| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Say No to Schistosoma Forum
Thread: SN2S: General discussion/problems |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 75
|
|
| Author |
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges: 
|
I rebooted my laptop today and 7 out of the 8 tasks running finished (in error) right at startup. 32 Hours of runtime down the drain...
---------------------------------------- Probably it was just a glitch in my laptop, but I thought I'd mention it anyway as I have never experienced this before on any of my systems. If required, I can post some more data.   |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Not the first report in recent the weeks. Suspecting again the [first appearing with Vista] too aggressive shutdown or the overbusy system during startup. Continue to recommend to set the startup delay in the cc_config.xml to 60 seconds with the <start_delay>60</start_delay> option, and when BOINC is installed as service to mod the BOINC service properties to "Automatic, (delayed start)". Better [IMHO] to loose 2 minutes at boot then 32 hours.
--//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
<comment>
Talk about old BOINC versions used by some of my wingWUs, the oldest I have come across thus far: BOINC_v6.4.5. Is the usage of that BOINC version even legal ? </comment> |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges:
|
Thanks Sekerob, I'll keep an eye on it and will do the modifications if it happens again.
---------------------------------------- |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges: 
|
The Ubuntu Synaptic Depot offers 6.12.33 for probably a year now if not longer. Hi SekeRob, I thank you for your answer, but Synaptic is still offering to me only 6.10.17 for Ubuntu 10.04 x64. Should I add an additional depot? After some tries and investigations, I am not able to go ahead. Surely, I missed something. Yves |
||
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
An affinity problem? More like a persistent problem AMD has with VINA. Don't think that is going to get fixed lest it snowballs... the invalid, essentially that the 2 wingman, often Intel meeting AMD but also AMD-AMD, produce slightly different results, so much so that the random wingman lets the validation tip one or the other's way. BTW, 6.10.17? Don't fix what's not broken, but if this is on Linux, pretty plz upgrade. This is a generally poorly benchmarking release (on Linux), meaning it's understating the FLOPS it processes. --//-- Just for clarification, this is not an AMD problem, it is a Socket A problem. They will only validate against themselves. The techs know about it and it has been gone over and over. Just don't run any VINA applications on any Socket A system.  Distributed computing volunteer since September 27, 2000 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Had not filtered the problem down to Socket A, but Socket A and AMD are according to a wiki [yes a wiki], inextricably coming together. http://en.wikipedia.org/wiki/Socket_A .The problem being that there's no way for the WCG servers to identify if a CPU sits in any particular motherboard. Does look like there's no reliable way to define a homogeneity group for this.
--//-- |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: 
|
Had not filtered the problem down to Socket A, but Socket A and AMD are according to a wiki [yes a wiki], inextricably coming together. http://en.wikipedia.org/wiki/Socket_A .The problem being that there's no way for the WCG servers to identify if a CPU sits in any particular motherboard. Does look like there's no reliable way to define a homogeneity group for this. Huh? According to the WIKI, Socket A is 32-bit only, and for AMD this means cpu's that only has SSE-type instruction-set. AMD only added SSE2 and higher with their 64-bit cpu's, like the Opteron, Amd64 and Amd-X2 and so on, but all these 64-bit cpu's uses other socets, like the 940 for Opteron and so on. Grouping the Socket-A-cpu's together is therefore very easy, just check for "if AMD and not SSE2". BTW, my guess is this is really a "doesn't have SSE2"-problem, so intel-cpu's that doesn't have SSE2 probably has similar problems with validation. For intel this means P3 and earlier cpu's, and since these cpu's is by todays standards very slow they're probably few and far between... edit - hang on, the system in question was a Athlon II, and this is 64-bit. So, how this is a socket-A-problem doesn't make any sence...  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." [Edit 1 times, last edit by Ingleside at Mar 4, 2012 12:47:55 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Makes me ponder if a general homogeneity separating 64 bit and 32 bit would help to lower the wingman matching issue for all the VINA sciences. Frankly, only looking at repair and inconclusive job quorums, seeing really very few that via that route go invalid... it's 99 out of 100 an wingman upfront error result or a No Reply. On the 400 my hosts completed of SN2S, not even 1% needing that roundabout.
--//-- |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges:
|
A couple of WUs are invalid today.
----------------------------------------I am not really able to understand why it was OK during the 5 first days and why it is not OK anymore now? If the Socket A is responsible, every WU on this particular host should be invalid and not only "some of them", even if it represents "only" around 25%-35% of the daily production of this host. Yves   |
||
|
|
|