| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home
Thread: Release of new FightAIDS@Home application versions on BOINC |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 43
|
|
| Author |
|
|
Protego
Cruncher Joined: Apr 26, 2007 Post Count: 33 Status: Offline |
I have switched all slow computers to temporarily run on the Human Proteome Folding project, and the fast ones now load a single work unit at a time.
----------------------------------------I am puzzled by these notes, that each work unit is actually duplicated. I thought that checking an obtained solution to a work unit was quick, so it took place in the project server. Clearly, a brilliant programmer, could speed up the computation if we could get rid of this duplication issue.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Protego,
I think that you are referring to the HPF2 project. Each work unit is sent out ?19? times, but each computer uses a different random variable, which produces a slightly different result. Then a complicated verification program checks to see if the results are ?similar?. If they differ too much, they are considered to be errors. The end result is that there is no duplication of results for HPF2, unlike all our other projects. Lawrence |
||
|
|
Movieman
Veteran Cruncher Joined: Sep 9, 2006 Post Count: 1042 Status: Offline |
Lawrence:
----------------------------------------Suggestion, Go back to 3 days. This 4,then3,then2, is creating a mess. Seroiusly, I'm carrying a 2 day queue of 50+FAAH WU a day and I can't keep up with your changes. Using return dates as an example: I'll have a queue of 100 WU dated for the 11th, you send in say 20 more that are dated the 10th, preempts the 11's, so I suspend the 10's till I finish the 11's that are already running, then resume the 10's, then you send in 9's and the problem compiles all over again... You need to add in one more day to your return time so that the machines can complete what they have to keep up and I'm turning 8 Faah WU every 2.75-3.5 hours on that machine.. I finally set the machines to "don't get new work" to clean up this mess. Then tonight I get 50 HP whatever WU as I think you ran out of FAAH and then 10 mins later I start getting FAAH units again. I won't need to see my barber this month because I've pulled out all my hair!  Help a fella out here, add an extra day to the return time. Thanks!   [Edit 1 times, last edit by Movieman at Aug 8, 2007 5:47:31 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Movieman, you are a victim of micromanaging. The simple fact is, your FAAH queue MUST run dry. Deadlines will return to normal with the new FAAH application version. BOINC is quite happy to attempt to make all the deadlines on its own. It may fail, with these short deadlines - but you won't be penalised for that.
I suspect the odd rescheduling behaviour is caused by your particular version of BOINC. The behaviour has changed so many times that I can't remember which behaviour goes with which version. Just let it run.... |
||
|
|
Protego
Cruncher Joined: Apr 26, 2007 Post Count: 33 Status: Offline |
W.U.:s which is in error, is inconclusive, is late, or have some other problem, need to be sent out again, so that all WU complete OK before the switch. I think.
----------------------------------------So your fast machine would be good to clear up the queue. Some recommendation(s) from The Server Administrators would be appreciated here. Meanwhile I have the same problem as you have, on a much smaller scale: Two work units, complete normally in 45 hours, one aported due to the shorter deadline, and both work units are now stuck in the machine. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Seems all my 'old' FA@H jobs that sat in Pending Validation for near a week, were validated last night with quorum 2, the 3rd waiting 'In Progress'.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So has the change started? I have received some Folding WU's but am not checked for these, other than the send me something if my choice is not availabe.
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Think the dry-up has started and the hiatus of 1 day for FA@H is underway.
----------------------------------------Think the validation 2 for old FA@Hs is an accelerated way to clean the system. Anyone who has not returned an older FA@H result is not likely to ask for a new one, or get one because of the programmed gap. If returned later it simply will get the credit of the canonical, or find a 'too late', ?no credit award?
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Aug 8, 2007 11:39:55 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
...however another of my machines has just received some more of the FAAH ones?
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Just have a look in the Result Status page WU detail if those are backups for missing items (no reply or error e.g.) to make up quorum 3 or some other filler. Don't abort as the system will keep on sending new copies until it gets what it wants.... and that cycle ends i think at 4 or 5 errored returns (and dries up the client supply for a while if done too many from 1 machine).
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|