| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: Does offline calculation results in invalids? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 60
|
|
| Author |
|
|
phytell
Cruncher Joined: Sep 8, 2014 Post Count: 39 Status: Offline |
Can anyone confirm for me whether invalid results for this project are actually doing anything for the research?
If everything that's being sent in is still making a difference then I'm happy, but if invalid means "doesn't matter and will be redone", then I'm wasting a whole bunch of time and there's really no reason to continue working on this project until the invalid issue is resolved (reached silver badge recently on 6 threads - if that's wasted its going to be rather disheartening). |
||
|
|
petehardy
Senior Cruncher USA Joined: May 4, 2007 Post Count: 318 Status: Offline Project Badges: 
|
Got my first Invalid.
----------------------------------------Result Name: FAH2_ avx38783-ls_ 000030_ 0013_ 024_ wcgfahb00020000_ 0--  "Patience is a virtue", I can't wait to learn it! |
||
|
|
phytell
Cruncher Joined: Sep 8, 2014 Post Count: 39 Status: Offline |
Update: results appear to validate if and only if:
(a) all trickles are reported via scheduler request prior to the unit itself (b) there is a delay between when trickles are reported and when the WU returns In both cases, the valid units were reported several hours after their trickles, leading me to believe that the delay should be substantial. Both of these units were calculated offline, with network connection only used to refresh the cache and report completion. Since invalid units are wasting dozens of hours (I recently returned a 63 hour WU) and nobody seems to be able to tell me whether these units are actually doing anything or not, it looks like this project is going to be require a bit to much management for me. I'll be watching these forums in hopes of a fix for this, but if none comes, best of luck to the rest of the crunchers. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
This all just gets more and more stupid as time goes on ....
If it sees a net disconnection that makes the WU invalid it should abort the WU post a message under the NOTICE's tab so we can read it and allow us to correct it or not..... I wish the people in charge of all this would start doing their jobs I no longer have questions as to why most people just drop WCG |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges: 
|
This all just gets more and more stupid as time goes on .... If it sees a net disconnection that makes the WU invalid it should abort the WU post a message under the NOTICE's tab so we can read it and allow us to correct it or not..... I wish the people in charge of all this would start doing their jobs I no longer have questions as to why most people just drop WCG I am sorry to hear you are having problems with this project as well as CEP2. This is a project which is still rather new, so all of the bugs are not fully crushed. Others have mentioned the problems with having intermittent internet service or other breaks having to do trickles and reporting. By WCG standards these are heavier than normal WU's which take more time than many of the other projects. This is also a more complex project with having to string together the results from 30 or 40 separate WU's to get a complete job finished. It appears like most of the crunchers are sending back valid units, because if there was a high error rate the techs would stop the project to figure out the problem. They do not want to see wasted cycles any more than you do. If this project is not compatible with your setup, for what ever reason, opt out. Choose a diferent project that is not so demanding of your resources. There is no use in getting high blood pressure over something that is bothering you. The techs would like to see all of the projects go smoothly for everyone, but that is probably an unrealistic goal due to the myriad of different setups that are crunching. At the very least just calling the system stupid without offering any possible solutions gets us nowhere. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sgt Joe it all boils down to one thing .... keeping the numbers up ....
----------------------------------------If not having a putter hooked to the net can cause a invalid result they could of told us in the notices section as fast as they do when there is some new video or news feed If all that matters REALLY is the first save on CHP2 then why not let us be sure to crunch that out and then impose a 18 hour limit and avoid sending valid WU's back out for re-validation until all good again It's the only valid reason I can find..... If you divide the daily CPU time by the number of computers connected on WCG (data found at) Community page top page they all work something close to 30 min a day each .... REALLY is that all we do ??? Posting the number of active computers just wouldn't look as IMPRESSIVE [Edit 1 times, last edit by Former Member at Dec 13, 2015 7:40:39 PM] |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges:
|
Sgt Joe it all boils down to one thing .... keeping the numbers up .... If not having a putter hooked to the net can cause a invalid result they could of told us in the notices section as fast as they do when there is some new video or news feed If all that matters REALLY is the first save on CHP2 then why not let us be sure to crunch that out and then impose a 18 hour limit and avoid sending valid WU's back out for re-validation until all good again It's the only valid reason I can find..... If you divide the daily CPU time by the number of computers connected on WCG (data found at) Community page top page they all work something close to 30 min a day each .... REALLY is that all we do ??? Posting the number of active computers just wouldn't look as IMPRESSIVE These figures are pretty old news. There are about 11% of the total users who are active. There is about 8% of the total hosts (about 2.5 million) which are active( about 200,000). Users come and go. For instance on the day I started, July 4, 2006, out of the 82 people who started on that day only 8 remain active, slightly less than 10%. I do agree the total hosts figure is deceptive. Just for me I have 48 device installations listed, but have retired 39 of them, either because they went belly up, or because they were too old to be effective crunchers any more. If you have any ideas on how to retain crunchers or make them more involved, don't hesitate to speak up. One of the areas frequently mentioned is more and better communication from the project scientists and better communication from the techs when needed. However, they have limited time, a limited budget and they need their time off to be with their families. I think in most instances they are reasonably prompt with addressing problems and keeping us informed. The uptime of the projects here is probably the envy of many other projects. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If you have any ideas on how to retain crunchers or make them more involved, don't hesitate to speak up. ok ... if you want people to hang around don't waste their time and resorces 1) Fix the problem with cutting off CHP2 before the first save point avoiding unneeded time run errors that cause an unneeded error to the computers record that causes units to be validated..... I see it as better for my computers tracking record and better for WCG as a whole to just abort the WU if after 17hr + if it hadn't saved the first job ... really that's a very sad way fix that problem because no one else can seem to make an effort 2) When there is a problem noted like the computer has to be connected to send up trickles or it comes up invalid just stand up and say Huston We Have A Problem and tell us in the notices section of the program and/or in the event log....don't allow our time to be wasted and force us to research it here not all systems can be connected 24/7....we shouldn't have to research what known problems are WEE SHOULD BE TOLD!!! Well they do any of that?? I'm betting no As far as a computer getting to old to make an effort if it has the ability to run it even if it's slower then others ... please be real about managing resources and do consider just what your saying there ... other wise no harm done to me at least I can rebuild them and over clock if I want to I just don't want to Good thing I can edit this stuff or I'd be banned for sure [Edit 7 times, last edit by Former Member at Dec 14, 2015 7:33:29 PM] |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges:
|
ok ... if you want people to hang around don't waste their time and resorces I agree. You have made two suggestions. On suggestion #1, if it is possible to do just that, I would be the first to applaud. There may be reasons it is not possible so if one of the scientists or techs could offer an explanation, that would nice. I don't think this problem is widespread, but it does occur due to, as I understand it, either the computer is too slow for the molecule size or the calculations lead to a divergent condition. On suggestion #2, it would have been nice if this would have been caught in the Beta, but it wasn't. I agree that if intermittent internet connectivity is a problem for this project, it should be noted in the project requirements page. As far as a computer getting to old to make an effort if it has the ability to run it even if it's slower then others ... please be real about managing resources and do consider just what your saying there I retired some machines as being too old because the electricity use per unit crunched was too high. These were PII's, PIII's, P4's. They were repurposed to other uses if they were still operable. I obtained most of them for next to nothing, used them while they were useful and upgraded when I was able to obtain newer machines others were getting rid of. I know a person who deals in older lab and diagnostic equipment. Sometimes they need older machines to replace the computer components which have gone bad or only certain models work with certain machines. I have given some of these older machines to them in order to help them rehab the equipment and extend its useful life. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I just aborted a CEP2 WU that ran 17.78 hr with no save point and I looked at one that I spent 18 hours on and a different computer spent 18 on it also with only one hour into the 2nd job. So the problem isn't the computers running the WU's it's that unholy 18 hour limit and there is no good reason to pick up the 18 hour limit before the first job is done
----------------------------------------No harm I would of received nothing for it and my computer would of taken a hit As far as FAH2 goes I can keep it connected but of the 3 WU's that came up bad one I can't find I did get credit for the other two but they both got sent out twice.... so that's 3 times the CPU time for any one of them.....Dumb very dumb and both problems need to be corrected.... But I bet they never will [Edit 2 times, last edit by Former Member at Dec 16, 2015 2:36:46 PM] |
||
|
|
|