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: 6
|
![]() |
Author |
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I have a triple core AMD and the buffer is set to .2 days. That means the buffer will request another WU when the time remaining reaches 14:24 hours (3 CPU * .2 * 24hours). The last WU I received is estimated at 5:51. Therefore the maximum time I think I should have had is 14.4 hours plus 5:51 for a total of 20:15 (give or take some minutes of overlap).
Currently I have 3 IP and 3 waiting (all NRW) and the time remaining for all 6 WU is 22:42. The last WU was DL 1:50 hours ago (requesting 72 seconds) meaning I have also chewed up another 5:30 since the DL (3 * 1:50). All 6 are estimated total time at 5:51 - 6:01). Therefore at the time the last WU was downloaded, I had a total of 28:12 of work (22:42 + 5:30) when my buffer is at 14:24. I have no local preferences in place since the last time I synched to my online Default setting was 21 hours before that last WU DLd and my connect time is .01 days. This should guarantee I am on the online default settings. I realize the answer could be explained by something with the buffer size being out, but I can't see where the difference is generated, or where my thinking might be wrong. |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Is this client 7.5.0 ?
----------------------------------------BarneyBadass likes these challenges I think. It's an area I understand and have learned to ignore. Note: Each completed result the rDCF changes too, thus results in buffer have their TTC times adjusted too. The Connect of 0.01 also funks into the equation, lest that changed too. Added: the client will request X amount of work, but the duration of work then received is not that exact length.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Mar 19, 2009 3:30:32 PM] |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Lucky for you, client 6.6.16 alpha and up show Elapsed time not CPU time. I've compared this value with what can be seen in the process explorer properties view, where BOINC CPU time agrees with the CPU User time.
----------------------------------------The definition for Elapsed time I've not looked up in so far that it seems to account for the total time a science is allowed to run including CPU inefficiency. On one of the current jobs it's 6 minutes more than the CPU time shown in the properties view.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
p3nguin53
Advanced Cruncher USA Joined: Dec 8, 2008 Post Count: 95 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I've got my buffer and connect time both set to .05 for my duo core. It looks like my downloads occur based on the formula (1CPU * .05 * 24h) instead of the formula (2CPU * .05 * 24h) as noted above.
On my machine, when an IP WU reaches about 80 min remaining another WU is downloaded. It doesn't matter how much time is left on the other IP WU. So each core has one WU running and maybe one WU waiting in the wings. I had my buffer and connect time set to .5 until a few weeks ago. I looked back thru my log and I've always downloaded only 1 WU except when the server was down. So it looks like it worked the same way with the higher settings. So does the buffer always look at time remaining for 1 core no matter how many cores you have or does it use a different formula depending on the number of cores? |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
It really uses the a multi core formula. Ignoring the fractions, the essence is that a cache set to 1 day on a 4 core device will strive to have 4 times 1 core day computing in the queue across all active projects attached to a client. The formula is tweaked constantly, so release 6 will show yet again different behavior due further refinements.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Is this client 7.5.0 ? 6.2.28 Windows Vista After the next WU was DLd, the WU waiting went from 6:09 to 6:51 and after chewing for 1:28 the WU waiting are now showing 6:27. So there is something changing on the fly. What makes me more interested is that this computer is 100% 24x7 on a battery backup and I have not changed the screen saver/power settings for months, all of which suggest consistency and stability, especially on NRW. This computer has done 116 WU since 07/03/09 23:15:32 all of which are NRW except 6 HFCC (ranging from 2.68 to 6.63 hours) and all are Valid, except 7 PV and the now 5 IP, no errors/invalid. [Edit 2 times, last edit by Former Member at Mar 19, 2009 5:55:45 PM] |
||
|
|
![]() |