Message boards :
Number crunching :
A Resolution to the Credit Disagreement
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
Lord_Vader Send message Joined: 7 May 05 Posts: 217 Credit: 10,386,105 RAC: 12 |
How is the "deadline" for a WU calculated? Fear will keep the local systems in line. Fear of this battle station. - Grand Moff Tarkin |
SargeD@SETI.USA Send message Joined: 24 Nov 02 Posts: 957 Credit: 3,848,754 RAC: 0 |
Now you are confusing me. I always got the same credits as everyone else in the quorum and most of the time it was less than I claimed. So where was I getting a bonus? From the fact that my computer could do more science in an hour than some others?? That is as it should be. Now, with the enhanced WUs I can still do more science in an hour but I do not get an equal amount of credits for that science over the amount of time it takes to do it. That is a major change and should have called for a reset to "truly" make the playing field level. |
Saenger Send message Joined: 3 Apr 99 Posts: 2452 Credit: 33,281 RAC: 0 |
Now you are confusing me. I always got the same credits as everyone else in the quorum and most of the time it was less than I claimed. So where was I getting a bonus? From the fact that my computer could do more science in an hour than some others?? That is as it should be. Now, with the enhanced WUs I can still do more science in an hour but I do not get an equal amount of credits for that science over the amount of time it takes to do it. That is a major change and should have called for a reset to "truly" make the playing field level. Why reset when nothing should be changed? The BOINC approach was: Count the work done on the puter regardles of what kind of science is done. As a viable flop-count was not available at first, the (interim?) solution with the benchmark was created. It's science independent, it works (in theory) on every puter, and the faster ones will claim more, while the slow ones will claim less. A puter should claim the same per hour for every different project and WU. If the WUs vary in length, no prob, time will be taken into consideration. The litmus test for a fair credit system would be: Is it project independent and will it be scalable for much faster computers? So the new approach is direct count of the flops. That's fine! Just make the conversion factor to match the once set norm. |
Eric Korpela Send message Joined: 3 Apr 99 Posts: 1382 Credit: 54,506,847 RAC: 60 |
How is the "deadline" for a WU calculated? The deadline is the estimated crunch time on a 50 MFLOP machine running 100% SETI@home. @SETIEric@qoto.org (Mastodon) |
Lord_Vader Send message Joined: 7 May 05 Posts: 217 Credit: 10,386,105 RAC: 12 |
How is the "deadline" for a WU calculated? So how about we convert the deadline (as a number of days)into a predetermined credit (some multiple of the days) and just reward it when the WU is returned? Then the faster machines will get more credits and everyone will get exactly the same credit for the same WU. Some people will just return them faster. x hours * fixed multiple = credit granted Is that too simple? Fear will keep the local systems in line. Fear of this battle station. - Grand Moff Tarkin |
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
How is the "deadline" for a WU calculated? Because the original estimate of time to be used is not always accurate. For example, if the WU is very noisy, it will stop processing in a few minutes at most. These do not deserve full credit. The run time is also dependent upon how many spikes and triplets are found. Having the server pre-determine the credit is IMHO a step backwards. BOINC WIKI |
Lord_Vader Send message Joined: 7 May 05 Posts: 217 Credit: 10,386,105 RAC: 12 |
How is the "deadline" for a WU calculated? I would only give credit for sucessfully reurned units. Also, the advantage to presetting the credits is that there is no argument or confusion. Everyone is under the same rules. It also allows the optimized apps to focus on the science without worrying about skewing the credit determination. Just thinking out loud.... Fear will keep the local systems in line. Fear of this battle station. - Grand Moff Tarkin |
Beach Bum Send message Joined: 13 Nov 05 Posts: 178 Credit: 611,717 RAC: 0 |
I think Sarge has a point on the credit base. With the serious differance between SETI@Home and Enhanced. I also believe the Credits should have been reset like they were for classic. Or like CPDN does, it becomes a totally different project that has seperate stats. I believe the current credit system is close to fair, other than maybe taking the multipler to maybe 4 instead of 3.35. With a reset of everybodies credit. The competetion would return to see who could be top dog. As it is now the top teams just got a 1 or 2 year stay at the top due to the amount of time it will take to close the gap now under the new credit system. I really think this is where a lot of the cries about the serious RAC drops are coming from. With the multipler at 4 or close to 4, credit would work out more online with other projects, or maybe just a small bit higher. So I say this would most likely be the best fix. 1. Bump the multiplier up to close to or to 4, to level the credit more towards the other projects 2. Reset everybodies score, teams and individuals. 3. pull a back credit system like you did for classic. 4. Get the bugs out of the APP and Splitters. As I see it, its a new project/way of looking for signals. Just as @home was to Classic. So it should have been treated in the same manner as it was done for classic. Reset, give the @home crowd a final spec they could post on their stats. Start from 0 again, and let the competetions begin, those slamming the science which is what SETI is really looking for. I don't think any of the teams or most of the members would have an issue with that as a solution. It truely levels the field for the power users, and the regular users. We all start at 0 and may the best system, tuned the best , get you the extra points you are looking for. Just my humble 2 cents to stop the dang argueing. Beach Bums Current Stats: Come Join us at Hawaiian Beach Bums |
RDC Send message Joined: 17 May 99 Posts: 544 Credit: 1,215,728 RAC: 0 |
The problem I see is that your then punishing the older teams for daring to create themselves early on by saying "it's not fair that your so far ahead and our new team can't catch up so we want you dragged down so we can pass you". Basically that is what the zeroing out of credit and starting fresh for everyone would be. Doesn't matter one way or the other to me personally but by doing such a thing, you then set precedent. A year or two down the road when new teams develop and start wanting a level playing field as well so they can be top dog, will S@H have to reset things again due to cries of unfair treatment and strike threats? To truly explore, one must keep an open mind... |
Beach Bum Send message Joined: 13 Nov 05 Posts: 178 Credit: 611,717 RAC: 0 |
Its not about the teams. It comes right down to the differance between Enhanced and @Home. It should have been handled as Classic was. Its as much a differance. It should be leveled. Heck I stand to lose more credit than a lot of the older crunchers with a couple machines here. The way you are looking at it, then the classic folks should have had their credits added to what they have now. They got penelized for sticking with it for years. It is no differant. New Project Design, New Credit System. Reset the credits, now there is no beef about anybody haveing more credits due to the old system of doing the credits. Its all a level field then. But lets look at it through your eyes, If my team is say 1 million points behind yours, and under the old system my team is putting out say 5000 Rac to your 4000 Rac a day. I know it will take us X amount of days to catch you. Now the project decides to change the way they do the credits, We both fall by say 1/2. You have just gained a lead of double the amount of time to catch you. I did nothing to have this extra time thrown on me, but hey you get to set there for a much longer time because of a rule change, not because you did something to make me slow my catching up to you, like produce more RAC. That is where zeroing the credits is needed. the playing ground was radically changed, the competetion was just fixed to the top runners. Beach Bums Current Stats: Come Join us at Hawaiian Beach Bums |
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
Yes, the method of calculating the credits is slightly different, but they are supposed to match (stock setup to stock setup). If they don't match reasonably, then that is a problem. Please note that classic counted 1 for each result reported. Non-enhanced reported X credits per hour with the stock setup on a particular system. With the same setup, enhanced is supposed to report the same number of credits per hour. A machine that is half as fast is supposed to earn half the credits per hour, and one that is twice as fast is supposed to earn twice the credits per hour. This is the same for both non-enhanced and enhanced. I really don't see why this is such a major change that credits need to be reset. If it is not quite right, I am certain that the credits earned per hour will be adjusted. BOINC WIKI |
Saenger Send message Joined: 3 Apr 99 Posts: 2452 Credit: 33,281 RAC: 0 |
If they don't match, something is (or was) seriously flawed. The credit calculation was the same for all projects, and afaik it never changed much for the stock client/application setup. Only the bonuses for the optimized clients and applications are reduced. That's something that's supposed to have happened, and it's fine and fair to cut bonuses if no extra is delivered any more. The stock setup is the measurement. That has to be fair across all projects. Extra credits for optimizers are only valid if they really do extra work now, not have done so in the past. And the project is in no way responsible for garanting these bonuses at all. The good thing with the "old" credit system is, that it's (in theory) independent of WU-size and underlying science. It's flawed concerning the benchmark calculation, as this is imho far to easy to manipulate to increase your claims withouit doing anything extra, but that's another discussion. To go even one step further (only for argumentational purpose, not really claiming this): You could claim that Seti has got far to much credits in the past, as it was poorly programmed, so that an ordinary application got points for useless calculations, only the optimized would have had the right to the credits calculated the ordinary way. The bonuses through inflated benchmarks should be scraped, and same goes for a certain percentage of the standard app, so that every seti-user will lose, say, 50% of all credits granted since the first optimized clients arrived on scene. But as I said, this will not happen, and I don't want it to happen, and imho no other project wants this to happen. But the argument is valid nevertheless. Gruesse vom Saenger For questions about Boinc look in the BOINC-Wiki |
W-K 666 Send message Joined: 18 May 99 Posts: 19064 Credit: 40,757,560 RAC: 67 |
I sense that most of the people who think the fpops credit system is flawed, did not join the beta test project. Or if they did, because they are mainly credit chasers, didn't do any real work there. There were 6000+ who joined, but a lot did no, or very little work because the credits were not exported to the BOINC totals. Even when the totals were eventually exported, Beta had not been optimised and credits were like 3/hr on a computer that could get 15+/hr on all other projects. Eventually the beta app was optimised but with so few (in the 100's) computers actually participating the credit calculation was formulated from a small database. So if it is not correct, who's fault is it. In the main it is probably close to being correct, at the extreme's of the angle_ranges it is probably less accurate. But it can be adjusted when the developers have some time. So next time, for Astropulse, join, do some crunching and it is possible the credit calculation will meet your expectations. Don't just sit there moaning because you got a few units that don't give what you regard as the correct cr/hr. I've not heard anybody complain yet where the calculation gives more than the average. |
Jim-R. Send message Joined: 7 Feb 06 Posts: 1494 Credit: 194,148 RAC: 0 |
Neither have I, and I have seen many wu's where this has happened. That would be something to see! "Hey, you guys, my credits aren't fair!!! You gave me too much credit on this last work unit!!!" Haha! /edit/ BTW, Winterknight, I totally agree with the rest of your post also! Jim Some people plan their life out and look back at the wealth they've had. Others live life day by day and look back at the wealth of experiences and enjoyment they've had. |
Chilean Send message Joined: 6 Apr 03 Posts: 498 Credit: 3,200,504 RAC: 0 |
|
JV Send message Joined: 16 Nov 05 Posts: 17 Credit: 91,295 RAC: 0 |
Welp, if they ever want someone to crunch just to collect data for figuring out a credit scheme I'll do it. I honestly don't care about the credit and they can zero my credit every day. Fine with me. I'll be willing to crunch 24/7 for their statistics even if the results don't carry over to count for credit. I'd go back to crunching the actual app when they were done. As long as I can walk by my computer, wiggle the mouse and see pretty colors dancing on the screen and numbers scrolling by real fast, I'm satisfied. Simple things for simple minds. I only have one computer on SETI and my RAC goes up and down like a rabbit hopping across a field. Just another pretty squiggly line to me. The only part of this entire evolution that I understand is looking for radio signals. Outside of that, it is all FM to me. (FM = Frickin Magic) Joe |
David Stites Send message Joined: 22 Jul 99 Posts: 286 Credit: 10,113,361 RAC: 0 |
{tone: friendly} If anybody wants my credits you can have them. I don't mind starting over. As long as the numbers are going up I know my computers are actually working and not screwing off when my back is turned. If the numbers are what you care about take mine. What I really want is to have a postive result, proof that ET exists. (Actually, I want a working stargate but I don't even dream about that) Keep crunching, David Stites Pullman, WA USA |
paul Send message Joined: 29 Jul 01 Posts: 42 Credit: 23,126,185 RAC: 0 |
With me not really trying to understand the method behind the old and new credit system, I still don't get why people are upset with the new credit system. Didn't we know back then that we were actually getting inflated credits with 4.1.X? I had a 21K RAC back then, and have dropped to 9K. I also had the fastest Intel dual core in the world running Seti for a while. I adopted Enhanced as soon as it came out, even though I lost the crown immediately. I say this, because I am one of those who build large fleets with my own money, and tweak them as fast as I can. Given the power users arguments, I should be on their side I see absolutely no problem with the new credit system, since everyone else's RAC has dropped just as much. It did bother me a bit at first, until I wondered about the litmus test.....How does Enhanced RAC compare to other projects? Will two equally configured rigs report the same RAC, with each running a different project? I setup my D @4Ghz running Seti Enhanced, and the wife's D @4Ghz running Rosetta. Same fast ddr2 memory, same timings and speed. Gave them both several weeks to equalize a bit, the Seti rig kept dropping, but has leveled off, the Rosetta rig kept rising, has leveled off. Results? My Seti D is claiming 1545 RAC, wife's Rosetta D is claiming 1612. This is how it should be, in my book. The playing field has been leveled. Paul Team Starfire World BOINC IRC- irc//irc.teamstarfire.net:6667/team_starfire |
W-K 666 Send message Joined: 18 May 99 Posts: 19064 Credit: 40,757,560 RAC: 67 |
With me not really trying to understand the method behind the old and new credit system, I still don't get why people are upset with the new credit system. Didn't we know back then that we were actually getting inflated credits with 4.1.X? Thanks for your support for the levelling of the playing field. It would help Eric K. if you could post some your credit/time info in the Cross Project Credit Equalization and Adjustment Andy |
Sandtiger@seti.usa Send message Joined: 24 Feb 06 Posts: 37 Credit: 560,736 RAC: 0 |
I see absolutely no problem with the new credit system, since everyone else's RAC has dropped just as much. It did bother me a bit at first, until I wondered about the litmus test.....How does Enhanced RAC compare to other projects? Will two equally configured rigs report the same RAC, with each running a different project? Does BBR know you are posting this? As a memeber of BBR I would think you would be at the Spearhead of the arguement for increasing it seeing how you guys pretty much had a lock down on the #1 spot you guys were out pretty much staying even with SETI.USA and we were all beating S.G by at least 100K a day. You were poised to pass S.G in 800 days and then we were going to pass you in something shy of 1300 days. Now as it is you are getting completly dominated by S.G and SETI.USA Im by no means trying to flame but I would think you guys would want to have the credit fixed a little bit in your favor. I'm not in it for the science...only the competition. |
©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.