Message boards :
Number crunching :
Average turnaround time
Message board moderation
Author | Message |
---|---|
Send message Joined: 29 Sep 04 Posts: 196 Credit: 207,040 RAC: 0 |
Since the upgrade to Boinc Server v5, Average Turnaround Time is still not being reported accurately. Back on 14 April 2005 I reported a problem with it on all of the computers on my account which kept the ATT at 13.89 days no matter the actual value. It seems to have gotten worse. Presently one of my machines is reporting an ATT of 236378.64 days. Uhm, that's 647.6 years per workunit. :) Any idea when/if that will be tackled? I realize it's a low priority issue. Hopefully the scheduler doesn't take it into consideration when sending out WUs.. Travis |
Send message Joined: 15 Jul 05 Posts: 17 Credit: 16,521 RAC: 0 |
I've been getting work constantly on PrimeGrid since no work is available here. It shouldn't have an effect on the server sending work since BOINC Manager regulates the requests for work. As long as the server has work and you haven't exceeded your quota, you should get work. |
Send message Joined: 29 Sep 04 Posts: 196 Credit: 207,040 RAC: 0 |
I've been getting work constantly on PrimeGrid since no work is available here. It shouldn't have an effect on the server sending work since BOINC Manager regulates the requests for work. As long as the server has work and you haven't exceeded your quota, you should get work. That's what I hope - however statistically speaking your computer "67437" has a 475330.18 day avg turnaround time with 13 returned results while your other machine "67449" has a 0 day avg turnaround time with 2 returned results. Stastically speaking it's impossible given this project's age and the date your PCs joined LHC. Granted, my PCs receive WUs just like any other, it would still be nice if the feature either 1) worked properly 2) was turned off. On PrimeGrid, both your computers are responding with an ATT of 0.12 & 0.14 days, respectively. Their statistics work. LHC's hasn't in months. My 2 ยข PS: You need to merge one of your hosts on PrimeGrid :) |
Send message Joined: 13 Jul 05 Posts: 51 Credit: 10,626 RAC: 0 |
LOLLLL!!!! I just checked my own Average turnaround time, and it's 324068.08 days! (887,86 years!!!) But I still have a Maximum daily WU quota per CPU on 100/day, maybe it's broken?! ;-) "I'm trying to maintain a shred of dignity in this world" - Me |
Send message Joined: 13 Jul 05 Posts: 143 Credit: 263,300 RAC: 0 |
887,86 years ??!!?? Fuzzy - are you related to Mathusilia by any chance??? If I've lived this long, I've gotta be that old |
Send message Joined: 13 Jul 05 Posts: 51 Credit: 10,626 RAC: 0 |
887,86 years ??!!?? Aren't we all? I think he was a close relative to Adam and Eve, so we all must be descendants, and with that relatives to him! ;-) It's just a kind of funny, as I should be one of the WU hoarders with my turnaround time on about 888 years, and a download of 100 WU's per day! What a cache that would be! But I don't have any LHC WU's right now and returned my last one several days ago! ;-) EDIT: And BTW, that should be 887.86 years! Where I live, we use the comma as the decimal point, so I copy - pasted from the calculator! So "only" about 888 years, not 887,860 years! "I'm trying to maintain a shred of dignity in this world" - Me |
Send message Joined: 15 Jul 05 Posts: 17 Credit: 16,521 RAC: 0 |
Nope. I had BOINC on my laptop, ran PrimeGrid, noticed I was claiming 13+ cobblestones and only receiving 8.00. PrimeGrid apparently has a max credit/result setting. I do this for the science, but I'm also a numbers freak. If I'm not going to get credit for 40-45% of my work, why should I do the work? |
Send message Joined: 2 Sep 04 Posts: 17 Credit: 22,605 RAC: 0 |
Nope. I had BOINC on my laptop, ran PrimeGrid, noticed I was claiming 13+ cobblestones and only receiving 8.00. PrimeGrid apparently has a max credit/result setting. I do this for the science, but I'm also a numbers freak. If I'm not going to get credit for 40-45% of my work, why should I do the work? We have a up-limit for claimed credits at PG to prevent cheating, as we are only sending one result per workunit. The average claimed credit is approx 5.3, so the upper limit was chosen to allow some fluctuation from the average and not to offend anyone. Seems like that didn't work :) As for LHC avg turnaround, it might be that recent host table corruption did something with that, like inserted incorrect values and we will be doomed to see incorrect values forever unless we re-attach our hosts (to create new id) or avg_turnaround value is reset to 0 globally. PrimeGrid Administrator |
Send message Joined: 15 Jul 05 Posts: 17 Credit: 16,521 RAC: 0 |
Nope. I had BOINC on my laptop, ran PrimeGrid, noticed I was claiming 13+ cobblestones and only receiving 8.00. PrimeGrid apparently has a max credit/result setting. I do this for the science, but I'm also a numbers freak. If I'm not going to get credit for 40-45% of my work, why should I do the work? I understand the upper limit. The biggest reason I stopped doing any BOINC on the laptop was because I could not guarantee it would be used continuously, and it seemed way off the charts for it to take over an hour to run a single PG result when my other systems take less than 30 minutes. One claims 3+ cobblestones, and the other claims 5+, so the laptop claiming 13+ just didn't seem right. *shrug* |
Send message Joined: 29 Sep 04 Posts: 196 Credit: 207,040 RAC: 0 |
As for LHC avg turnaround, it might be that recent host table corruption did something with that, like inserted incorrect values and we will be doomed to see incorrect values forever unless we re-attach our hosts (to create new id) or avg_turnaround value is reset to 0 globally. Not true. Problem existed before v5 server upgrade & prior to 14 April 2005. See 1st post & link. |
©2024 CERN