| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: Africa Rainfall Project
Thread: Welcome to the Africa Rainfall Project |
| No member browsing this thread |
|
Thread Status: Active Thread Type: Sticky Thread Total posts in this thread: 68
|
|
| Author |
|
|
sjcsystems
Cruncher Joined: Dec 17, 2008 Post Count: 11 Status: Offline Project Badges: 
|
Any ideas when WU are going to be released for this project please? I've been signed up since the announcement and nothing here at all yet.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Yesterday at WCG 2,213 ARP1 got validated , a second highest since the project start. The WCG validation totals were Results Returned 1,265,129, so hardly 2%, which is slim.
Since the release is still in chunks at 8 minutes after the whole and half hour at nightly UTC hours, your computer has to ask there and then or they're gone. A change of the distribution has been in the works though... it's going to be randomly timed (see posts giba/uplinger) |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
Very interesting have a new Project to colaborate, but in this case the tasks are very rare, and have a strange schedule in its distribution. I have several machines in random fetch asking for new tasks of African Climate and since de day of start don1t get anyone after the start day. Could you please Professor, ask for the WCG change the way of distribution putting a randomic distribution during teh 24 hours of the day, for this Project. This simple change could improve the number of entusiasta to help the Project today or in the future, once will provide mnore equity in the distribution criterias which today appear be based on specific times. Think about please. This idea have the pótential to help much more your Project.      Giba, I see what you are saying about the distribution. We are hopeful that the limitation on the speed of this project will cause this to be a non issue in the future. I'll look into randomizing the release, no promises as it'll need to be something simple to change since I hope we can run more at a time in the future. (Note: I do not have an eta on when that future is) Thanks, -Uplinger Giba, I have a solution to randomizing the distribution of results in testing at the moment on our test environment. I'll let you know when it is pushed to production. Thanks, -Uplinger While at it, kindly apply to HSTB too, and make it random to the second if that's possible. This has been implemented for both HSTB and FAHB. Results should be loaded on a random sleep interval. This will help give everyone a chance to get work units. Thanks, -Uplinger |
||
|
|
TLD
Veteran Cruncher USA Joined: Jul 22, 2005 Post Count: 806 Status: Offline Project Badges: 
|
 Good news. |
||
|
|
giba
Veteran Cruncher Brazil Joined: Dec 2, 2004 Post Count: 846 Status: Offline Project Badges: 
|
Very interesting have a new Project to colaborate, but in this case the tasks are very rare, and have a strange schedule in its distribution. I have several machines in random fetch asking for new tasks of African Climate and since de day of start don1t get anyone after the start day. Could you please Professor, ask for the WCG change the way of distribution putting a randomic distribution during teh 24 hours of the day, for this Project. This simple change could improve the number of entusiasta to help the Project today or in the future, once will provide mnore equity in the distribution criterias which today appear be based on specific times. Think about please. This idea have the pótential to help much more your Project. Giba, I see what you are saying about the distribution. We are hopeful that the limitation on the speed of this project will cause this to be a non issue in the future. I'll look into randomizing the release, no promises as it'll need to be something simple to change since I hope we can run more at a time in the future. (Note: I do not have an eta on when that future is) Thanks, -Uplinger Giba, I have a solution to randomizing the distribution of results in testing at the moment on our test environment. I'll let you know when it is pushed to production. Thanks, -Uplinger While at it, kindly apply to HSTB too, and make it random to the second if that's possible. This has been implemented for both HSTB and FAHB. Results should be loaded on a random sleep interval. This will help give everyone a chance to get work units. Thanks, -Uplinger Uplinger, thanks to capture the idea behind the texts and in very quick time implemented the changes for this good way of tasks distributions; It will help many crunchers get tasks from this specifcs projects and help all contribute with more abrangency. It was very fair and I will let my team mates know about. Thank you very much !   |
||
|
|
LoneWolf_53
Senior Cruncher Joined: Nov 11, 2005 Post Count: 238 Status: Offline Project Badges: 
|
Any ideas when WU are going to be released for this project please? I've been signed up since the announcement and nothing here at all yet. Yes the work units for this project are scarce as hens teeth. I set my computers to run only ARP and they sat idle for four days, then I activated MIP again so they're doing something. Guess ARP isn't ready for prime time yet. ---------------------------------------- [Edit 1 times, last edit by LoneWolf_53 at Nov 24, 2019 2:38:11 AM] |
||
|
|
tazzduke
Advanced Cruncher Australia Joined: Aug 8, 2009 Post Count: 56 Status: Offline Project Badges: 
|
Greetings All
----------------------------------------Yep same here, all excited for this new project but really scarce. So I started off filling up my caches with MIP (2 days worth for me) and then unselected MIP and selected ARP and HSTB, and sooner or later I managed to snag 1 or 2 but only on two of my 4 machines. Its just a case on timing and luck till there is a constant flow. Even managed to snag a couple of HSTB workunits which is pretty cool as well. Cheers PS, I only said MIP as thats the one I am working on of the 2 that are still flowing full time, but I have switched one machine across to MCM.  |
||
|
|
rgarvey
Cruncher Joined: Nov 22, 2004 Post Count: 23 Status: Offline Project Badges:
|
It seems that this project is being assigned to CPUs that cannot return a result in the time allowed when the task is not assigned based on the CPU involved.
----------------------------------------An African Rainfall Project task that takes 23 hours on a modern CPU is taking 5 days on older processors. There does not seem to be a system requirement for processor age/capabilities and/or a BOINiC scheduling adjustment for aging processors to account for the many day processing lag for older generations. A system requirement for a minimum processor seems to be required as well to avoid "No reply" wasted effort  |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7670 Status: Offline Project Badges:
|
It seems that this project is being assigned to CPUs that cannot return a result in the time allowed when the task is not assigned based on the CPU involved. An African Rainfall Project task that takes 23 hours on a modern CPU is taking 5 days on older processors. There does not seem to be a system requirement for processor age/capabilities and/or a BOINiC scheduling adjustment for aging processors to account for the many day processing lag for older generations. A system requirement for a minimum processor seems to be required as well to avoid "No reply" wasted effort There is currently not a way to assign work units based on processor. The issue has been discussed in the past, but has not been deemed a priority at this time. These work units are set for a 7 day turnaround. That is more than enough time for even an older processor to complete a unit (depending, of course, on the amount of time the unit is turned on). The problem stems from the queue being set to too high a number of days. For instance, if the work unit sits in the queue for 4 days before it is started, then you only have 3 days to complete the work unit, which in some cases will not be enough time. Proper queue management is the key here, especially noting the lengthy time between checkpoints. If you know your machine is not going to be able to process these units in a timely fashion, you should probably not opt in to this project, even if you really believe in the cause. Also, if the techs start seeing an inordinate number of "no reply" responses, I am quite sure they would make some adjustments. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
rgarvey
Cruncher Joined: Nov 22, 2004 Post Count: 23 Status: Offline Project Badges:
|
Got it, except when there is one system with Intel(R) Xeon(tm) CPU 2.66GHz in pool, while all others are more suitable, BIONIC settings don't currently allow this system to exclude it self from this work load based on CPU while allowing all others to take on the task.....
---------------------------------------- |
||
|
|
|