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: 3204
|
![]() |
Author |
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12359 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
There are now 3,361,696 units to go.
On the basis of 2 results per unit, the average returned over the last 10 days has increased to 10,077.8 units. For 7 days there were over 20,000 results returned. This would mean an end date of 7 August 2022, however Kevin has explained that the extra is intermittent and already seems to have stopped. However, this also would require for there not to be a tail off at the end, which is highly unlikely as all projects have a tail off at the end and ARP1 is more likely to tail off because of the stragglers. That is not referring to the current stragglers but to those units from the last few generations at the end. This also does not allow for invalid results or errors. I expect the end to be nearer to the end of 2022. I will try to update this weekly or whenever Kevin might update us with the stragglers. Mike |
||
|
leloft
Cruncher Joined: Jun 8, 2017 Post Count: 23 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Thanks for the update.
They have their system set up to download a single job, run immediately it and then return it (I believe they use the command line options --allow_multiple_clients --fetch_minimal_work --exit_when_idle and have each client running look like a single core machine). That's interesting. How does that differ from simply running N concurrent ARP workunits and no other type of workunit? I am starting to see a pattern from the various N/0/0 (ARP/OPN/MCM) strategies. 1) As expected, if N is greater than half the number of cores, the processing time for each workunit increases noticeably with each additional workunit. 2) However, if N is less than half the number of cores, the processing time is not decreased as noticeably; although up to a point, there may be a slight decrease (but see below). 3) If the number of workunits running falls below a (as yet undetermined, very probably machine specific) threshold, the Intel (c) TurboBoost doesn't kick in; paradoxically below this threshold, it takes longer to process a fewer number of workunits than a higher number above it (subject to (1) above). 4) It is easily possible to operate a 'download on demand' policy through the use of the various Device Profiles where the value of N can be varied by ticking the ARP box only (also actively exclude the possibility of downloading other projects during a hiatus in workunit availability such as we saw a couple of days ago) and setting N with the dropdown menu limiting the number of concurrent ARP workunits. This makes the use of app_config redundant. The time 'lost' from upload of a completed unit to running of a new one is of the order of 3 minutes, which is not significant with a processing time of 11h, and less so with one of 27h. 5) It has also helped to keep the machines cooler during the hot spell; I haven't detected any thermal throttling, but the i7 cores run at 90 deg on 2N/0/0 and <75 with N/0/0, with no real loss of productivity: 8 units every 30h and 4 units every 16h respectively. Best leloft |
||
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
What's going on? I'm not getting any new WUs! It's happening again. Edit: And as I post, I get some to fill my queue! ![]() ---------------------------------------- [Edit 1 times, last edit by BladeD at Sep 13, 2021 6:50:31 AM] |
||
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12359 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
There is only a finite number of work units ever available. 35,609 x 2 = 71,218 to be precise. Once they have gone out, we have to wait for some to be returned. What slows things down is that whole pairs have to be returned and validated before the next generation can be created.
Anyone who holds more than 1 spare is slowing down this process. Mike |
||
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Latest stats:
Average generation: 89.9 Pace (average time to complete a generation): 4.78 (7-day average) first_indexed generation num_units_currently_on_generation num_units_completed_last_day |
||
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12359 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Thank you, Kevin. Are you moving to Toronto?
I see that all the generation 001's have started to move. 087 is the latest generation identified as labelled 'priority' but I will assume that 088 has now reached that status. I will base this response on that. If anyone knows of a later batch please advise and I will adjust my figures. 27% of the units are in the priority generations. There are now 3,316,732 Units to be crunched to finish the project. The average number returned since the last of Kevin occasional high performance machine last stopped was 7891.6 units per day so my target for completion is now calculated as 8 November 2022. This assumes no invalids or errors are included in those returns. I suspect we are unlikely to see the project finish before the end of 2022. Mike |
||
|
alanb1951
Veteran Cruncher Joined: Jan 20, 2006 Post Count: 952 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Mike,
----------------------------------------I doubt this is actually of any use in your analysis, but it can resolve your assumption about generation 088.... I received a full-duration 088 task at about 11:00 [UTC] on 14th September and a priority 088 task first thing on 15th September, so that transition actually happened about 12..24 hours after you posted (which is, I suspect, unlikely to disrupt your analysis!) Thanks for all the effort you put in here and on the OPN1/G threads... Cheers - Al. [Edit for missing words!] [Edit 1 times, last edit by alanb1951 at Sep 15, 2021 2:51:17 PM] |
||
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12359 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Thanks, Al.
I thought it was about time for 088, but I don't seem to get many new priority units, I mostly get resends, which, although they are priority cases, don't classify as priority generations. I run a PC with 4 of its 8 threads on ARP with a cache of 1 extra unit and the rest onOPN, MCM and occasionally HST, each with 1 spare unit. My laptop overheats periodically so I don't run ARP on it, nor on my phone or 2 tablets. The extra tablet is because I cracked the screen but it still crunches although operating it is problematic. I reiterate a previous post for general consumption by saying that the occasional lack of units is because there is a maximum of 71,218 units available (35,609 x 2). New units cannot be created until both of the previous pair have been returned and validated. Anyone who hold a substantial cache is cutting down that availability. Please cut your cache to a maximum of one more than the number of threads actually used for ARP. That way we boost the availability and crunchers can get their units validated more quickly. Mike |
||
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1320 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
088 is surely now priority batch. I got 8 tasks with a shorter deadline. 2 of them:
ARP1_0014526_088_0 deadline 19 Sep 19:46:56 |
||
|
Mike.Gibson
Ace Cruncher England Joined: Aug 23, 2007 Post Count: 12359 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Just to confirm that at the time of Kevin's post, 088 was not a priority generation, so only 8% of the then current units were from priority generations.
Since then 088 has become a priority generation but some of the 27% will have moved on. Mike |
||
|
|
![]() |