll have to get out my Kill A Watt after everything's settled down and see how it measures. I know it's bad, though, because I retired an X2-3800 and an Opteron 165, and this thing appears to be using more power than both combined. :o
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Sun Oct 03, 2010 8:01 pm

A couple of updates.

I noticed today that TR's team suddenly saw 28 new folders show up today. I'm glad to see that, but it causes me to wonder why there hasn't been a corresponding increase in the number of posts to this forum.

I've also noticed in the last few weeks some WU's of mine not being credited when I uploaded them, at least for more than an hour after they were really uploaded. Which is odd. You'd think there'd be lots of complaints about that on foldingforum.org, but I haven't checked yet.

Also, it appears that mmmmmdonuts21 / rappr quit folding on September 14. Gerbil Jedidiah has also declined significantly in output, and also quit posting a lot. I forget if he mentioned something about that somewhere.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5354
Joined: Sun Jan 20, 2002 7:00 pm

Re: As the Work Unit Crunches

Postposted on Sun Oct 03, 2010 8:07 pm

Ragnar Dan wrote:What the BIOS calls the MCH is running hot.
That's your motherboard chipset. Does your chipset have one of those dinky little 40mm fans that always fail?
JustAnEngineer
Gerbil God
Gold subscriber
 
 
Posts: 15343
Joined: Sat Jan 26, 2002 7:00 pm
Location: The Heart of Dixie

PreviousNext

Return to TR Distributed Computing Effort

Who is online

Users browsing this forum: No registered users and 1 guest