Message boards :
Number crunching :
Seti Enhanced Credit Fair?
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 23 · Next
Author | Message |
---|---|
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
Fairness, like beauty, is in the eyes of the beholder. It will be fair to some and not to others. The fear I have is that there may be a mass exodus of people away from Seti. Many people crunch solely to help the science (and some claim to do so), but many more crunch for the competition and the bragging rights and they will be the ones who will leave when everybody starts getting the same credit regardless of the time spent and CPU used. The current enhanced credit system will remove the incentive to update to faster machines which coupled with people leaving the project is ultimately hurting the project. IMHO, the project should be setup to attract people, not run them off. Well, let's see, "classic", if a fast computer used 1h/wu and a slow 6h/wu, the fast would get 24 "wu-credit"/day while the slow 4 "wu-credit"/day, meaning fast gets 6x more credit/day. SETI@Home/BOINC v4.18, if a fast computer uses 1h/wu and a slow 6h/wu, and on average gets 24 Cobblestones/wu, the fast would get 576 Cobblestones/day while the slow 96 CS/day, and again you've got fast gets 6x more credit/day. With Seti_Enhanced v5.12, what has changed? On ar=0.426535, fast computer uses 2.6h while slow uses 15.6h on the same wu, and both gets 62.36 Cobblestones. On ar=0.693026, fast computer uses 1.41h while slow uses 8.48h, and both gets 33.90 Cobblestones... But hang on, on fast computer this means ar=0.426 is 24 Cobblestones/hour, while on ar=0.693 it's 24 Cobblestones/hour. For both angle-ranges, it means fast computer gets 576 CS/day. For slow computer, ar=0.426 is 4 Cobblestones/hour, while on ar=0.693 it's 4 Cobblestones/hour. For both angle-ranges, it means slow computer gets 96 CS/day. Meaning, appart for different amount of cpu-time for a specific wu, and a different amount of credit for this wu, the average granted credit/day is nearly unchanged between v4.18 and v5.12. This also means, a fast computer that used 1/6th the cpu-time under v4.18 and therefore got 6x more credit/day than a slow computer, will still get 6x more credit/day under Seti_Enhanced. Well, since not all computers is exactly the same and doesn't crunch all things exactly as good, and the "flops-counting" isn't completely linear with cpu-time, a fast computer can possibly get 24 CS/h on ar=0.4 and maybe 24.2 CS/h at ar=0.5 and 23.8 CS/h at ar=0.6 and so on. Also, while fast computer under v4.18 got average 24 CS/h and slow got only 4 CS/h, it's possible the fast on average gets only 23.5 CS/h under Seti_Enhanced while slow gets 4.2 CS/h, meaning 5.6x more instead of 6x. But, for 2 other fast and slow computers, it's maybe changed to 24.5 CS/h and 3.8 CS/h, meaning 6.4x now. Bottom line is, Seti_Enhanced just like v4.18 and earlier applications, gives more Cobblestones/day to a fast computer than to a slow. So, the incentive to upgrade to whatever is the "latest", "greatest" and "fastest" computer is just like it has been in SETI@Home/BOINC v4.18 and SETI@Home "classic". |
Crunch3r Send message Joined: 15 Apr 99 Posts: 1546 Credit: 3,438,823 RAC: 0 |
... Well it depended on the benchmarks run by boinc. Where did you pull that 24 credits off ? Thought it was intended to be more like 32 credits per WU on average ?
Again... 1 hour with official 4.18 ??? Never seen that happen ;) Maybe a P4 with 4 GHz not running HT...
Drop in credit claimed/granted. P.S. Anyhow... my personal note and again what i've seen/read on various boards is that credit claims/granted is way to less. Join BOINC United now! |
The MariahNet Network Send message Joined: 14 Jul 99 Posts: 173 Credit: 2,469,357 RAC: 0 |
Doesn't anyone care? Credit is not fair! I really could swear. To say, would I dare? Yea, Credit is not fair! Like cookies eaten by a bear. Clothes ripped and nothing to wear. Lost the spray for my hair. And my pants is no longer a pair. Credit is not fair! *a little head banging action* ;-) |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
First of all: no good deed ever goes unpunished. Any time the credit system changes, some are going to do better under the new system, and others not so good. Those who see a jump in credits will see the new system as "fair" and those who don't will see it as unfair. Simple as that. |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13848 Credit: 208,696,464 RAC: 304 |
my personal note and again what i've seen/read on various boards is that credit claims/granted is way to less. Still reminds me of pinball machines. The early ones the maximum possible score was 9,999. Bonuses were 10 points, 100 for a huge one. The last pinball machine i saw the maximum possible score was something like 10 million, small bonuses were 10,000. People are just greedy; it's that simple. Fair has nothing to do with it. Grant Darwin NT |
Pappa Send message Joined: 9 Jan 00 Posts: 2562 Credit: 12,301,681 RAC: 0 |
Well, let's see, "classic", if a fast computer used 1h/wu and a slow 6h/wu, the fast would get 24 "wu-credit"/day while the slow 4 "wu-credit"/day, meaning fast gets 6x more credit/day. If what I say upsets people you are welcome to say so... I do not own the computer hardware that you are making your claims from (fast)... I have a lot of computers working here, Seti Enahnced Beta and some doing Einstein just because... Pappa Please consider a Donation to the Seti Project. |
Ananas Send message Joined: 14 Dec 01 Posts: 195 Credit: 2,503,252 RAC: 0 |
Enhanced 5.12 claimed about the same with BOINC CC 4.19 and CC 5.x, so Enhanced 5.12 must habe been quite well adjusted to the previously claimed credits. CC 4.19 ignores the Fpops counter, that's why it can be used quite well as a reference for this comparison. With Enhanced 5.13, BOINC CC 5.x boxes claim about 50% of the amount that CC 4.19 claims. From all versions that I have seen in Beta, no version got closer to the CC 4.19 values than Enhanced 5.12. |
thinksnow Send message Joined: 26 Dec 02 Posts: 41 Credit: 505,798 RAC: 0 |
here is a link to a 3.6Ghz Xeon doing Seti Enhanced Beta that takes 19,465.36 Seconds (5.4 hours)56.96 CC AR 0.4847 or 5.4 Credits/hour... Opteron 165, O/C'ed to 2.46 (~on par with an FX-55) came in @ 8,081 seconds and I'm usually right in there at 7.5k-8.5k on the enhanced units. The longest it's taken me for an enhanced WU is 17,773 seconds with a short run of some that took 12k. 2,600 seconds on the old 4.18 units, meaning ~32WU/day. Now, if you're talking about "pure," non-Crunch3r processing, then nevermind, but it can be done, relatively easily, by a non-server machine. Main rig: AMD Opteron 165 Dual Core, 2Gb PC3200, 450Gb onboard + 1Tb RAID-5 NAS HTPC1: P4 2.8E, 1.5Gb PC3200, 3x80Gb SATA, Fusion5Lite + PVR-250, Sony 50A10 50" HD-LCD HTPC2: P4 2.8E, 1GB PC3200, 200Gb PATA, PVR-250, JVC 32" tube |
W-K 666 Send message Joined: 18 May 99 Posts: 19376 Credit: 40,757,560 RAC: 67 |
.... Where did you pull that 24 credits off ? Thought it was intended to be more like 32 credits per WU on average ? .... The average claim of 32 credits was for a reference unit. It was found that the reference unit took about 20% to 30% (depending on computer system) longer than an average unit so the correct average is about 75% of 32 = 24/unit. Andy |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
... Note that was "classic" (before BOINC) where 1 WU = 1 "credit". Joe |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 66302 Credit: 55,293,173 RAC: 49 |
... I can't help But agree with You Crunch3r about the drop in Credits claimed/granted, As I'm expected to do 4.5 times the work and for only about twice the credits, And to get any 4.18WU's I have to allow 5.12WU's on to My systems. I use Your App for Seti and I use Truxs 5.3.12.tx36 client to keep the cobblestones correct and from claiming more than is right and fair, Now I'm being told You need to do this extra work, And oh BTW here's a few extra credits tossed in, If It isn't the same amount of credits as before they think so what? They ought to come out of their ivory tower, I won't run their 5.4.9 client until there is a 5.4.9 compatible calibrating client as I do more than Seti, I do Einstein and Predictor too. As long as the correct amount of cobblesones were being gennerated I didn't see a problem. So am I a science Nerd? Not on Your life. You guys could have used some of that $250,000.00 on Newer hardware to keep up with the flood, Sun hardware might be nice, But 400MHz or 440MHz cpus? When were those relics made 2000? AMD owns near 50% of the Server market cause of the superior Opteron cpu, Isn't It about time the older Sun servers were changed out for Opterons? They have 4 and 8 cpu servers now with real dual core cpu capability and sata hdd drives in 10,000rpm spindle speeds too. All I want is the 1:1 ratio in credits vs. work to be respected, Until then No 5.12 Work Units will be processed, Only 4.18WU's will be allowed to be crunched. And I can abort every one of them as I have the time(takes very little time really or about the time to type this captioned line of text)......... Savoir-Faire is everywhere! The T1 Trust, T1 Class 4-4-4-4 #5550, America's First HST |
W-K 666 Send message Joined: 18 May 99 Posts: 19376 Credit: 40,757,560 RAC: 67 |
... Beginning to think the reason you are a juker is because you are not all there. I hope the village you originated from realises their idiot has gone missing. You seem to think the excess credits you have benn granted by using Crunch3r's optimised app and Trux's BOINC client are the right figure's. But you live in cloud cockoo land. The correct credits for version 4.18 were about 24 and version 5.12 has been adjusted in testing to reflect that. Live with it and stop your useless ranting. Andy |
BF Send message Joined: 7 Aug 05 Posts: 17 Credit: 1,594,515 RAC: 0 |
my personal note and again what i've seen/read on various boards is that credit claims/granted is way to less. Very good analogy and accurate assessment.. |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 66302 Credit: 55,293,173 RAC: 49 |
... Keep My handle out of It, Or go freeze Yourself on a Winters night. If It had been set for 24 someone would've objected to that too I'd think. Savoir-Faire is everywhere! The T1 Trust, T1 Class 4-4-4-4 #5550, America's First HST |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13848 Credit: 208,696,464 RAC: 304 |
I use Your App for Seti and I use Truxs 5.3.12.tx36 client to keep the cobblestones correct and from claiming more than is right and fair, And yet you bleat about now only being given what is right & fair for doing Seti work??? As long as the correct amount of cobblesones were being gennerated I didn't see a problem. The correct amounts are now being claimed & received. The perceived problem before hand was the excessive claims for credit, then not getting what was claimed (even though it was higher than the claims of others). Now you're upset because you're not getting as much as you were used to, even though previously you were getting more than you should have & you say that all you want is what is fair? You guys could have used some of that $250,000.00 on Newer hardware to keep up with the flood, Sun hardware might be nice, But 400MHz or 440MHz cpus? When were those relics made 2000? AMD owns near 50% of the Server market cause of the superior Opteron cpu, Isn't It about time the older Sun servers were changed out for Opterons? From the server status page. jocelyn: Sun V40z (2 x 1.8GHz Opteron, 8 GB RAM) This might come as a shock to you- but those systems aren't cheap. Then once you put come CPUs, memory, HDs in them they cost even more. As well as new hardware & maintaining the present hardware they also need to pay the people working on the system (they need to eat, have somewhere to sleep), pay for the power bills, software licences, other replacement hardware (switches, UPSs, batteries) etc, etc. All I want is the 1:1 ratio in credits vs. work to be respected, That is what is now happening, yet you continue whine about how it's not fair. Make your mind up. Until then No 5.12 Work Units will be processed, Only 4.18WU's will be allowed to be crunched. And I can abort every one of them as I have the time How old are you? 12? It'll be interesting to see if you ever do grow up. Grant Darwin NT |
Jack Gulley Send message Joined: 4 Mar 03 Posts: 423 Credit: 526,566 RAC: 0 |
You seem to think the excess credits you have benn granted by using Crunch3r's optimized app and Trux's BOINC client are the right figure's. But you live in cloud cockoo land. The correct credits for version 4.18 were about 24 and version 5.12 has been adjusted in testing to reflect that. Those "excess credits" were for using our machines more efficiently than the stock program would, and doing four times more science calculations in the same amount of time and expense. And for making the effort to use better programming skills, better compilers, better math libraries and advanced features of our processors. All of which seti "management" could have done with the stock programs, even back in "Classic" where a lot of computer processing power (and electricity) was wasted by inefficient programs. So now Seti Enhanced is using most of those user developed improvements (under the cover of more in depth analysis) to get a lot more useful work done, yet you are insisting all this additional useful work is now worth nothing, and only credit based on what the old inefficient programs would have done should be given. So, I and many others are now doing the same amount of useful work that we were doing (at our expense) and are only getting half to one third the credit of what we were getting for our efforts. And those who have made no effort to do more efficient work in the past are now getting a free ride. But then, no one has ever guaranteed LIFE, or not being "The Management", is fair. |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13848 Credit: 208,696,464 RAC: 304 |
So, I and many others are now doing the same amount of useful work that we were doing (at our expense) and are only getting half to one third the credit of what we were getting for our efforts. And those who have made no effort to do more efficient work in the past are now getting a free ride. The reality distortion fields appear to be running at full power. Grant Darwin NT |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 66302 Credit: 55,293,173 RAC: 49 |
I use Your App for Seti and I use Truxs 5.3.12.tx36 client to keep the cobblestones correct and from claiming more than is right and fair, I have nearly the equivilent of jocelyn(2 x 1.8GHz Opteron, 8 GB RAM) right now, It has only 2Gb of ram of course, It sure doesn't cost $100,000.00. And jocelyn existed before the $250,000.00 campaign and is the only modern nearly upto date computer that Seti has that I know of. Sorry about the work stuff, It was going to be deleted, The 1:1 was meant of course. Right now It's more work and less credit, What next forbid overclocked computers from running Boinc??? Whole teams are running that way. I think I have more grey hairs than You, Junior. Savoir-Faire is everywhere! The T1 Trust, T1 Class 4-4-4-4 #5550, America's First HST |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13848 Credit: 208,696,464 RAC: 304 |
I have nearly the equivilent of jocelyn(2 x 1.8GHz Opteron, 8 GB RAM) right now, It has only 2Gb of ram of course, It sure doesn't cost $100,000.00. Desktop system, or rackmount? You can get "decent" priced dual core dual CPU work station systems, but once you start going for managable rackmount gear with redundant power supplies etc, the prices skyrocket. Also the problem isn't just servers, it's network equipment & suitable UPSs, airconditioning so it all keeps going when the temperature's rise & on & on it goes. Right now It's more work and less credit, As has happened before in the days of Seti classic. There was a lot of wailing & knashing of teeth back then as well. People get used to seeing a certain number next to so many seconds of processing. And if that number gets smaller over time instead of bigger then the world must be ending, even though it's the same for everyone. *shrug* What next forbid overclocked computers from running Boinc??? Whole teams are running that way. BOINC has resulted in just that- because many of the Seti classic Work Units returned were nothing but errors. As long as the work done is valid, it will be accepted & people will get their credits. The more Work Units they do, the more credits they will get per hour. So overclockers have nothing to be concerned about- as long as they produce results, not noise. I think I have more grey hairs than You, Junior. Geez, i hope not. I started going grey in my late 20s & that was some time ago now. *too depressing to think about* Grant Darwin NT |
Jack Gulley Send message Joined: 4 Mar 03 Posts: 423 Credit: 526,566 RAC: 0 |
Now you're upset because you're not getting as much as you were used to, even though previously you were getting more than you should have & you say that all you want is what is fair? "Upset"? No just very disappointed, yet again, in how the Setiathome project is now being ran for the sake of BOINC and it's "new wave computer theology". "you were getting more than you should have" - Not when I look at most of my Credit Granted numbers. While Trux's calibration usually claimed 32, usually only 21 or so was granted on average, and often much less. Trux pointed out that his use of 32 credits was not to actually receive that much, but to bring the project average granted back up closer to the expected levels. There were a lot of big anonymous farms out there using optimized applications who's machines were only claiming 6 credits or so and this was dragging down the whole setiathome average, despite the efforts of some groups of cheaters to inflate their granted credits. |
©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.