| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Discovering Dengue Drugs - Together - Phase 2 Forum
Thread: DDDT-2 Work Unit distribution updated |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 121
|
|
| Author |
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
Greetings all,
We are making a few changes to how work units are distributed to members for DDDT-2. The reason for these changes, we would like to have more members being able to participate. We have been saying that when the type C's come, more work will be available to the members. So with that thought in mind, we are going to make these changes to help get the type C work units running. 1. New type A work units will be sent to reliable hosts only with a 7 day turnaround. This means they will get scheduled first on the grid to the generally faster 24/7 machines on the grid. 2. Type B work units will be sent out with priority 1. This means they will go to all members, but will be pushed to the top of the queue when sending work units, meaning they'll get B work units if they're available over type C work units. 3. On type A work units, some error out within seconds of the work unit starting. These work units are automatically counted as complete on our end to help off load the batches quicker with less intervention on our part. 4. If a type A work unit has completely exit 29 out and there is runtime associated with it, then credit is granted to all members on that work unit. We realise these work units may have run 24 hours before failing. So members should get credit for this work. (Please note that this is starting now) We are still working closely with the researchers in hopes to get more new work. Please be patient with us and remember, this project will have periods of no work being available (this being one of them). Let me know if you have any questions, -Uplinger |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
-Uplinger,
----------------------------------------From this mornings validations, 54, with an avg run time of 43 hours each one can derive these were technically of the Too Late / No Reply class, over 14 days after distribution. Unless there's a hard cap on Too Late, think there continue to be group 25 hold ups, which as you computed there having been only a 3% statistical chance at the end of the original 14 period for a group to come together on the first round, which was thus 1 of 40 groups to be complete. Suggestion: Stop announcing A / B type batch launches and you'll find that backfilling of work will be on a one/few per device basis, then there being a much wider quorum finding in the first 48-72 hours also on this A/B part of the project, also for those who like contributing in set and forget and no less dedicated to WCG. 2 cents on getting the science done.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
pball1224
Cruncher USA Joined: Nov 22, 2005 Post Count: 45 Status: Offline Project Badges: 
|
I like what I see in #1 and #2 !!!
----------------------------------------But I almost wonder if B units should be limited to reliable machines too. I wonder if any tech could tell us what percentage of all active machines are currently on the reliable list... [Edit 1 times, last edit by pball1224 at Mar 4, 2010 5:07:17 PM] |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges: 
|
Suggestion: Stop announcing A / B type batch launches and you'll find that backfilling of work will be on a one/few per device basis, then there being a much wider quorum finding in the first 48-72 hours also on this A/B part of the project, also for those who like contributing in set and forget and no less dedicated to WCG. I agree completely. This project is in production. Stop allowing lurkers (like me  ) to get an advantage on wu's. It may reduce the over caching problem too and allow more to participate.My two cents.    |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
I can keep the announcements to a minimum. The only reason I announced was I thought members would enjoy it :)
There is a little wiggle room on the too lates. I think it is set similar to the other projects where you still get credit if returned within 24 -48 hours of the deadline. Don't hold me to that as I'm not 100% sure on that. As of right now, I still don't show any new batches being complete. Remember patience is needed :) Also, we thought about setting type B to reliable only, but decided against it at this time. We may revisit this later when we get actual data back in on type B batches. Thanks, -Uplinger |
||
|
|
Dataman
Ace Cruncher Joined: Nov 16, 2004 Post Count: 4865 Status: Offline Project Badges:
|
I can keep the announcements to a minimum. The only reason I announced was I thought members would enjoy it :) Great. There is a little wiggle room on the too lates. I think it is set similar to the other projects where you still get credit if returned within 24 -48 hours of the deadline. Don't hold me to that as I'm not 100% sure on that. As of right now, I still don't show any new batches being complete. Remember patience is needed :) Also, we thought about setting type B to reliable only, but decided against it at this time. We may revisit this later when we get actual data back in on type B batches. Thanks, -Uplinger I think not announcing the release of new batches would help level the playing field for all and not give "retired old men with too much time on their hands" (like me )a distinct advantage at snatching work within moments of its release.BTW: It's not like me to be this alturistic.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hmmm,
don't you overestimate that 'reliable computer' effect? I bet most of those computers who got the lions share of the WUs were reliable (at least when they hoovered all those WUs). And a computer can become reliable within 1-2 days when running 24/7. So the same computers will get most of the work... :-( IMHO the 7 day deadline is much more effective. |
||
|
|
HutchNYC
Advanced Cruncher United States Joined: Nov 27, 2005 Post Count: 97 Status: Offline Project Badges: 
|
IMHO the 7 day deadline is much more effective. While I agree with the objective behind a 7 day deadline, I think in reality it would make things even worse overall. From just the first short release of the initial 1000 Type "A's", nearly 25% weren't returned in time with a 14 day deadline. I think cutting the 14 day deadline in half would only compound the issue. Seeing how this is only an "issue" here in the beginning where there is the least amount of work available, once the pipeline gets rolling and production of A's, B's, and C's all at the same time goes into effect, it is possible that in a few weeks it will be moot as much more work will be available for everyone. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The main benefit of 7 day reply is that a member cannot scoop up more than 7 days of work (per core) at once, so the scheduler will get to spread a wider net to more computers.
|
||
|
|
Hypernova
Master Cruncher Audaces Fortuna Juvat ! Vaud - Switzerland Joined: Dec 16, 2008 Post Count: 1908 Status: Offline Project Badges: 
|
Hey, doctor there's a pulse!
----------------------------------------  |
||
|
|
|