| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Website Support
Thread: "Device Statistics" not accurate? [Resolved] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 28
|
|
| Author |
|
|
ryan222h
Senior Cruncher Joined: Sep 4, 2006 Post Count: 425 Status: Offline |
Every day I check the "device statistics" page to check on individual machines, and the run time and points always are less than they should be. Is there a problem with "device statistics" I'm assuming there's a problem because I added up the points of all the individual machines on the device statistics page...and they still don't even come close to the statistics in "my statistics history". Here's several days worth of a 24 core machine: 01/24/2014 0:014:03:45:06 42,450 75 01/23/2014 0:012:18:05:08 54,767 93 01/22/2014 0:011:00:19:50 59,817 103 Averages about 12-13 core days/day? Not right. Machine is AMD Opteron running Linux. And another machine 8 cores: 01/24/2014 0:006:17:16:25 31,837 40 01/23/2014 0:006:07:20:44 32,740 29 01/22/2014 0:005:20:58:00 28,669 43 This one is running Windows 7-64 with an i7. The problem is not as bad but still noticeable. I thought it could be a PV issue but that doesn't explain the discrepancy between "device statistics" and "my statistics history" pages. Any help would be appreciated. Apparently I'm not the only one. Here's a repost from Mamajuanuk's build log: Mamajuanuk: Well after five days crunching FA@H the early stats are in! Current five day stats look like this: 100k PPD (Average - 99,817 to be exact) 7 PPW (7.03) 168 RPD (results 150-190) 33 (average) CPU days per calander day (27-39) I don't know what you guys think, but I think the stats are a little dissapointing! This beast is running 64 cores, no H/T. I would expect it to do closer to 50 CPU days/Calander Day, just on that along, it appears to be running at a little ove 50% I'm going to stop the desktop and run it in full server mode for now and see if it improves. Any tips would be appreciated, as would your comments.  [Edit 3 times, last edit by ryan222h at Feb 7, 2014 9:14:34 PM] |
||
|
|
Mamajuanauk
Master Cruncher United Kingdom Joined: Dec 15, 2012 Post Count: 1900 Status: Offline Project Badges: 
|
I've added up my stats on the 'Member Stats Page' these total
----------------------------------------1,556,268 2294 Adding up the individual device stats returns: 1,032,976 1371 In both cases the left number is points the right is results, they appear to be about 500k points and 900 odd results. Most of my machines are Ubuntu 13.xx although 2 are Win7 with MegaCruncher being Ubuntu Server 12.xxx LTS Something is clearly amiss here!
Mamajuanauk is the Name! Crunching is the Game!
  |
||
|
|
PMH_UK
Veteran Cruncher UK Joined: Apr 26, 2007 Post Count: 786 Status: Offline Project Badges: 
|
For me the long term Device totals are a little low compared to Project totals:
----------------------------------------# Total Projects #___32,756,675___68,334 # Total Devices #___32,448,541___67,405 # Device shortfall #____308,134______929 The shortfall is many times recent daily totals so it is not just difference in timing. Looking at individual devices they appear to average out but need to do so over many days due to random delays in validations. (only done this by eye so far for a couple of single core devices - when I have time I'll find the code I did to add up time). Paul.
Paul.
|
||
|
|
Mamajuanauk
Master Cruncher United Kingdom Joined: Dec 15, 2012 Post Count: 1900 Status: Offline Project Badges:
|
For me the long term Device totals are a little low compared to Project totals: Hi Paul, my numbers were a snapshot from 19/01/14 - Lunchtime update today the 25th as this came up due to the MegaCruncher (4CPU/64Core) returning disapointng low stats. This was build on the 19th.# Total Projects #___32,756,675___68,334 # Total Devices #___32,448,541___67,405 # Device shortfall #____308,134______929 The shortfall is many times recent daily totals so it is not just difference in timing. Looking at individual devices they appear to average out but need to do so over many days due to random delays in validations. (only done this by eye so far for a couple of single core devices - when I have time I'll find the code I did to add up time). Paul. If you could do your stats over the same period it may provide a good comparison...
Mamajuanauk is the Name! Crunching is the Game!
|
||
|
|
PMH_UK
Veteran Cruncher UK Joined: Apr 26, 2007 Post Count: 786 Status: Offline Project Badges:
|
For 19th to 25th noon:
----------------------------------------Seconds Points Results 6,632,315 _ 193,139 _ 213 6,596,192 _ 192,466 _ 212 __ 36,123 _____ 673 ___ 1 Days per day will be less than elapsed as some PCs part time etc. To date: _ _ _ _ _ _ _ _ _ _ _ _ Points _ Results _ y d h m s _ Seconds # Total Projects # 32,756,675 _ 68,334 _ 43 130 18 45 12 _ 1,367,347,512 # Total Devices # 32,448,541 _ 67,405 _ 43 _ 5 2 27 11 _ 1,356,488,831 # Device shortfall # _ 308,134 ____ 929 _ 0 125 16 18 1 ____ 10,858,681 Paul.
Paul.
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7848 Status: Offline Project Badges: 
|
There is a definite discrepancy in the totals listed on the "My Grid" page and the totals for all the devices listed under the "Device Statistics" page. The "My Grid" page lists 43 devices for me, but the "Device Statistics" page only lists 42. Also the point totals are are different and the results are different. "My Grid" has 79,786,627 points and the "Device Statistics" page only totals 78,736,657, a difference of 1,049,970. "My Grid" results lists 185,457 while "Device Statistics" lists 182,228, a difference of 3,229. Boincstats comes much closer showing 79,673,424 (after adjustments for 7 WCG points equal 1 BOINC point and adjusting for the 1 United Devices cruncher not included in Boincstats figures). Much closer, but Boincstats always runs behind WCG in updating totals. In conclusion, there is some discrepancy, but I do not know where it is. On an individual machine basis, most of the Boincstats figures are higher than the WCG figures, even for machines which have been out of production for a long time. For instance, I have one machine which I took out of production in April of 2013, so all of the figures listed should be final. The WCG points listed are 3,620,393 which translates to 517,199 BOINC points while Boincstats shows 519,935 points, a difference of 2736. ( I considered rounding errors, but this seems like too big a discrepancy for that.)
----------------------------------------Note: I really do not care all that much about the points or the accounting, I just care that the crunching is getting results done, and the results are in some way meaningful to the researchers. However, I am still curious that this is not a symptom of some of the results getting lost in the system or there is some other shortcoming which is affecting the research. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Mamajuanauk
Master Cruncher United Kingdom Joined: Dec 15, 2012 Post Count: 1900 Status: Offline Project Badges:
|
Joe, this came to light with the rather low returns/stats of my latest build, on looking closer it is strange that many people are seeing the discrepency...
----------------------------------------Having said that, I agree the crunching is the important thing, also that the results are getting through to the team analysing them. As you say, I would like to understand where the differences come from
Mamajuanauk is the Name! Crunching is the Game!
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7848 Status: Offline Project Badges:
|
I have not done it, but you could look on several different sites and compare the figures for a particular machine. I am curious but not curious enough to do a full scale investigation. Since Boincstats gets all of its figures from WCG, I suspect there is an internal accounting problem at WCG. Perhaps some errors occurred in their database(s) during one of their updates or reconfigurings. I really would like to know where that 43rd machine is though. It is possible the data just got lost.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
We have reviewed and there appears to be incomplete data at the device level. We are looking at what happened and why our monitoring did not catch this issue.
|
||
|
|
Mamajuanauk
Master Cruncher United Kingdom Joined: Dec 15, 2012 Post Count: 1900 Status: Offline Project Badges:
|
We have reviewed and there appears to be incomplete data at the device level. We are looking at what happened and why our monitoring did not catch this issue. Many thanks for the feedback Keith... Please advise of the outcome after the investigation...
Mamajuanauk is the Name! Crunching is the Game!
|
||
|
|
|