Posted on 01/21/2006 10:29:50 PM PST by texas booster
Priceless.
:O)
P
Have you thought about overclocking the flux capacitor to a full 1.21 gigawatts?
LOL! Man they're behind the 8 ball! We discussed that several weeks ago after someone posted that write up on HT'ing by Pande... I think I was running it for 24 hrs before turning it off after reading the Pande post. Last time I browsed their thread (yea...had Lysol handy), I recall seeing a post where the poster was bragging about pumping out more WU's with HT'ing...hahaha.
I purchased 2 used P4 (2.26GHz and 2.6GHz) motherboards (and needed accessories) on eBay last night to start building that diskless Folding Farm thatI've been jabbering about. Can't wait for the brown truck to show up!
My god man... have you considered how much shielding that will take? The Global warming process alone, will take 3 kyoto's worth of hot-air bypass's to overcome the thermal uptake.
A more elegant democratic design would be to bypass the binary polarization, solidify your worker processes by joining them together using Chad's process, and apply the theory of negativity.
Wow, no wonder you're rolling up the stats ;)
Although, I've been considering a more democratic approach to the folding procedure. I figure I can up my totals pretty significantly if I stop doing any work, and just tax everyone else for 40-50% of their work. I call it the "Teddy K Folding Fairness Plan" - after all, it's not fair that you have so many CPUs, and I have only a few, so I'm just going to quit and you can give me a free ride, right?
Does that plan only include future points earned, or must I sacrifice those that I've already put in the bank?
I'm not clear on the point that Pande was trying to make. Sometimes overclocking, and hyperthreading can cause errors in the results files, causing those workunits to be rejected, BUT, if you have a hyperthreaded CPU turning out two good workunits in 3 days, as opposed to that same CPU turning out one good workunit in 2.5 days, which is better for science?
He speaks of speed in turning around workunits, but a half day difference to get two workunits, vice one? Seems to me, HT (IF THERE ARE NO WU ERRORS), is better for the system.
Seems to me that overclocking, and feeding complex workunits to woefully underpowered processors, would be more harmful to the collection results.
Since you appear to be the undisputed champion at this process, I've got a couple of questions that you might be able to steer me toward the answers...
I've got 5 PCs dedicated to this process now. And two more that are giving me fits.
The first is a WinXP machine that hangs whenever it starts to process a WU (either the console version or the GUI). I've pretty much given up on that machine.
The second is a Windows NT server (running the 5.04beta console) that seems to connect, download and start a WU but almost immediately gives the following set of log entries:
...
[05:07:30] Folding@home Core Shutdown: EARLY_UNIT_END
[05:07:32] CoreStatus = 72 (114)
[05:07:32] Sending work to server
[05:07:32] + Attempting to send results
[05:07:33] + Results successfully sent
[05:07:33] Thank you for your contribution to Folding@Home.
[05:07:37] - Preparing to get new work unit...
...
It's done this several times, which leads me to my next question:
In looking at my stats, I'm noticing that I have credit for 26 WUs, but only completed successfully 1 or 2 of them, to my knowledge-- which my score indicates. Do these 0-score WUs fall off eventually? I'm assuming that these are the ones that the NT machine keeps failing, reporting results, and receiving new ones.
Any thoughts?
clean out your cash. That always works for me... especially when I launder it in someplace that sells adult beverages. ;)
First
re: the XP machine that hangs. Since you didn't mention that it had a problem downloading the WU and the core, the problem seems to be when it starts processing.
Can you boot this into safe mode, then launch the console after the computer boots, to see if it still hangs?
If it starts processing normally, in safe mode, then the problem is a software conflict with something else running on your system, if it still hangs, then it's a hardware error, or a corrupt kernel. First suggestion at that point, reload XP clean.
As for the NT computer... *sheesh*. It's been years since I've had to deal with NT.
The incomplete workunit problem, will be taken care of, after periodic reviews by the Stanford group. They'll eventually eliminate all your zero point work units.
But what is causing the "early work ends". First guess, I'd say permissions. Are you running this as a local user, a domain user, or an administrator? NT was funky about permissions.
Look in the service log, to see if any processes are aborting. Remember, there are two processes running... the
FAH console, and the FAH core. If the core crashes, the console says "done", "restart".
EARLY_UNIT END: Quite possibly the most common error found today. EARLY_UNIT_END is usually caused by one of two things: a bad WU or an unstable system.
If you get one isolated EARLY_UNIT_END, it's most likely just a WU that is bad. It's not a problem, and you shouldn't worry about it. It's usually caused when atoms in the WU reach impossible positions and Gromacs can't continue.
Multiple EARLY_UNIT_END errors are a sign of a severe problem with your machine. Machines that are clocked too high, have heat problems, or possibly have SSE forced on (AMD only) will generate this error. You should stop F@H if you get more than one EARLY_UNIT_END per week per machine, and certainly if you get two in a row. Make sure your machine is up to spec, with reasonable temperatures, reasonable clocking (CPU, FSB, and memory must all be stable), and a good, powerful PSU. EARLY_UNIT_END is most often caused by problems with a user's machine, and an abnormal number of them certainly merits examining your system.
This error may be accompanied by a LINCS WARNING message that gives more specific technical details on exactly what happened.
NOTE: See the description about "-forceasm" (3.x) or "-forceSSE" (4.x) causing SPECIAL_EXIT on certain AMD based systems. If you are running an AMD Athlon XP with the Thoroughbred or Barton cores, you should remove the "-forceasm" or "-forceSSE" switch, most likely fixing your problems. Ok... so that's probably greek. Completely uninstall both the GUI, and the console client. First, start run, see if you have a folding@home program group, then uninstall the gui. Under control panel, admin tools, services, set the FAH client to disabled. Delete both the folder where you placed the Console client, and the Program files\Folding@home directories. Re-install, but DO NOT use the advanced methods, and see if the problem persists. If it persists, then it's most likely hardware. 1. Verify your system boards settings, and take off any overclocks. 2. Try removing one simm/dimm at a time, to see if it's a memory parity error. 3. If you have the capability, "underclock" your processor.
Completed 0 out of 500000 steps <0>
Completed checkpoint files
Completed checkpoint files
Completed checkpoint files
Completed 5000 out of 500000 steps <1>
Writing checkpoint files
Writing checkpoint files
Writing checkpoint files
Writing local files.
And so forth and so on. I'm up to step 34 now, and all steps look like that.
There are 3 different cores, and all 3 of them post to the log file differently...
Tinker, Gromacs, and Amber.
I haven't paid much attention, to which, writes which way, but they all log and record the steps differently. As long as it's not posting any errors, I wouldn't worry about it.
Well, just wait until the DUmmies figure out that you work for Halliburton - then the $hit will really hit the fan...
Actually, I'm only a sub-contractor. I work for KBR... but I don't think they know how to differentiate.
Still plugging away with my two computers, one slow and the other relatively fast(er).
Reading the stuff on this thread, I have no CLUE what you guys are talking about.
Hyperthreading? Overclocking? Diskless folding?
As long as all this stuff won't be on the final exam!
Pleased to be a part of such a team though! We're well on our way to being in the top 100 by June I'd say!
DUmmies up to 175 computers
http://fah-web.stanford.edu/cgi-bin/main.py?qtype=teampage&teamnum=48157
| Free Republic Folders - A Tribute to Ronald Reagan |
| Date of last work unit | 2006-01-24 07:18:00 |
| Active CPUs within 50 days | 536 |
| Team Id | 36120 |
| Grand Score | 873729 (certificate) |
| Work Unit Count | 6924 (certificate) |
| Team Ranking (incl. aggregate) | 383 of 42365 |
| Home Page | http://www.freerepublic.com |
| Fast Teampage URL | http://fah-web.stanford.edu/teamstats/team36120.html |
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.