Message boards :
Number crunching :
Credit limits apparently exist ...
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
Send message Joined: 1 Sep 08 Posts: 520 Credit: 302,525,188 RAC: 1 ![]() ![]() ![]() |
Is the SETI benchmark based on the SETI *optimized* application (which offers more credit per CPU cycle then the 'regular' application)? Further, will the changes mean that the disadvantage for AMD processors (which SETI and Einstein among other projects have) gets *propagated* over here? The SETI applications are written in a way which (unintentionally I believe) results in AMD processors of any type to have a lower rate than an Intel processor. I would hope that the MilkyWay folks succeed in NOT making that disadvantage show up here. Seems that Travis has decided to kneel at the master's feet "DR" Anderson - and is reducing the credits. ![]() |
Send message Joined: 1 Sep 08 Posts: 520 Credit: 302,525,188 RAC: 1 ![]() ![]() ![]() |
|
![]() Send message Joined: 20 Sep 08 Posts: 1391 Credit: 203,563,566 RAC: 0 ![]() ![]() |
As a newcomer over here, but a cruncher for other projects as well, it does seem to me that David Anderson and Travis are getting some unfair stick here. The whole concept of Boinc is to use distributed computing, to allow the general public to contibute their computer time to worthwhile scientific and humanitarian projects. I think the vast majority of people do choose projects on that basis. But people being people, there will always be those that are in it for the credit, and will choose the highest paying projects because they like the competition and league tables element of it all. I personally can see no reason why all projects in Boinc shouldn't give the same credit for the same amount of work on the same computer. As for the computers themselves, clearly a quad or twin core will produce considerably more work in a given time than a P4. Therefore it seems only fair to award the faster computers more credit for their greater input to the project. It does seem unfair that AMD kit seems to lose out here, but as I understand it the Intel design is more efficient at floating point calculations than the AMD one. Look at it this way, If I bought a Ford family car for £10,000 which was speed-limited to 70mph I might accept it. However a chap buying a Ferrari for £100,000 capable of 200mph wouldn't. He'd probably move to Italy to drive on the Autostrada! But there aren't that many Ferrari owners compared to Ford. Just my 4C worth. Don't drink water, that's the stuff that rusts pipes |
Send message Joined: 4 Oct 08 Posts: 1734 Credit: 64,228,409 RAC: 0 ![]() ![]() |
Well reasoned Chris, and true on most points. People can choose based on their motivation and interest in the science and worth of different projects. Dr Anderson, as a broad principle of the BOINC platform is aiming to keep approximate comparable credit granted between different projects. Nothing can be exact, so a comparable position is being asked for, based on a mythical average computer and stock applications. Where the principle may get ragged is when optimised (or faster) applications are included. But as long as these do not become the recommended stock project client then only minor credit adjustments need be made. |
![]() ![]() Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,884,648 RAC: 0 ![]() ![]() |
There only appears to be fear mongering regarding credit reduction being spouted by a very vocal few. I think I complained about the low credits at LHC@home quite a while ago, but when the big dry occured I just stopped crunching and never went back.....but then again so many people think that it is good science that me not being there was irrelevant. |
![]() ![]() Send message Joined: 10 Nov 07 Posts: 42 Credit: 27,012,695 RAC: 0 ![]() ![]() |
I don't really understand what you people who are critizising DA are up to. Those who are crunching only for science really shouldn't care about credits at all. Those who only want to compare themselves with others in the same project shouldn't complain about the comparism to other projects. So it should be totally the same if a computer cycle is worth 1000 credits or only 1 - inside one project. The actual amount of credits can only be of importance to those who compare boinc-wide or team-wide. And for them it should be the aim to have comparable credits in all projects. And here some difficulties do actually start. A project like MW, in the moment giving about 4 to 10 times the credits of the other high crediting projects (4 times for modern quads and 10 times for P4 or P3 compared to PG, which gives really high credits) certainly utterly destroys any boinc-wide comparism. And this even after the credit reduction of today. (Not sure if there was one at all, since all my machines still work over the credit limit). I do not think that seti, with its high difference between optimized and unoptimized applications, should be the measure. Or its not so easy to take it therefore. The only advantage of seti being the measure is, that it is the project with the most crunchers. But to keep the fun of racing, comparing stats between teams and members, there has to be a limit and a range to crediting. So this, I assume, is the motivation for DA. But what is the motivation of those arguing against it? To destroy the keeping of stats? I don't think so. To horde high numbers? I don't think so either. For the majority of people numbers only have meaning in comparism to other numbers. To have high credits whereever they are crunching and therewith having an advantage over those who like other projects better? Quite probable. To further their teams forthcoming and position in the Boinc-wide stats in comparism to very strong teams bound to special projects (like the einstein-grid)? also quite probable. To have opened a battlefield with DA, which has originated somewhere else? possible, but I don't know enough of the crediting-battles of the past, to judge here. But than you would be abusing MW for personal reasons that really do not belong here. So my conclusion is that those crying for the independence of project crediting are up to the contrary of what they are claiming. They want the high credits to advance compared to others boinc-wide. I would be glad to have someone logically explain to me another reason and so restore the credibility of those argumenting this way. I really don't find one. Instead of logically arguments I've more and more often seen attempts to ridicule persons who do care for interproject comparability or even the use of strongly insulting expressions. I am really sorry for this development and would gladly see everyone relax a little but to keep a little interest in fair crunching conditions for everyone. Dragons can fly because they don't fit into pirate ships! |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 ![]() ![]() |
My position is: If the implied equality among projects is removed from BOINC's sphere of responsibility / influence, projects could grant whatever they wish and not have to deal with administrative overhead in keeping themselves "in line" per a poor definition of a "standard unit" (the "Cobblestone"). Place the calculation of the "Exchange Rate" with an independent arbitrator, someone who cannot be accused of bias. Until then, the only comparisions that are valid are comparisons within a project, not between projects. If I have 50,000,000 "credit" here, it does not necessarily mean that it is "more than" someone with 1,000 credit at SETI. |
![]() ![]() Send message Joined: 4 Dec 07 Posts: 45 Credit: 1,257,904 RAC: 0 ![]() ![]() |
Why bow down to SETI@HOME on the credit issue. If they (and any other project were worried about crunchers, then why don't they raise their credit system to the level of Milkyway@home. I liked the credit Milkyway was giving, but that didn't stop me from working on other projects. It was like a pick-me-up since I don't leave my computers on 24/7. Now I will have to rely on ABC for it. At least they have not bowed yet. |
![]() ![]() Send message Joined: 10 Nov 07 Posts: 42 Credit: 27,012,695 RAC: 0 ![]() ![]() |
My position is: If there were people who did that, it would be a huge improvement. Although I think a minimum of comparability is already existent. Its not good or even perfect but its good enough for teams to celebrate boinc-wide milestones or placements. But your idea is definetely better only those who are accepted an willing have to be found. But till then I think MW should come down from there 5 times higher crediting than any other project. Although I enjoy every credit the really great opt. application floats into my lair, I think MW should reduce the credits immediately and drastically. Dragons can fly because they don't fit into pirate ships! |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 ![]() ![]() |
Until everyone has the same performing application, nothing should be done. If you do, you will majorly irritate those who were minding their own business. They will see a massive reduction for "no reason". Additionally, part of the problem with raising and lowering credits is that it typically destroys comparisons within a project for any hosts that did not run in similar credit eras. To properly keep the ratio before and after the new application, my credit rate would need to go back down to 50/hr rather than the 215-216 it is at now. That might still be more than other projects, but at least it is the same as what it used to be, which means that there is consistency within the project, even if it is higher than someone's desired "standard" amount. Otherwise, if it is reduced to 15-20/hr, like SETI grants now, then someone with a system identical to mine will have to do around 3X more work to just catch up with me, much less overtake me. |
Send message Joined: 10 Nov 07 Posts: 96 Credit: 29,931,027 RAC: 0 ![]() ![]() |
Why bow down to SETI@HOME on the credit issue. If they (and any other project were worried about crunchers, then why don't they raise their credit system to the level of Milkyway@home. AFAICS it’s not about deferring to S@h, but of preserving the ratio of credit granted to Flops performed. This may be a pipe dream, but the idea is that one should be able to calculate the computational throughput of a project from the amount of credit it grants—indeed, many project and stats websites report their performance in TFlop/s, based solely on the credit figures AFAICT. Personally, I wouldn’t object to a ‘floating cobblestone’, leaving it to the stats sites to work out ‘exchange rates’ for cross-project comparisons, but I can also see the advantages of standardization from an administrative POV, theoretically allowing each project to derive a reasonable estimate of the available computing resources directly from its credit stats. ![]() |
![]() Send message Joined: 12 Nov 07 Posts: 20 Credit: 1,146,706 RAC: 0 ![]() ![]() |
SETI has lost and continues to lose crunchers due to their constant reduction of credits. Other projects are starting to feel the sting as they are pushed into lowering theirs as well by the "powers that be". It is the intra project comparisons that are suffering the most from this. Even those of us who are in it for the competition continue to crunch many projects regardless through "team projects of the month" and or races and other competitions. We just prefer to give the lions share to those projects that reward us the most. Many people like to compare the credit situation to the economics of work, so here is my comparison: I think pay in all jobs should be equal. Since the US government has set a standard with the "minimum wage" then all employers should lower their wages to the federal minimum wage. Now how well would that go over?? |
Send message Joined: 1 Sep 08 Posts: 520 Credit: 302,525,188 RAC: 1 ![]() ![]() ![]() |
SETI is probably also losing crunchers due to the problems they have keeping their servers running regularly. Whether it is the daily 5 hour outage every Tuesday (with about a 5 to 10 hour recovery period afterward due to backlog), or the general 'touchiness' of their servers (which frequently need to be 'kicked' for one problem or another), or the Astropulse mess, or the relatively low credits, or the optimized for Intel CPU's -- all those things add up to migration away from SETI. I have joined 10 projects over time, two are defunct (Climate BBC and that bad boy Predictor). This week MilkyWay moved to the top of my credit list. Here's hoping Milkyway continues to do well (though today it does appear we've got a sick server). SETI has lost and continues to lose crunchers due to their constant reduction of credits. Other projects are starting to feel the sting as they are pushed into lowering theirs as well by the "powers that be". It is the intra project comparisons that are suffering the most from this. Even those of us who are in it for the competition continue to crunch many projects regardless through "team projects of the month" and or races and other competitions. We just prefer to give the lions share to those projects that reward us the most. ![]() |
Send message Joined: 12 May 08 Posts: 30 Credit: 1,589,348 RAC: 0 ![]() ![]() |
I think pay in all jobs should be equal. Since the US government has set a standard with the "minimum wage" then all employers should lower their wages to the federal minimum wage. for me? got my house, my porsche for summer, my subaru for winter, my LC8 for fun, don't pay no rent, don't need to work anymore - should i care? but retranslated for the project - collecting all those boat-anchors to chrunch which get credited like fast Xeons is a very humble approach - respect! let's dig out the ancient stuff and waste more energy... |
Send message Joined: 30 Oct 08 Posts: 32 Credit: 60,528 RAC: 0 ![]() ![]() |
Well, I don't pay too much attention to credits normally - I join projects for the science, but I do enjoy a good fat bunch of cobblestones when I get it. But this seems strange enough for me to want to give my opionion about it. To make it clear, I am very much for giving out fair credit between projects. We had the same "problem" over at Einstein about a year ago, that we were getting too much credit (though probably less than what Milky Way was giving crunchers until the fix). Even then I was in favor of lowering credits to get closer to what SETI was giving and my opionion has not changed. Let credits be something that can be compared, and try to give all projects a fair chance of attracting volunteers. Still, I'm not happy with the current situation. Give us less credits per WU, so we get the same as crunchers at SETI or Einstein - fine with me. But this "cap" is, in my eyes, a bad idea to say the very least. To me that is punishing people with fast boxes, and let's face it - those are the ones that get most work done for science. With the credit scheme we have now, my cute little Core Duo laptop would probably make the same amount of credit as my brand-new quadcore desktop - although it only has rougly 1/4 the power. Hell, I could attach my old Acer with the Celeron CPU and would probably get way more credit, compared to faster crunchers, than the speed of that box suggests. What's next? If we are still getting too much credit, would you lower the cap so far that the faster boxes reach it in the time it takes to go for the first cup of coffee in the morning after switching on the computer? I might be exaggerating but you get my point: People who only go after credits would consider much of their work "wasted" (and probably decrease res share, which would be bad for science) and even for those crunchers who, like me, see the credits as merely a nice little extra, it is kinda demotivating. And there are faster comps out there than mine, much as I hate to admit it ;-) take Xeons alone. That said, I want to make clear that I have no intention of leaving this project any time soon and I want to express my solidarity to the project staff, who seem to have done a very good job so far. My comment on crediting is meant as constructive criticism and I hope it comes across as such. Greetings, Annika |
![]() Send message Joined: 12 Nov 07 Posts: 20 Credit: 1,146,706 RAC: 0 ![]() ![]() |
Well, I don't pay too much attention to credits normally - I join projects for the science, but I do enjoy a good fat bunch of cobblestones when I get it. But this seems strange enough for me to want to give my opionion about it. If the science is worth while, then the project will have very little problem getting crunchers anyway. The kredit kops are always saying that those of us who crunch for competition are in the minority, so why is there so much concern for credit parity? If we are the minority, then it should not matter to the projects if we credit shop because the majority of crunchers are doing it for the science and care less about the credits. This whole parity thing still does not make sense to me. Let each project set their own credit level as they want it and let the "minority" shop for credits if they want to. |
![]() ![]() Send message Joined: 4 Dec 07 Posts: 45 Credit: 1,257,904 RAC: 0 ![]() ![]() |
Crunch3r Right on the money...... I don't have a massive number of computers like some users do nor do I have ones that are really that fast, nor do I keep them going 24/7, so I value every credit that I can get. The more the better. Oh yeah, I still do SETI along with many other projects (56 in total when they are up and running). |
![]() ![]() Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,884,648 RAC: 0 ![]() ![]() |
Crunch3r, I don't think people have said that those who crunch for competition are in the minority - ever. It is just a founding concept of BOINC that each project gives about the same amount of credit for the same cpu. It's as easy as that. It is also a plain fact that if a project gives unreasonably high credit, people will flock to it - no matter what the science is about. |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 ![]() ![]() |
The issue I have with that is it assumes that all of us are dirty, rotten, mangy, credit-hungry people, thus we are only interested in the credits. If that were the case, why did I not attach to RieselSieve? Why is my Pentium 4 system still just slowly churning through Einstein@Home tasks that have a payout of only 9-10 cr/hr? Following your logic, I should be so rabid that when I saw things going on here I immediately attached it over here and tried to "rack 'em up"... Why do I participate in LHC which is equally as low? Why am I still keeping an eye on Orbit and will give resources to them once they get to Beta stages? The projects I participate in are Physics or Astronomy related. I don't do the life sciences stuff, nor do I do the Math problem projects (RCN, Riesel, 3X+1, etc...). It was not my fault that MW or Cosmology happened to have higher grant ratios, and bear in mind I am still participating in LHC and Einstein, both of which are either at or below the "SETI Benchmark" rate. Additionally, what constitutes "flocking"? From what I've seen the people who chide the "high paying" projects typically consider a 1% increase in participation to be "a flocking". |
Send message Joined: 3 Apr 08 Posts: 7 Credit: 638,859 RAC: 0 ![]() ![]() |
It has been fun, while it lasted! I for one can no longer justify the electricity consumption to have my computer running 24-7. My credits went from 1200+ to 400- 500 a day. Good Luck! |
©2024 Astroinformatics Group