| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Cure Muscular Dystrophy - Phase 2 Forum
Thread: Monster WU on the loose... |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 98
|
|
| Author |
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: 
|
Kremmen, if that can help you for your estimates, difficult positions on my Q6600 under Ubuntu 64 took between 4 and 5 hours each and the "worst" WU had 4 such difficult positions to pass. I don't know which ratio you must use to convert this info to your machines but I am afraid that could give big numbers...
----------------------------------------And I forgot, this quad is overclocked at 20 %. Good luck. Jean. |
||
|
|
oliverstirling
Advanced Cruncher United Kingdom Joined: May 7, 2007 Post Count: 107 Status: Offline Project Badges:
|
Jean,
----------------------------------------You mention you're quod is overclocked. Does this have any effect on wu's aside from computing them faster? Have been wanting to overclock my systems for a while but was unsure of any possible side effects it would have on crunching. [Edit 1 times, last edit by oliverstirling at May 30, 2009 11:26:27 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
the "worst" WU had 4 such difficult positions to pass. I don't know which ratio you must use to convert this info to your machines but I am afraid that could give big numbers... That's the important point right there: How many difficult positions might turn up in a single WU. Obviously, I've hit 3 so far. If there are only 3 or 4, that's not too bad. Most machines are faster than mine, but my wingman hasn't finished either. And then there's always the possibility that the WU will end up being marked as invalid anyhow, though that's "only" been happening to about 10% of HCMD2 613 WUs on this machine. |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges:
|
Does this have any effect on wu's aside from computing them faster? Have been wanting to overclock my systems for a while but was unsure of any possible side effects it would have on crunching. No side effect in my case because this particular Q6600 processor can be pushed by HW gurus up to 4 GHz and mine is only at 2.88 GHz instead of 2.40 normally. Since I am not an overclocking expert and I did not want to make any effort to become one I have followed some expert's comment that the Q6600 can be pushed 20 % without having to tweak other parameters (voltages mainly) or fitting it with a special heat sink, which is true. I have simply run the usual stability tests and I am carefully monitoring temperatures and cleaning the dust when necessary. If you are not a HW geek you must be very careful with overclocking. Not all processors are as tolerant as the Q6600, and you can have bad consequences from an unstable system with errors up to damaging your processor. Last, obviously, WCG does not recommend overclocking and support is normally for not overclocked systems. Cheers. Jean. |
||
|
|
oliverstirling
Advanced Cruncher United Kingdom Joined: May 7, 2007 Post Count: 107 Status: Offline Project Badges:
|
Thanks, that's good to know. Didn't want to overclock my system only to find the majority of results returned would be marked error.
I've got a water cooled system installed so temp control should be fine, my only problem now is finding a stable optimal overclcoking point...let the experimenting begin! Btw nice disclaimer  |
||
|
|
gdlxn
Cruncher Joined: Nov 16, 2004 Post Count: 24 Status: Offline Project Badges: 
|
I just aborted one, CMD2_0001_2IJM_B.clustersOccur_2O72_A.clustersOccurs_191_2, that had been running for over 61 hours on wth no change in the percent complete for quite some time. The system is 2.40 GHz Pentium 4 running openSUSE 11.1.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Maybe after all, P3 (or a P4 that has hyperthreading) on should be excluded from running this project, or an additional 'yes I read and understand the consequence' option before getting any more of these jobs.  Sekerob, this is a bit off topic, but since you mentioned it I'd like to know your opinion. I'm guessing you're against the idea of using HT on P4 for BOINC, but see this temperature chart from one of my P4s:  The green line is CPU temperature in °C. On the left I was running with 2 (HT) CPUs and 50% CPU time, on the right it was 1 CPU and 100% CPU time. As you can see, there's a big difference on temperature, so I prefer the first approach. I've been doing this in all my P4 that support HT. What's your advice? [Edit 1 times, last edit by Former Member at May 30, 2009 5:39:41 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello bono_vox,
I will toss in my (unasked-for) 2 cents. I assume that the 58C temperature is for a P4 with hyper-threading enabled running 2 BOINC projects at 50% (1 second on -- 1 second off). I assume that the 64C temperature is for the same P4 running 1 BOINC project 100% of the time. My rule of thumb is that hyper-threading will speed up floating-point heavy programs no more than 15%, so I estimate total throughput in the first case to be 115% * 50% time = 58% of the throughput of the second case. So the second case turns out 70+% more work for a 6C temperature increase. This is why I like to run P4 computers with hyper-threading enabled (to run OS subroutines simultaneously with BOINC) but with only 1 BOINC thread. Lawrence |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello bono_vox, I will toss in my (unasked-for) 2 cents. I assume that the 58C temperature is for a P4 with hyper-threading enabled running 2 BOINC projects at 50% (1 second on -- 1 second off). I assume that the 64C temperature is for the same P4 running 1 BOINC project 100% of the time. Hi Lawrence and thanks for your (unasked-for) reply  . Your assumption is correct. The moment I switched from 2 processes @ 50% usage to 1 process @ 100% usage temperature went up. The last part part of the chart shows a decrease on temperature because I switched back to the original configuration.I'm worried about those extra 6°C. I've read that P4s should not go over 66~67°C (at least 90nm Prescotts) and 64°C is too much close for my comfort (and that's without doing anything else on the computer). What do you think? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello bono_vox,
You probably have a max Tc of around 73C. I do not like 64C myself, but Intel has an onchip Power Management feature that cuts power and speed if the temperature hits the cutoff value. My guess is that you will not need to use the onchip protection at 64C. As long as you are running a desktop and are not bothered by a noisy fan, it should be all right. Lawrence |
||
|
|
|