| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: Long computing times with short Due date? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 98
|
|
| Author |
|
|
enels
Senior Cruncher Joined: Apr 25, 2008 Post Count: 286 Status: Offline Project Badges: 
|
Isn't the trickle up supposed to take care of this? We get credit up to our last report? Although I haven't seen this in action.
|
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
There will be some proration factor against the current grant for a full 100K result. It's a dynamic value, and not all take quasi equal time to compute on the same device, but given the screw-ball ways of work assignment, i.e. unstable/buggy average runtime tracking, someone getting 120 WUs on a 0.5 day buffer device, considering another project with 'anything goes at WCG'. Bottom line: Credit [points, if that was what the comment included] continues a waste of time to pense over.
|
||
|
|
RTorpey
Advanced Cruncher Joined: Aug 24, 2005 Post Count: 67 Status: Offline Project Badges: 
|
So, why are we now seeing "Aborted by project" for WU's due on 10/9?
|
||
|
|
deltavee
Ace Cruncher Texas Hill Country Joined: Nov 17, 2004 Post Count: 4891 Status: Offline Project Badges: 
|
So, why are we now seeing "Aborted by project" for WU's due on 10/9? Here's one reason why. The number 1 wingman went valid before I started crunching. The server didn't want me to waste my time so it aborted the workunit. This is not so unusual with resends.FAH2_ avx101140-ls_ 000074_ 0004_ 001_ 2-- 714 Server Aborted 10/5/15 18:15:43 10/6/15 09:29:48 0.00 0.0 / 0.0 <--me FAH2_ avx101140-ls_ 000074_ 0004_ 001_ 1-- 714 Valid 10/5/15 13:32:45 10/6/15 09:22:43 12.94 367.3 / 367.3 FAH2_ avx101140-ls_ 000074_ 0004_ 001_ 0-- - Invalid 10/1/15 16:19:06 10/5/15 18:13:21 0.00 0.0 / 367.3 |
||
|
|
Hans-Martin
Cruncher Germany Joined: Nov 29, 2013 Post Count: 10 Status: Offline Project Badges: 
|
I basically understand the reasoning behind the work unit schedule, but it's still sad that it is really incompatible with my computer use pattern (home desktop PC, BOINC only running while I'm using the computer for activities that demand little CPU power). With just 2-3 hours of runtime per day, FAH2 WUs can't be finished. The one WU currently being worked on can be finished on time because it's a weekend, but after that, no more FAH2 for me :-(
Hans-Martin |
||
|
|
deltavee
Ace Cruncher Texas Hill Country Joined: Nov 17, 2004 Post Count: 4891 Status: Offline Project Badges:
|
I basically understand the reasoning behind the work unit schedule, but it's still sad that it is really incompatible with my computer use pattern (home desktop PC, BOINC only running while I'm using the computer for activities that demand little CPU power). With just 2-3 hours of runtime per day, FAH2 WUs can't be finished. The one WU currently being worked on can be finished on time because it's a weekend, but after that, no more FAH2 for me :-( Hans-Martin This project should be perfect for you because it uploads results with every 10% of workunit completion. Even if you don't reach 100% you have still made a contribution and get credit for it. From a previous post: What's going on is that it doesn't matter to the WCG that the WUs aren't finishing. The WU will not be resent. Instead a new WU will be sent that starts where the old one left off. No work is wasted. No work is duplicated. The science moves along faster. Everyone gets credit for the work they accomplished, even if the WU did not finish. |
||
|
|
Hans-Martin
Cruncher Germany Joined: Nov 29, 2013 Post Count: 10 Status: Offline Project Badges:
|
Hmmm... still feels a bit dissatisfying to not be able to complete a WU, but I'll get over it. If the work isn't lost it's good enough :-)
(Project enabled again) Cheers, Hans-Martin |
||
|
|
enels
Senior Cruncher Joined: Apr 25, 2008 Post Count: 286 Status: Offline Project Badges:
|
In case you aren't following the Oct 8 Beta thread, it looks like a big speedup for Linux is in the works:
----------------------------------------http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,38490 edit: for Linux [Edit 1 times, last edit by enels at Oct 10, 2015 7:12:06 PM] |
||
|
|
NixChix
Veteran Cruncher United States Joined: Apr 29, 2007 Post Count: 1187 Status: Offline Project Badges: 
|
What's going on is that it doesn't matter to the WCG that the WUs aren't finishing. The WU will not be resent. Instead a new WU will be sent that starts where the old one left off. No work is wasted. No work is duplicated. The science moves along faster. Everyone gets credit for the work they accomplished, even if the WU did not finish. Yes, there will always be work wasted under this method, up to 9.99%. On some of my slower machines that could be several hours. I've already experienced that. Are there more FLOPS per day being processed? Exactly how does this move the science along faster? It seems that the science is actually slowed by throwing away alot of work because we're in such a hurry. Cheers   |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
With the soft stop methodology nothing gets throw away, nothing. The client gets an instruction to finish next trickle, then stops. Only when that next trickle is reported, is the next task generated starting at the point where the last reported trickle ended.
|
||
|
|
|