| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: Africa Rainfall Project
Thread: Work Available |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 3595
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
nanoprobe - And just how do you suppose they are doing that? Sounds like you suspect they are somehow cheating. Wow that's a stretch, thanks for just taking that off to some wile conclusion of your own - I made an observation and possible theory which has a plausible answer. Moving on... DrMason - I get maybe 10 units a day to process, so not a huge amount of the daily amounts. Even with 176 dedicated cores, I don't have sapphire yet I run 44 dedicated cores (40 on one beefcake, 4 on a laptop - all Debian 10) and have seen 3 ARP units total make it my way so far since the beginning. These two devices turn over only about 250-300 units per day (usually closer to 250, sometimes lower), not a huge performer at all but it's strange how you can get 10 a day and I've seen 3 total since we started ARP. :) |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
And just how do you suppose they are doing that? You didn't answer my question. Please explain how you think there are teams in your challenge( your exact words>) pulling every single ARP unit the minute it's offered? but it's strange how you can get 10 a day and I've seen 3 total since we started ARP. Maybe because I have 270 cores at my disposal.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
----------------------------------------  [Edit 5 times, last edit by nanoprobe at Dec 2, 2019 11:47:23 PM] |
||
|
|
DrMason
Senior Cruncher Joined: Mar 16, 2007 Post Count: 153 Status: Offline Project Badges: 
|
DrMason - I get maybe 10 units a day to process, so not a huge amount of the daily amounts. Even with 176 dedicated cores, I don't have sapphire yet I run 44 dedicated cores (40 on one beefcake, 4 on a laptop - all Debian 10) and have seen 3 ARP units total make it my way so far since the beginning. These two devices turn over only about 250-300 units per day (usually closer to 250, sometimes lower), not a huge performer at all but it's strange how you can get 10 a day and I've seen 3 total since we started ARP. :) I wish I had an answer for the alleged disparity. (Don't take "alleged" the wrong way - there may be a disparity, but I don't know all of the variables so I don't have a way of proving or disproving whether there actually is one.) If I had an answer, I would offer it. To the extent that I had information and speculation, I've provided my perspective here and in other threads. I understand that we're all wanting more work units, but I feel like there's a leap here going from "Huh, some people are getting more work units than others. That's interesting and I wonder what might be causing that. Let's seek information and crunch some data to find out what we can about the algorithm!" to "I'm getting less than others so I suspect that there is some sort of unfair play and I'm getting the short shrift." The only thing I have to speculate is that I have more cores turning over work faster (average of about 950 a day), and my machines are largely dedicated to only crunching WCG, so they're more frequently available to receive work. *shrug*  [Edit 1 times, last edit by DrMason at Dec 3, 2019 12:31:11 AM] |
||
|
|
DCS1955
Veteran Cruncher USA Joined: May 24, 2016 Post Count: 668 Status: Offline Project Badges:
|
Today I got 5 WU's, my first of the randomized release. Two reconditioned computers circa 2012 &2014 running Windows 10.
----------------------------------------  |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges:
|
The only thing I have to speculate is that I have more cores turning over work faster (average of about 950 a day), and my machines are largely dedicated to only crunching WCG, so they're more frequently available to receive work. *shrug* BINGO!
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges: 
|
I usually have between 90 to 100 cores running, 80 of them run 24/7. I usually run with about 1/2 day queue. The key I think is to limit the work units called for. For instance, on one of my 8 core machines I have 4 mcm, 4 mips, 1 ARP, and 1 TB listed in the device profile. This means I have a maximum of 10 work units on the machine at any one time. The MCM and MIPS units keep my minimum number of tasks at 8, one unit for every thread. These units are very available, so I am pretty much never short of having all my threads full. However, since my queue minimum is 1/2 day, BOINC manager will query looking for at least 1 more ARP and one more TB unit at whatever interval it is programmed to do, something like once every ten minutes or so I think. If it gets an ARP and a TB unit, it is full. If it does not, it waits the approriate interval and queries again. If one of the MCM or MIPS units finishes, then it will query for one of them. If I get an ARP or TB unit, they are so long , that it will only then query for the much shorter MCM and MIPS units. If I have both an ARP and TB unit in the queue, then the only thing I am querying for is either MIPS or MCM.
----------------------------------------On the 24 thread machines the maximum number of work units held in the queue is 31, 2 ARP, 2 TB, 20 MCM and 7 MIPS. This is enough to keep the machine running on all threads and do enough querying to get maybe 1 or 2 ARP per day some days and about 1 TB unit every other day. I am not trying to hog any of these units, but just let them come as they may. Even if I get neither ARP nor TB, the queue stays full enough to keep every thread busy. Good luck on your search for them. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
DCS1955
Veteran Cruncher USA Joined: May 24, 2016 Post Count: 668 Status: Offline Project Badges:
|
That makes sense SGT. Joe. I got my silver by scheduling. I prefer the luck of the draw + ready to pounce.
----------------------------------------[Edit 3 times, last edit by dcs1955 at Dec 3, 2019 6:17:33 AM] |
||
|
|
Jack007
Master Cruncher CANADA Joined: Feb 25, 2005 Post Count: 1604 Status: Offline Project Badges: 
|
I got rid of my queue, and that seems to help a bit.
----------------------------------------snagged 5 arp and 3 TB and with the 10 MIP, I actually had 2 paused while 2 more finished (16 cores).  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So I just excluded ARP from the projects I run, as many of you want them so badly.. Not that I crunched many... MCM and MIP have plenty of important work anyway, so I concentrate on these.
ARP seems to be a long-term project, so I will postpone getting some badges until there is more work to do. |
||
|
|
5TEVE
Cruncher Joined: Sep 4, 2006 Post Count: 34 Status: Offline Project Badges:
|
My team is in one of the challenges for ARP only and it's interesting that out of 31 teams, only 8 of those teams even get ARP units and out of those 8, one team has over 4 times as many ARP than the next team. I suspect there are machines out there pulling every single ARP unit the minute it's offered, making the odds swing wildly away from 1 in 1000. 23 teams have zero ARP results (although it's only been a day or two since the challenge started). https://www.worldcommunitygrid.org/team/chall...lenge.do?challengeId=9745 That's a pretty top-heavy looking list of teams and doesn't "seem like natural distribution" to my eye. $0.02 Personally I think it's Ludicrous that a ARP Challenge would be run this early with the limited amount of Wu's being released ...  |
||
|
|
|