Message boards :
Number crunching :
BOINC needs a overhaul
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · Next
Author | Message |
---|---|
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
BUG REPORT Thank you Gary well put! Official Abuser of Boinc Buttons... And no good credit hound! |
Odan Send message Joined: 8 May 03 Posts: 91 Credit: 15,331,177 RAC: 0 |
Thank you Richard for your rational discussion of the matter. I think someone has been at the catnip :) |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
SNIP Sorry, you have not made your case. 1) Multi projects are encouraged. So if you cannot get work from one, you could get work from another. 2) There are real cases where a limit is required as explained. If you can come up with a better limit, propose it and how to calculate it.[/quote] I am so tired of people in this Seti forum trying to get me to run other projects. I signed on to Seti in the Beginning and that is what I want to run. I never liked the idea of Boinc and that was why I left for awhile. If I didn't run seti I would probably Fold. Official Abuser of Boinc Buttons... And no good credit hound! |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
Well, by my reckoning, out of the 22 or so people who have expressed their opinion about 15 would like to see an overhaul... 70% or so. Boy, How about all the times I have 2 units running on one core? How about the way Boinc gets confused when the cache is not really that big? Buy yourself a Cuda card or 2(Of course boinc isn't designed for) and watch how fast boinc gets totally confused. How many other problems have I posted in the past Ned? It is a whole lot more than the 24 hour backoff. Wonder why I have a need to push the buttons all the Darn time? If it were working like it should I wouldn't need the buttons. Personnally I saw this thread as a way for some users to let off steam after the recent crash but it seems the ones against Lukes idea are making it a bigger issue. Official Abuser of Boinc Buttons... And no good credit hound! |
DJStarfox Send message Joined: 23 May 01 Posts: 1066 Credit: 1,226,053 RAC: 2 |
I have a few comments about various posters' comments. Comparing it to a commercial business, isn't it good customer service to reply or answer your calls and emails so the customer knows they have been acknowledged? Comparing a nil-funded, open source project with a commercial business is an entirely unfair comparison. These are volunteers who contribute their time on the side of their real jobs. Disaster??? If you've studied queuing theory, then you'll know that there is some logic to having an exponential backoff when communication fails. Certainly, the logic needs a good analysis, but there are mathematical ways to calculate an appropriate value without BOINC waiting 23 hours. Did the BOINC developers do the math before programming those values? I don't know. [BOINC Developers] need to train themselves... [for] multi-resource, multi-core, multi-threading, multi-application, multi-everything projects... [with] a thorough code review (not re-write), concentrating on areas where the entity-scopes are wrong. Richard made a lot of good points, and I agree some quality control and a lot of beta testers would be very helpful for the BOINC core code. All of that is contingent on a code review and communicating a clear vision of what BOINC should and should not do. As a side note, I believe the future of distributed grid computing lies in virtualization technology. A good start is the CernVM application. Without a clear vision, communicated to the developers, testers, and the public volunteers, this project will continue to struggle with acceptance and technical limitations. |
Luke Send message Joined: 31 Dec 06 Posts: 2546 Credit: 817,560 RAC: 0 |
|
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Thank you Richard for your rational discussion of the matter. I think it was some bad cats on a modem pool in the Ukraine. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Thank you Richard for taking a civil approach on the discussion, even if our views don't completely align. Exactly, Luke. We should be able to 'argue' a bit without going down each others throats. I can be guilty at times myself. A pity we can not all be so well mannered and insightful as Mr. Haselgrove. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Thank you Richard for taking a civil approach on the discussion, even if our views don't completely align. I disagree. The first rule of an argument is to win! One of my favorite quotes. "The worst part in an argument is when you realize you are wrong." SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
RFP Send message Joined: 21 Jan 10 Posts: 44 Credit: 29,197 RAC: 0 |
Sigh. What ever happened to "anything worth doing is worth doing correctly"? True this not a commercial endever, but habits established on this type of project tends to follow ones into the commercial world. Is it any wonder why everyday devices such as cars which use computer controls go toes up. Again back to the fact that this is not a commercial effort so there should not be any TTM constraints or project managers looking over you shoulder going 'is it done yet?'. As so many have pointed out this is done on a shoestring budget and uses volunteer developers and testers so why not try for perfection? |
Bill Walker Send message Joined: 4 Sep 99 Posts: 3868 Credit: 2,697,267 RAC: 0 |
... so why not try for perfection? Can anyone prove they are not trying? As you said, shoestring budgets mean it could take a very long time to get there. Any time you have finite resources, in a commercial project or otherwise, you have to prioratize those resources. Things like getting the next grant and doing the science are probably more important that keeping us credit hounds busy. And before everybody jumps on me, I still haven't seen any proof that server outages, upload/download delays, credit issues, etc. effect the quality or the timeliness of the science. Sure, in general a delay in an upload might delay the science, but can you prove that is actually the limiting factor right now in SETI? Or in any other BOINC project? I'm much more concerned about the lack of new data from the Telescope. How will quicker uploads or downloads help that? |
Gary Charpentier Send message Joined: 25 Dec 00 Posts: 30651 Credit: 53,134,872 RAC: 32 |
Effective interfaces do not concern the user with the inner workings of the system. That goes for error reporting too! Specific things can get fixed, general ill ease is a lot harder to deal with. |
Gary Charpentier Send message Joined: 25 Dec 00 Posts: 30651 Credit: 53,134,872 RAC: 32 |
This is a BOINC part that needs a total re-design! BUG REPORT Scheduler Credit/Debit Situation: User attaches to LHC -- or any bursty project -- and sets his crunch percent high in the project because he wants to grab a work unit if one becomes available. Short term this does not cause a big issue, or not an issue if the project has work. Long term it is a big issue. Consider that two years down the pike ALL the other projects will be claiming they have over crunched two years of data. Even if the user detaches from LHC this two years of credit will still exist on the other projects. This can cause the scheduler to not fetch work units that would complete on time from a project with a short deadline because the crunch time is owed to someone else who no longer exists. A denial of crunch time. Possible Solution: Write the credit/debit routine to be a debt driven one, not a credit driven one. That way when a user detaches from a project the rest of the projects will reset to normal expected values. Better, use a double entry system of bookkeeping, where both the credit and debit are kept track of by each project as switching to a debt only driven system may have its own pitfalls. Accountants do double entry for a reason. Additional, don't accumulate debit/credit if the project is unable to supply any work units over a time frame. Perhaps if the project can't deliver a work unit in a weeks time then further work unit requests that are not fulfilled will not incur debit/credit until the project sends a work unit. This would prevent dead projects for running up infinite balances. There may also be need for a sanity check on the total amount of debit/credit allowed, perhaps a month's maximum. END BUG REPORT |
Pappa Send message Joined: 9 Jan 00 Posts: 2562 Credit: 12,301,681 RAC: 0 |
This is a BOINC part that needs a total re-design! Go signup for the Boinc Alpha mailing list and file it! You are perfectly capable of doing that, you should not require a "middleman!" Please bring the appropiate Debug Logs. Regards Edit: You do not even have to be an Boinc Alpha Tester, valid input is welcome. Please consider a Donation to the Seti Project. |
Gary Charpentier Send message Joined: 25 Dec 00 Posts: 30651 Credit: 53,134,872 RAC: 32 |
You are perfectly capable of doing that, you should not require a "middleman!" Try to understand this! It is the error, yes, a human factors one, which pervades BOINC and much open source software. |
Sirius B Send message Joined: 26 Dec 00 Posts: 24879 Credit: 3,081,182 RAC: 7 |
It's not just open source software. IMV, it's down to programming. Look at Microsoft, NVidia etc etc.... As a system builder, I'm getting customers coming back complaining that their display is out of whack...why? simple....Windows Automatic Updates with updated display drivers amongst other problems. On my own systems, I've had to return to 185.85 & 182.50 to keep my displays working as they should & keep NVidia updates unchecked. As for Boinc, I no longer keep on top of the development cycle as 6.10.24 crashed several systems. Find 6.10.19 ok. It would be nice for developers to return to structured programming. Also, even though I'm not a programmer, with technology moving at it's current pace, Boinc should be keeping pace as well. |
Lionel Send message Joined: 25 Mar 00 Posts: 680 Credit: 563,640,304 RAC: 597 |
I agree...and it needs to be from the ground up... |
The Gas Giant Send message Joined: 22 Nov 01 Posts: 1904 Credit: 2,646,654 RAC: 0 |
All up I think BOINC works very well. I don't always like how it handles work fetch and how it works out what to crunch and what it preempts. I also don't like how it can loose all the stats if your computer freezes at the wrong point in time - which happens all to regularly with 'doze. I also don't like the way GPU jobs are scheduled (FIFO is just all wrong and needs to be fixed). I'm no fan boy - but it has come a very long way since it's inception and is many times better than the very early versions. And OMG look at the number of projects using it. It looks like it's doing exactly what it was designed to do. I also wish reporting bugs / interacting with the devs was easier. YMMV |
The Gas Giant Send message Joined: 22 Nov 01 Posts: 1904 Credit: 2,646,654 RAC: 0 |
Speaking of not liking the work schedular, see this link PrimeGrid has gotten into deadline issues due to some other weired BOINC issue (I'll try and catch that one next) so now BOINC is going to crunch every wu it thinks is in deadline trouble until the sum of the estimated time to completion is OK again. This means that each wu will remain in RAM (removing them is not the answer) and cause my machine to start paging to disk. UGH! |
Kibble (KB7TIB) Send message Joined: 6 Dec 99 Posts: 27 Credit: 10,121,469 RAC: 2 |
@ WinterKnight:
Please try this point out at the LHC@home fora. LOL! Regards |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.