| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 99
|
|
| Author |
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges:
|
One host (Athlon II x4) - which usually earns between 3'750 and 4'200 points daily - just completed 4 HST1 WUs after over 40 hours each:
----------------------------------------- Claimed: 174 points /WU !!! - Granted: < 500 points /WU. Instead to earn about 6'600 points for 40 computation hours, the effective yield will be < 2'000 points. Questions: - Is it ineffective for a host (without HT, 4 real cores) to compute 4 HST1 WUs at the same time (see for example CEP2 limitation)? - Is HST1 so badly designed that it is impossible for host to claim a real and reasonable amount of points? (see a high clamp at 174 points whatever the duration is) - Is any affinity missing between HST1 and AMD CPU? I would be interested to better understand the reason(s) for so many troubles with HST1. We are not anymore in the project starting phase. Many members provided already remarks and observations, it would be fine to be able to notice some improvements. Cheers, Yves |
||
|
|
Sandvika
Advanced Cruncher United Kingdom Joined: Apr 27, 2007 Post Count: 112 Status: Offline Project Badges:
|
Yes, anecdotally, my daily points tally has been about 70K when crunching other projects and about 50K when crunching HST or FAAH2 with their long running WUs.
----------------------------------------I'm also seeing huge mismatches between claimed and awarded points and a low average points per hour (which obviously translates into low points per day) A couple of recent example WUs.... Here we see 28 and 35 hours of crunching getting almost the same points: HST1_ 007374_ 000042_ KC0019_ T350_ F00040_ S00008_ 0-- wcg Valid 26/08/16 10:36:29 31/08/16 11:25:28 34.91 / 35.06 376.4 / 415.6 HST1_ 007421_ 000077_ MC0022_ T300_ F00072_ S00008_ 0-- wcg Valid 27/08/16 11:37:33 01/09/16 05:06:24 28.10 / 28.42 348.2 / 408.5 Here we see very similar crunching times getting wildly different points: HST1_ 007421_ 000077_ MC0022_ T300_ F00072_ S00008_ 0-- wcg Valid 27/08/16 11:37:33 01/09/16 05:06:24 28.10 / 28.42 348.2 / 408.5 HST1_ 007372_ 000063_ KC0011_ T350_ F00080_ S00008_ 0-- wcg Valid 26/08/16 10:28:11 31/08/16 22:17:12 27.86 / 28.16 355.7 / 322.8 I've not seen rewards vary so wildly for other projects but then I've not seen such huge variability in WU crunching times either. My points for my Emerald year of HST were 1.2 million, after getting Sapphire it was 2.0 million, so only 0.8 million more. I guess it is what it is and if you're in it for the stats then this project will disappoint! It's hard to get the WUs so I'll probably pack it in when I get to 5 year diamond. Maybe CEP2 will have returned by then with the silly 18 hour processing time limit removed. This project proves that 18 hours does not indicate there's a problem! ![]() ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Got 3 TB units to crunch...aborted 1 ie ran out of time to report, 2 others still running, all are taking over 3 days to complete...running W10...please get this sorted guys...
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges:
|
Got 3 TB units to crunch...aborted 1 ie ran out of time to report, 2 others still running, all are taking over 3 days to complete...running W10...please get this sorted guys... What are your system specs and what other programs are you running at the same time ? Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Eric_Kaiser
Veteran Cruncher Germany (Hessen) Joined: May 7, 2013 Post Count: 1047 Status: Offline Project Badges:
|
That's a good question. Even on my AMD-Kabini, a real lowcost cpu, a wu is finished within 36 hours.
----------------------------------------![]() |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
All HSTb get 10 days deadline, so not sure what's there to sort. v.v. W10, install BOINC as service and sign-out when not using system... on 2 of 3 systems here this resulted in 30-35 percent more throughput because W10 loves to load up on bunches of background processes to snoop and index and resnoop on the signed inuser, plus some they've really deeply hidden. /OT
|
||
|
|
supdood
Senior Cruncher USA Joined: Aug 6, 2015 Post Count: 333 Status: Offline Project Badges:
|
Another head-scratcher:
----------------------------------------24.78 168.6 / 547.7 (me) 16.89 449.2 / 547.7 (wingman) 26.21 460.8 / 246.9 (me) 8.59 170.0 / 246.9 (wingman) |
||
|
|
jay_Orlando
Senior Cruncher USA Joined: Jan 4, 2006 Post Count: 189 Status: Offline Project Badges:
|
I am concerned with Wu with T350 in the name. They seem to take 24 hours -
----------------------------------------Not the estaimated 7 hours. Question: Is there a separate forum thread somewhere that adresses these running times??? I have looked - just didn't see it I am not that concerned with points. I am concerned that I could be running WU for WCG that produce results in a timely fashion. Am I wrong? Is the long-running WU important because it takes a long time? FOUND the answer: "Wide Variation" is actually the norm at WCG, except the different runtime ranges are here easily recognized by their designators (as scientist explained, first part of a task name is for WCG indexing purposes, part is for the themselves) T000+T001 are shorter T300+T325+T350+T400 are long to very long. The closest answer on the why is here https://secure.worldcommunitygrid.org/forums/wcg/viewpostinthread?post=517439 ... different simulation systems. Thanks to SekeRob here is what BOINC says about my CPU:
T H A N K S in advance, Jay (edited to clarify question) ![]() [Edit 4 times, last edit by jay_Orlando at Sep 24, 2016 3:42:37 PM] |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges:
|
Hi Jay Orlando,
----------------------------------------just for the record: For a host (Athlon II x4, LinuxMint 17.3 x64) earning usually 40 point per hour per core, the WU HST1_007888_000079_KT0001_T350_F00065_S00006 took over 39 hours, the result has been declared "invalid" (no error. reason unknown) and 180 points have been granted. HST1 is a real pity. I am still contributing to this project because I think that it represents an important cause, however it is more than frustrating. Cheers, Yves --- PS: A second similar WU - HST1_ 007888_ 000051_ KT0001_ T350_ F00018_ S00006_ 1-- - taken over 39 hours is still pending ! Question: What would happen if all contributors would rigorously stop to support HST1? Hope for improvement? |
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1326 Status: Offline Project Badges:
|
Question: What would happen if all contributors would rigorously stop to support HST1? Hope for improvement? No I personally feel people that aren't worried about points would enjoy getting to work units easily ![]() |
||
|
|
|