Response to concerns regarding the new credit system.

Hi Everybody,

We at the project are aware that there are questions and complaints regarding the mechanism that the enhanced client uses to grant credits. We do take these concerns seriously and I will try to adress some of them here.

The first concern seems to be that claimed credit per hour has dropped for fast machines, especially those that use BOINC clients with optimized benchmarking code. That is undoubtedly true. Fast machines were generally claiming more credit per workunit than their slower counterparts. For those using optimized benchmarking code this was doubly so, since they claim more credit per hour than the same machine running a standard BOINC client without doing any more real work per hour. I am not suggesting that this was a deliberate attempt by people to claim more credit than they were due. I believe that in most of the cases this was done to match the BOINC benchmarks to the work done by optimized SETI@home clients.

The second complaint is that a large fraction of users are seeing a large drop in their granted credit per hour. This is not true. I was very careful in attempting to choose a credit per floating point operation (FLOP) that would be credit neutral overall. If there has been a change in average credit per hour it is of order 10%. Future revisions of SETI@home enhanced may have their credit multiplier adjusted in order to maintain parity with credit granted in other projects.

The current multiplier gives 3.35 FLOPs of credit per FLOP actually done. Despite claims to the contrary, this number was not calculated based upon processing times of Sun processors. It was, in fact, calculated based upon the compute times of a large number of machines that were running both the SETI@home and the SETI@home Enhanced beta. The vast majority of these machines were using Intel processors.

Another claim is that this change hurts people who were running optimized SETI@home clients. That is true to some extent, but there isn't a lot I can do about it without being unfair to even more people. The previous versions of SETI@home were not as highly optimized as the enhanced version is. It was relatively easy for optimizers to create versions that would run in a small fraction of the time. It is still possible to create optimized versions of SETI@home enhanced, but the gains will be smaller. So whereas people running optimized version were, in the past, able to boost the credit per hour of their machines by factors of 5 by running optimized versions, now they may only be able to boost their credit per hour by a factor of two. To offset that effect, our credit multiplier would need to be increased to around 8. That would more than double the rate at which everyone gets credit and double the rate at which the project as a whole generates credit. This would be obvious in the statistics, and it would be unfair to other BOINC projects if we were to offer twice the credit per operation that they do.

It has been suggested that I was ordered by David Anderson to reduce the SETI@home credits in order drive people to other BOINC projects. It is a great misunderstanding of our organizational structure to assume that David could order me to do anything, much less something that would hurt the project that pays (at least part of) my salary. I get no salary from the BOINC grants. I have no incentive to drive people away.

Eric
____________

Hi Eric, thanks for clearing things up.

There's a nice graph at Team OcUK, showing total stats for seti@home.

There has been no significant change in the last days:




Regards Hans
____________

[quote]

1. The first concern seems to be that claimed credit per hour has dropped for fast machines, especially those that use BOINC clients with optimized benchmarking code. That is undoubtedly true. Fast machines were generally claiming more credit per workunit than their slower counterparts. For those using optimized benchmarking code this was doubly so, since they claim more credit per hour than the same machine running a standard BOINC client without doing any more real work per hour. I am not suggesting that this was a deliberate attempt by people to claim more credit than they were due. I believe that in most of the cases this was done to match the BOINC benchmarks to the work done by optimized SETI@home clients.
2. The second complaint is that a large fraction of users are seeing a large drop in their granted credit per hour. This is not true. I was very careful in attempting to choose a credit per floating point operation (FLOP) that would be credit neutral overall. If there has been a change in average credit per hour it is of order 10%. Future revisions of SETI@home enhanced may have their credit multiplier adjusted in order to maintain parity with credit granted in other projects.
3. Another claim is that this change hurts people who were running optimized SETI@home clients. That is true to some extent, but there isn't a lot I can do about it without being unfair to even more people. The previous versions of SETI@home were not as highly optimized as the enhanced version is. It was relatively easy for optimizers to create versions that would run in a small fraction of the time. It is still possible to create optimized versions of SETI@home enhanced, but the gains will be smaller. So whereas people running optimized version were, in the past, able to boost the credit per hour of their machines by factors of 5 by running optimized versions, now they may only be able to boost their credit per hour by a factor of two. To offset that effect, our credit multiplier would need to be increased to around 8. That would more than double the rate at which everyone gets credit and double the rate at which the project as a whole generates credit. This would be obvious in the statistics, and it would be unfair to other BOINC projects if we were to offer twice the credit per operation that they do.[/quote}

1. I am having a hard time understanding this. Even using the optimized client, none of my fast machines ever claimed nor were granted more than the 32 (average) credits that were supposed to be the baseline. While many of the slower machines that I saw were claiming 40-44 credits per WU.

2. Again, this is not what most of us are seeing. If this were true there should have been very little if any drop in RAC (once credits started being awarded). However, just the opposite apears to be happening. I have only a few slower machines running enhanced at present, but each of them has a much lower RAC than they had without enhanced.

3. As to this, I have never heard anyone from the other projects complaining about the amount of credit being awarded to Seti participants. If they had an issue why not have them raise their amount of credit rather than Seti decreasing theirs? That would seem more beneficial to all rather than alienating a large portion of your volunteers.

Respectfully
Earl

____________

Again, this is not what most of us are seeing. If this were true there should have been very little if any drop in RAC (once credits started being awarded). However, just the opposite apears to be happening. I have only a few slower machines running enhanced at present, but each of them has a much lower RAC than they had without enhanced.

The first concern seems to be that claimed credit per hour has dropped for fast machines, especially those that use BOINC clients with optimized benchmarking code. That is undoubtedly true. Fast machines were generally claiming more credit per workunit than their slower counterparts. For those using optimized benchmarking code this was doubly so, since they claim more credit per hour than the same machine running a standard BOINC client without doing any more real work per hour. I am not suggesting that this was a deliberate attempt by people to claim more credit than they were due. I believe that in most of the cases this was done to match the BOINC benchmarks to the work done by optimized SETI@home clients.

This is not just about ME. This is about ANYONE who has put fourth the time, effort and money to build farms of the newest, fastest hardware, only to find that, by Eric's own admission, that it is this subset of users that are seeing a drastic cuts on their daily production of credit. The more faster computers you own and operate, the bigger your loss will be. This is what is so upsetting. Just because it's "happening to EVERYONE" is not a justification. It's a cop out. The fact is, that those with the fastest machines, and the most of them, are in effect being punished the most. The only ones who have any net GAIN from this are older machines running the stock clients and apps. Those that have spent tens of thousands of dollars, and the effort to use the most efficient apps, to do the MOST SCIENCE, are being HURT the WORST. It is a slap in the face for those of us that are in this position. And there are a great many of us. I am working on a way to let everyone know just how MUCH this means to us, and just what the potential loss to this project, at least in crunching terms, would be like.

Regards, Daniel.

I can tell you now that the only reason I am/was crunching SETI was for the increased credit when compared to the other projects. My resource share was no where near 100% SETI but sufficient so that I could hold my own while giving a majority of my resource share to other projects. In reality 1hr of computing time on SETI should give me the same credit as 1hr of computing time on LHC@home or MalariaControl.net or Leiden Classical etc for the same machine. If by going back to that "ideal" means that people stop crunching SETI then so be it, they were only in it for inflated credit in any case and not the science. By SETI releasing the source code and allowing optimisations by the general public they have sullied the idea behind a consistant credit per machine hr of crunching across BOINC based projects for that machine. If SETI or any other project wants to allow people to optimise their code, while still supplying non-optimised apps and the BOINC admin allow people to alter the credit claim then BOINC will not grow into what it could do as there will be no incentive to crunch for projects that rely purely on the standard BOINC framework. Afterall the whole idea behind BOINC was a consistant framework resulting in a consistant manner in giving credit for work done per machine hr of crunching. If you want more credit per hr, then you buy a faster machine, not change projects.

Non-project based optimisations should not be allowed. If a project wishes to optimise their application to increase the amount of science then it should occur completely within the framework of the project (like LHC@home).

Single quorum projects that grant credit purely based on a single result are also allowing escalated credit to be achieved and have a legion of crunchers purely doing those projects soley for the artificially high credit (yes I have done some of those as well).

The below is something that I posted regarding this in a different thread...

"Acknowledgement of people’s contribution is required to keep a majority of people involved with distributed computing. The conundrum that the developers of BOINC have been working on is how best to do this fairly. The benchmark idea was great at the time when only 1 or 2 projects were on BOINC and only a few individuals were optimising, but as more projects came online and optimised apps and clients became available en-masse, its limitations were discovered pretty quickly. Resulting in peoples preference for projects that gave good credit. The idea of counting the number of operations performed came out as a good replacement, but now very early in its public infancy we are seeing that it has problems as well. So how do the developers ensure that the same computer claims the same amount of credit no matter which project is crunched?

Do we just want to rely on having faster or more computers to boost our overall RAC? Do we, the volunteers, want to see a fair credit system? Are we for a socialist credit system or a free-market capitalist style of credit system? Do we want a credit system that is transportable across all BOINC based projects? Do we want to see a credit system that is only good on a specific project where we end up only (effectively) counting the number of wu’s completed and validated per project?"

I know what I want out of BOINC. What do you want out of BOINC?

Live long and crunch.

____________
Paul
(S@H1 8888)
And proud of it!

Saw the above response and decided not to expose myself to anything similar and removed my post.
____________
Kind regards
Kirsten

I guess that we have to wait a month ot two to see how things (stats) will develope/settle.
(My RAC is falling quicly while the number og pending WU's are rising and so is the number of preemted WU's - all 6 hosts running 24/7)

Most people like to compete - and almost all people like to compete on equal terms.
I guess that the staff at SETI is trying to provide us with an equal playground to compete on.

As of now it doesn't look like they have sucseeded in doing that - they have only managed to put 'a stick in the wheel' on the fastest competitors so far (like putting a 50 kilo backpack on the worlds best marathon-runners = a kind of 'handicap' system).

PS. When reading Eric's first post here - I got a sence of that the fastest Intel CPU's are very disliked amongst the SETI developers. Please correct me if I'm wrong.

Kiva


____________
Greetings from Norway

Crunch3er & AK-V8 Inside

Hi,

The second complaint is that a large fraction of users are seeing a large drop in their granted credit per hour. This is not true.

Wrong.

What you all don't understand is that by SETI allowing optimisations outside of the project and that by the BOINC admin (remember they are 2 different teams) allowing their source code to be optimised it has created this situation. Remember that BOINC is multi-project, NOT JUST SETI. If by bringing the amount of granted credit back to the same amount as the same machine would get on other projects this upsets 100? 200? 300? 400? 500? people and they stop crunching, at the end of the day it is but a small bump in the road. Does it mean that SETI looses 3000 hosts..then pfffft. SETI is not being fair to the other BOINC projects. Neither is Einstein. Neither is Rosetta. Neither is QMC. Neither is any other project that allows more credit to be claimed and granted than is defined by a typical BOINC install.

Again...just a few more cents worth.

Live and long and crunch (fairly).

Well said.

Mike

____________

Mate, you really have got a serious problem.
Whether you decide to leave or stay i reckon you need to just get away for a few weeks & get some perspective & take a good long hard look at things from the outside.
Leave your systems running, take them off line. Whatever you do just get away from the forums & everything else for a while & if you do decide to come back you can at least do so with a fresh outlook on things.
Just get a grip before you loose it completely.

Daniel, It's probably as tough for you to watch your rac fall as it was for those of us sticking with NON optimized Boinc core clients to watch ours stay the same as yours (and others) rose with artificially inflated claimed credits. You/others will still keep their credits and thus still be further ahead. I'm sitting here with 5 computers (two amd 64 3700's) and the best RAC I can muster is 1200 for all boinc projects. It's always been in that area. Yet you claim to get a RAC of 3500 from ONE computer.

In the current system advantage will be had by having more and faster computers. Guess what? You'll still have the advantage, just the numbers will be different.

I appreciate the help you've given me in the past and hope this doesn't cause a rift between us, but I'm having a hard time feeling sorry for you.

tony
____________

The other serious problem I have is with those like yourself who can never know what it is like to lose more RAC on ONE MACHINE that your total RAC. By simply switching to enhanced on ONE machine, it's RAC was still climbing at that point, went from about 1850 down to it's current 1408, and still falling. There are many others that are afflicted by this, and are just as unhappy as I am.

Daniel,
I don't know where you get this idea that many are affected by the granting of the real value of credits. I appreciate you are in the top 20 participants, but if this household which runs three separate accounts, where to combine forces and only crunch for Seti, with only our top computers, our rac using 4.18 would have been over 5,000, which would put us in the top 300, easily if the other 4 old computers were added.

I don't consider our computers, equals a farm, its a family. I also suspect that at least half of those in the top 200 participants are using computers to which they have access and permission to use from their places of employment. So these are not really in the same position as you are.

Therefore we are talking about maybe 100 participants. Not a lot even if we only consider active particpants, it may equal one in two thousand.

The Seti personnel don't have any real choice on the credits they allocate per time to a host. That was decided by the original BOINC/Seti setup using standard client and app. Any other value can only bring competion between the projects and that will not be tolerated, by management or the users.

It is all very clear on the projects other that Seti that there is at least resentment against users of optimised clients who over claim, especially on Rosetta where the validation of the result is done post granting of credit.

We have been using optimised apps will all see a drop in out RAC proportional to the speed of our computers. The only way you will see your climb up the BOINC (not Seti) ladder slow down is if you compare yourself to a simarly dedicated cruncher for another project, with the possible exception of Eistein crunchers who have a brilliant optimised app available, but the life of that is limited, probably to the end of the S4 project, and that is due to end in about a month.

Andy

@Zaak

Hmmm - didn't know about any Trux calibrating thing.
So I guess, that I have only been 'cheating' moderately - with an optimized 4.11 and 4.45 Boinc-client ;-)

You can help me to understand this 'Trux thing' by telling me the exact specs (if they are OC or not) of your 2 Pentium4 3Ghz host - so I can compare these with my own P4's (2,8 - 3,2 and 3,4Ghz).

Each of your two P4 3Ghz is having almost double RAC than my Prescott 3,2@3,7Ghz (s.478) and mine is cruncing WU's at 100% 24/7.
(and crunching a 4.18 reference WU in 4311 seconds as seen here (rated second): reference WU

Kiva
____________
Greetings from Norway

Crunch3er & AK-V8 Inside

Quote from the Gas Giant:

SETI is not being fair to the other BOINC projects. Neither is Einstein. Neither is Rosetta. Neither is QMC. Neither is any other project that allows more credit to be claimed and granted than is defined by a typical BOINC install.

as one of the other 99% of the world i would just like to say that i have been following the enhanced threads on all of the project sites but as yet have not bothered to use one .

i can understand that people who are using optimised apps are "loseing" credits

BUT if a FASTER machine crunches a w/u in 5 hours and clame 30 credits
and i crunch the same w/u in 10 hours and claim 30 credits they still have 5 hours in which to finish another w/u and claim another 30 credits
meaning that they will outpace me very quiickly
and if i want to keep up with them then i will have to spend money and upgrade to a faster machine

i have always thought that the size of a w/u should be the defining factor of a credit system as opposed to a timed one (less chance of anyone "fiddling" the credit system)i.e everyone gets the same credit for crunching the same w/u irrispective of time taken (can`t say fairer than that can you )
faster machines will STILL CRUNCH faster
i also think that if all projects could go to this credit system then a very level playing field would be had (also this way people would not just crunch for the highest granting projects )meaning more users for the projects

i agree that credit is nice to show how much work you have done (and i am quite happy with mine thank you )
and the apps that acually made the machine FASTER are fair enough but the ones that just multiplied claimed credits(i beleive that there are some please let me know if i am wrong ) i think are not so fair
why should someone claim more credit than me when we took the same TIME to crunch a w/u
i also have no trouble with overclocking as this makes the machine FASTER and there for if i want to keep up with you again i know what to do



just my thoughts NOT slagging anyone off
____________

Hi Daniel,

The effect of this being that those people who have spent an enormous amount of money and time on building a farm of the fastest crunchers, that do the MOST SCIENCE are, in effect, being PUNISHED for doing so. This is how we feel, as if we're being punished for doing the most work, in the most efficient mannor.

Hi Eric,

what's your position w.r.t to further possible optimisations of the enhanced app - Should I consider the current credit scheme as stable?

I might post some patches during the next weeks, but I don't want to cause further grief by triggering another adjustment.


Regards Hans
____________

what's your position w.r.t to further possible optimisations of the enhanced app - Should I consider the current credit scheme as stable?

@Zaak

Hi Zaak,

Thx - but I wasn't interested in the specs of your impressive Pentium D farm - but interested in the specs regarding your two 'slowest' host which are listed as Pentium 4(four) 3.00Ghz in your profile. ;-)

These two Pentium 4 (four not D) should be comparable with my own three Pentium 4's - if you will give me those the specs?
I will then be able to see how much this Trux has to say ;-)

Kiva

____________
Greetings from Norway

Crunch3er & AK-V8 Inside

Bump

Hans Graph is showing the "overall" granting of credit... This over time may show what is more "real."

Hi Eric, thanks for clearing things up.

There's a nice graph at Team OcUK, showing total stats for seti@home.

There has been no significant change in the last days:



Regards Hans

Consider this alternative scenario. Suppose that instead of creating the enhanced client and modifying the credit scheme we had just started distributing one of Cruncher's optimized versions. Would you have considered that to be unfair to people already using optimized apps?

Eric

No I would not have considered that unfair. As a matter of fact I had a mission to try to get everybody I could on optimized because it was helping the science by doing WUs quicker. But those of us who had spent time and money building farms of high speed machines for the competition would not have minded if optimized became the standard because it would not have adversely affected our production that we had built up.

No I would not have considered that unfair. As a matter of fact I had a mission to try to get everybody I could on optimized because it was helping the science by doing WUs quicker. But those of us who had spent time and money building farms of high speed machines for the competition would not have minded if optimized became the standard because it would not have adversely affected our production that we had built up.

The point of my post was that if everyone had transferred to an optimized application it WOULD have adversely affected your credit production (even more so than the enhanced application does) because everyone would have been claiming fewer credits due to shorter run times.

Eric

Hi Daniel,

I hope you don't consider this to be punishment. It certainly not meant to be. We are not deliberately reducing the credit of optimizers.

Consider this alternative scenario. Suppose that instead of creating the enhanced client and modifying the credit scheme we had just started distributing one of Cruncher's optimized versions. Would you have considered that to be unfair to people already using optimized apps?

The end result on the recent average credit of those already using the optimized version would have been much the same as with enhanced. People (not running a Trux BOINC client) would have started requesting less credit because their workunits would have finished much faster. Everyone who was already running an optimized client would see their recent average credit plummet by even more than has happened with enhanced.

In other words, would it be unfair to people running optimized versions if everyone started using an optimized version?

Eric

If everyone were using an optimized application and the optimized client, that would not be true. It would have had very little if any affect on my granted credit. :)

If everyone were using an optimized application and the optimized client, that would not be true. It would have had very little if any affect on my granted credit. :)

Well, yeah. But getting everybody to download an optimized core client would be pretty difficult.....

Eric

If everyone were using an optimized application and the optimized client, that would not be true. It would have had very little if any affect on my granted credit. :)

Actually it would. Running the optimized client and the regular optimized boinc core client (not trux) resulted in my pentium D machines claiming around 5 credits per wu. Trux changed that equation, and only by reporting inaccurate benchmarks and crunch times.

I think you would be even more surprised at the small percentage of the total number of active users that have even heard of optimised clients, let alone use them.

Edit- fixed spelling.

On or about the 8th of October, I pulled 76 wus from my result id's and pasted the contents to Excel. I examined the results, marking which were from Optimized seti applications and which weren't. I then found out that:

313 results were included in the 76 WUs.
Of those 313 results, 92 were from optimized apps (29.39%), Of course all my results (76) were optimized and included in my study, so I subtracted my 76 from the 92 and find that only 16 results of the (313-76=)237 other results came from optimized apps (6.75%). So at that time 6.75% of results returned were from optimized apps.
Of those 76 WUs, 12 contained 2 results from optimized apps, 3 contained 3 from optimized apps.
8 times out of 76 (10.5%)was an optimized app involved in the quorum and validation process(remember all mine are from optimized apps, so this percentage is HIGH). Only twice would 2 opt apps been in the quorum if I took out my results.(2.6%)

Remember this was OCTOBER and certainly there would be a higher percentage now. Anyone feel like spending hours doing this?

Funny Tony reported his finding just now.
I was doing my own look at granted credit etc. taken from the CSV file generated by BoincLogX.

This is a Pent M, mainly running optimised apps, first from TMR and then Crunch3r's, three units ate the beginng and 24 at the end after enhanced was introduced ran the standard app. The normal claimed credits using Crunch3r's optimised app was in the range of 6 to 7 credits. It has never run an optimised BOINC client.
The totals include all units including ones with errors. So the true average would be higher if I could be bothered to find and delete those entries.
Most, but not all, of the time it ran with a three of four day cache to increase chances it's low claim was not included in initial validation/granting of credits process.

Start Date _ _ _ __ _ _ 2005.06.29 21:10:43
Last 4.11/4.18 date _ _ 2006.05.19 11:23:04
Total units crunched _ _ _ _ _ _ _ _ _4059
Total Credit Granted _ _ _ _ _ _ _ _ 101031
Average Granted _ _ _ _ _ _ 24.890613451589

As the average granted for 4.11/4.18 Seti units is about 24 or 25 this shows that a sensible use of optimised app's did not require an optimised client. The only thing that could have drawn me to use an optimised client was the ability to use cpu affinity, but I had cracked that problem long before on the dual P3.

Andy

DON'T FORGET.....NO-ONE FORCED YOU TO CRUNCH!
YOU ARE A VOLUNTEER!
I've notice my RAC drop.
I've notice my Ranking drop recently....
It's not the end of life as we know it!
____________

I was just a volunteer in the beginning. But after Berkeley began asking for donations, I also became a sponsor/investor/sugar daddy. So I think my opinion should matter at least a little bit.

Trux's calibrated boinc core client simply adjusted the benchmarks to compensate for the improperly low credit being claimed by faster optimized computers. Slower PCs were getting closer to 32 credits per completed WU unoptimized. Faster unoptimized computers were getting less credit simply because it didn't take as long to finish a WU, and optimizing the SETI app (without optimizing the boinc client with calibration) only reduced claimed credits even more on faster machines.

Agreed, but the entire intended credit system for Seti (before enhanced) wasn't meant to have faster machines claim the same credit per wu as slower machines, it was meant to grant credit more on a time spent crunching approach.

Now, let me show you why those of us who care about stats are a wee bit upset with this...

Our team, SETI.USA, is currently the #1 SETI team for credits/day. We are still a young team (14 mos old). We motivate ourselves to overtake the older established teams by forecasting the amount of time it will take us to surpass the other teams' total credit. SETI.Germany has the largest total credit of all SETI teams - more than double ours. They accumulated this credit under the prior system, which is now changing. Even though both our team and SETI.Germany will be affected by the lower credit granted per day under enhanced, it will now take us much longer to pass their total credit.

Look here for the current projections - http://www.boincsynergy.com/stats/overtake.php?project=sah&team=115396

As you can see, we are currently projected to overtake SETI.Germany's total credit in approximately 781 days - a little over 2 years. With the credit changes being seen with enhanced, it will now take us probably 4-6 years to pass their credit - or longer.

Wow, I never realized the entire seti project exists just to allow Seti.usa to pass other teams faster. And here I thought I was crunching for a science project ;-).

funny how this "I have some spare cpu-cycles, let me help science" turned into "gimme my credit or else...."
____________

I normally ignore you but I am going to respond here. What Trux did was accepted by the project and a large number of the volunteers so do not go there.

I normally ignore you but I am going to respond here. What Trux did was accepted by the project and a large number of the volunteers so do not go there.

This implies that the project could have blocked modified clients somehow.

They could have, by simply not releasing the BOINC source, but means we'd never see BOINC for anything but a "mainstream" OS, or something that is 100% binary compatible.

... and if they did start trying to block "improved" BOINC clients, those clients could just report standard versions -- a "stealth" enhancement.

"Acceptance" means a whole lot less than one might think -- blocking enhanced BOINC clients may not be impossible, but it is impractical.

I was very rarely if ever given the points requested by my computers.

Please read more carefully the statements about credit awards were in reference to Optimized Seti from Trux and or Cruncher3.
It has nothing about Enhanced points except to express a contradiction to what Berkeley stated about the difference in point awards was only 10%.

You may want to re-read the original statement from Eric Korpela as well.

He stated that Enhanced credits are within 10% of standard 4.x clients, not within 10% of optimized.

However, a person running 10 CPUs that made 5000 credits per day under 4.x and now is only seeing 4500 credits from the same 10 CPUs under 5.x might well perceive it much differently. He's now seeing a net loss under 5.x of more than the single CPU cruncher above is making per day - effectively, the loss of a full machine's production, plus a little bit more.

Sorry Ned, but I have to disagree. My RAC has dropped well in excess of 10%: 16 days ago it was 1,137 and during the past week -- when I've been crunching mostly enhanced WUs the same number of hours each day -- it's dropped to 876. I could understand and live with a 10% reduction; 23% is beyond belief. And why does it have to be a level playing field? People who own faster PCs or have several PCs crunching for them deserve more credit.

RAC isn't relevant.
I think we'd be better off if it was removed all together.

RAC isn't relevant.
I think we'd be better off if it was removed all together.

@Grant
Offcause RAC is relevant. Specially after SETI started linking to different Cross-project stats-pages on 'Our Account' pages. The statspages are really striding with heavy traffic/load now.

Most of my hosts are living their own life in a dedicated room (due to heavy fan-noise) with some small tasks (besides crunching 24/7) for the house like ex. video-surveiliance, fileserver etc. They are unatended most of the time.

I use RAC to monitor their health and keep them sound and healthy.
As a native born competitior I also use RAC to see when I will overtake or will be overtaken by other 'competitors'. Wouldn't like to be without it. ;-)

@Xaak
Thx for providing the specs for the P4's. I now have a clue to the effect/impact of the 'Trux thing' ;-)

PS. Resetting the stats now seems to be a good idea, as this is a major change! So we will have Classic, Pre-enhanced & Enhanced Stats = Fair playing field.

Kiva

Hi,

So I don't see where your statement of a 10% difference holds any water or is factual.

@Grant

Thanks for having a close look on my stats. ;-)
These bad runs occured before I migrated to Enhanced - which I did approx on May 12.

I had crashes on some of the hosts (overcloked) due to severe heatproblems at the time where these bad runs occure.
Here in Norway we had an extremely hot Heat-wawe in the beginning of May. Hottets ever recorded in Norwegian weather-history. (Normaly there's no need for air-conditioning (cooling) here in Norway).
After these crashes, BOINC wouldn't resume on WU's and had to download a complete set of new WU's. WU-cache is set to max of 10 days.
I think I might had deleted to many old WU's, that I thought was not 'in use' any more.
Do you think that this 'deleting' could have caused all the 'bad runs' ??

PS. I still think thou, that by increasing crunhing time by a factor of 2-4 is a major change and have caused many new users to the forums ;-)

Kiva



____________
Greetings from Norway

Crunch3er & AK-V8 Inside

... Most of my hosts are living their own life in a dedicated room (due to heavy fan-noise) with some small tasks (besides crunching 24/7) for the house like ex. video-surveiliance, fileserver etc. They are unatended most of the time.

I use RAC to monitor their health and keep them sound and healthy. ...

Most people are competitive by nature. You start crunching and focus on the one ahead of you and run the calculations of how long it will take to catch the one ahead of you. You add machines to shorten the time span of conquest. You get used to the current playing field and rate of progress. With enhanced, everyone was left on the same playing field, but had the rug pulled out from under them on rate of progress. The time to catch the one in front of you has multiplied many times over. The motivation slackens at this point. People looked at the credits as payment for their efforts. Many crunch for pure science, which is great. Many others crunch for the credits, which is also great. It is time consuming to keep up with a lot of machines and to keep them tweaked to top performance. You have your average drivers and your professional race car drivers. The pros have all just had governors put on their cars that they spent thousands of dollars on. They are not happy. The average driver does not understand that. They are still operational and driving just fine. The short time to catch the vehicle in front of you has been extended to the extreme. This has severely challenged the competitive nature and thus has created protest. These professional crunchers give a lot to the project. Higher credit for a Work Unit hurts no one and brings everyone into the perspective of the old playing field. Higher credit also brings everyone to the professional level of output per machine. This is something to be considered. Higher credit per WU will not upset those crunching for the science because they will still be doing just that and it will silence us credit mongers. Just my 2 cents.

As I crunch for several projects I have seen my overall average/day decrease slightly and that is mainly due to my increase in pending here on Seti due to the longer crunching times.[/quote]

I think we've all seen a marked decrease in RAC, but this will work itself over a little time, as pending credits are converted to actual credits. As for time taken to overtake the "one ahead of you", (re: post from Lazy) since the credit granted for each WU is now correspondingly larger, this argument has to be considered nonsense. Although each individual WU takes longer to crunch, the total credit gained will be approximately the same. The person ahead of you is also earning in the same way, so if that is what you are after, you will still achieve your aims.


____________

It got into the mid 20s?
;-)

Since all but 3 of my machines are still crunching standard work, my stats have not as yet bottomed out. But, I am forecasting that once they are all crunching enhanced, my RAC total, which would have ended up at around 31,000 with my newest upgrade, will end up at 10,000- If I'm lucky. I just don't see the reward in that. It feels more like a punishment.
Respectfully, Daniel.

@Xaak
Thx for providing the specs for the P4's. I now have a clue to the effect/impact of the 'Trux thing' ;-)

I forgot to mention, both were running Crunch3r's SSE3 optimized seti client for windows. That's where some of the increased numbers came from.

Trux added an average of about 8 credits per wu when I was part of the granting quorum vs. no trux. With an average of 85 wu/day from the 630, that worked out to approximately 680 additional credits granted just from the trux calibrations.

Trux's Client may make the claim for that computer higher, but there are no gaurantees that it will increase the granted credits. Occasionally it will get another unit where it is paired with another high claimer and increase the grante for that unit. But it is just as likely to be either the last to report and be excluded fron the validation/granting process or be paired with two low claimers.
There is no evidence that any method of claiming high has increased the granted credit. It was about 24 before any of the optimisations were released and still is today for versions before enhanced was released.

Andy

Actually, if you're in the quorum (first 3), it pretty much guarantees you'll get more credit. Here's how.

Without trux that box was claiming 7, which was almost always the lowest, so the next lowest, usually around 15 - 20 became the grant. After trux, that same claim was 32 or so, making it one of the highest. So now that 15-20 became the lowest, mine became the middle or highest, giving me anywhere from the same or a little more credit granted to 32 granted. Average increase was around 8 credits per wu.

True, there's no guarantee, but the data I collected at the time I swiched proved that increase, and the increase was soley due to the change Trux made in claimed credit.

I'm fine with the fact that we can't do that anymore. The option for an unfair advantage is no longer present. I'll still optimize and get the most I can out of the tools available under the current scoring system, but I won't cry about a drop in rac due to using loopholes in the old system that allowed me and others to overclaim credit.

Eric, your post is well-understood and accepted by a great many of us.

The Enhanced setup is infinitely more fair and is the closest we've come by far to distributing credits equitably, not only here but across the BOINC world. People complained when we switched over from Classic because they wanted their 1 credit for 1 workunit. On Classic it didn't matter how fast or slow your machine was... if you completed the workunit you got your reward, and it was the same reward everyone else got. Now we have basically the same thing here, and people are still complaining.

I do understand that it must sting to see your RAC drop but this is fair and necessary to level the playing field. Those folks are now joining the crowd, not being punished by it. And it wouldn't have mattered one bit if it had been decided to raise everyone else's credit rather than removing the optimized-client bias. The end result is still the same. Faster machines process workunits faster and get more credit, just the way it has been for seven years. This is what I thought the BOINC credit system was striving for to begin with, an equitable reward like Classic plus the added benefit of compensating for anomalous units.

I never quite understood why optimized clients were even permitted. Optimize the apps certainly, build your farms, tweak your machines, crunch faster and more efficiently, and compete for more credit that way. That is fair. Artificially inflating credits is not.

I never quite understood why optimized clients were even permitted. Optimize the apps certainly, build your farms, tweak your machines, crunch faster and more efficiently, and compete for more credit that way. That is fair. Artificially inflating credits is not.

It depends what you call "artificially inflating credits" manual ajusting the credits or shall we criple the compiler doing a great job?

I never quite understood why optimized clients were even permitted. Optimize the apps certainly, build your farms, tweak your machines, crunch faster and more efficiently, and compete for more credit that way. That is fair. Artificially inflating credits is not.

I never quite understood why optimized clients were even permitted. Optimize the apps certainly, build your farms, tweak your machines, crunch faster and more efficiently, and compete for more credit that way. That is fair. Artificially inflating credits is not.

So, the release of Enhanced voids all the optimized clients, and new ones are needed. For those who came to rely on them, well, with any luck there will be new, faster Optimized Enhanced clients that also request the right amount of credit for crunching the same WU.

I'm not sure how Work Units that were deleted manually would show up in the error logs, most likely just as being not returned by the dead line.

I'm not sure how Work Units that were deleted manually would show up in the error logs, most likely just as being not returned by the dead line.

I believe those say “Aborted by user” or something of the kind.

I'm not sure how Work Units that were deleted manually would show up in the error logs, most likely just as being not returned by the dead line.

I believe those say “Aborted by user” or something of the kind.

If the wu were simply deleted from the hd, then it would not be reported and would have to time out as not being returned by the deadline.

I'm not sure how Work Units that were deleted manually would show up in the error logs, most likely just as being not returned by the dead line.

I believe those say “Aborted by user” or something of the kind.

If I understand his question right, they would not be reported as aborted unless he/she had actually highlighted the wu and hit "abort" in boinc. If the wu were simply deleted from the hd, then it would not be reported and would have to time out as not being returned by the deadline.

21 hours for 1 WU and a whole 60 credits claimed and granted. Lets see, that equates to less than 3 credits per hour of work. Funny thing is the other two in the quorum were around 7 hours and they claimed and got the same 60 credits or an average of almost 9 credits per hour of work...... three times what I got....

Yeah that REALLY sounds fair to me...... NOT!!
____________

So, the release of Enhanced voids all the optimized clients, and new ones are needed. For those who came to rely on them, well, with any luck there will be new, faster Optimized Enhanced clients that also request the right amount of credit for crunching the same WU.

If you mean the optimized science applications, then I wholeheartedly agree with you. I am a huge fan of Crunch3r's apps and can't wait to get going with an optimized Enhanced app. It was the optimized BOINC clients that I was expressing concern over, and the fact that anyone was free to compile their own code and request whatever credit they felt they deserved. There needed to be standardization in credit distribution, and I think we now have that with Enhanced.

21 hours for 1 WU and a whole 60 credits claimed and granted. Lets see, that equates to less than 3 credits per hour of work. Funny thing is the other two in the quorum were around 7 hours and they claimed and got the same 60 credits or an average of almost 9 credits per hour of work...... three times what I got....

Yeah that REALLY sounds fair to me...... NOT!!

21 hours for 1 WU and a whole 60 credits claimed and granted. Lets see, that equates to less than 3 credits per hour of work. Funny thing is the other two in the quorum were around 7 hours and they claimed and got the same 60 credits or an average of almost 9 credits per hour of work...... three times what I got....

Yeah that REALLY sounds fair to me...... NOT!!

21 hours for 1 WU and a whole 60 credits claimed and granted. Lets see, that equates to less than 3 credits per hour of work. Funny thing is the other two in the quorum were around 7 hours and they claimed and got the same 60 credits or an average of almost 9 credits per hour of work...... three times what I got.... Yeah that REALLY sounds fair to me...... NOT!!

If you wish make more credits per hour now the only way is to bring more cpu's to the party.

Of use a client that is able to do more calculations per hour.

Yes, in theory... but I'm confused.

Take a look at this result for enhanced.
http://setiathome.berkeley.edu/workunit.php?wuid=79097680
My machine is 1924541 - athlon 64x2 4200+ running crunch3r's 5.12 sse3
This is not a slow machine, and yet it comes out with the slowest time in the set.
The credit system with enhanced seems to work the same as setiathome with the middle claim awarded.

I'm completely mystified!
Andy.

Yes, in theory... but I'm confused.

Take a look at this result for enhanced.
http://setiathome.berkeley.edu/workunit.php?wuid=79097680
My machine is 1924541 - athlon 64x2 4200+ running crunch3r's 5.12 sse3
This is not a slow machine, and yet it comes out with the slowest time in the set.
The credit system with enhanced seems to work the same as setiathome with the middle claim awarded.

I'm completely mystified!
Andy.

I wouldn't say 2,000 seconds faster is "easily" out performing, especially when the second CPU has a 1GHz clock advantage...
:-)

If you look in the Statistics, Top Computers, the highest X2 is down at position 93, and the only other AMD's above it are a couple of Opterons, the other 90 positions are filled with Quad core Mac's and Intels, mainly Pent D's and Xeons.

Andy

I can't figure out this whole SETI mess. I joined in Nov 1999 and I see my ID is 1100989,,a very high ID number compared to others that joined so long ago, near the start of this. I gave it almost 2,500 hours time and got hardly any credits.. in fact I have all of 447 credits with a RAC of 8.47 (getting only .11 credits on one session).. this credit thing seems pointless.

AFAICT the ID numbers on ‘auto-migrated’ Classic accounts represents the user’s rank in terms of the number of work-units completed at March 2005, when the penultimate database ‘snapshot’ was taken.

AFAICT the ID numbers on ‘auto-migrated’ Classic accounts represents the user’s rank in terms of the number of work-units completed at March 2005, when the penultimate database ‘snapshot’ was taken.

Correction, 14. May 2004, when the SETI/BOINC-database was initialized.

I can't figure out this whole SETI mess. I joined in Nov 1999 and I see my ID is 1100989,,a very high ID number compared to others that joined so long ago, near the start of this. I gave it almost 2,500 hours time and got hardly any credits.. in fact I have all of 447 credits with a RAC of 8.47 (getting only .11 credits on one session).. this credit thing seems pointless.

Too many postings to go through at this "late date". This may have been stated before, but I thought I would give everyone my experiences.

I started out crunching using the standard application and when I started reading about the optimzed applications and how much faster they were, I switched over to them.

I saw my credits per WU drop from about 32 to about 8-10, but I was processing at least 4 times faster, so it worked out in the end for me. No problem, right? Wrong.

Then the postings started about how the optimzed applications were pulling down the folks that were running the Standard Application. I checked my WUs and sure enough, if 2 of the first 3 results were optimized, the standard application got the short end of the stick. One solution was to run with a larger cache size to give the standard application users more time, but this didn't always work out.

Then Trux released his version of the Client that optionally corrected the credit reported. I switched over to this and my RAC and Total Credits started up at a fast pace. This may or may not have been proper, but I did it so that I would not have to have such a large cache and so that I would not adversly impact the standard application users.

Then came the Enhanced application. The amount of work being done by the application increased, but this was partially offset by the programming optimizations that had been developed by the optimized applications. (If you think the run times are long now, consider what they would have been like without the programming optimizations!)

I am currently running Crunch3r's optimized applications that are specificly compiled and linked for my CPUs. (This is something that the standard application can not do.) I know that he included additional code in the program to report additional information from the application, but I have to trust that the only "optimization" done had to do with the compiler and link libraries. (But then I am a trusting sole.)

I am currently reporting reporting my claimed credits on a par with those runnig the standard applications. I do see that I'm using 40-45% less CPU time.

And before someone starts up about different CPU speeds, etc. This is what I based my statement on...

Using 2 results from the SAME WORK UNIT (no AR impact this way), I compared the closest result to mine:

The other system (Pentium D, 3.46GHz, Measured Floating Point Speed: 1527.34, running BOINC Client 5.4.9, standard application, and XP Pro.) returned a claimed credit of 62.52 in 22,151 CPU seconds.

My system (Pentium D, 3.00GHz, Measured Floating Point Speed: 1497.11, running BOINC Client 5.4.9, Crunch3r's SSE3 optimized application, and XP Pro) returned a claimed credit fo 62.88 in 12,149 CPU seconds.

While I don't know if the other system was a Pentium D830 like mine, I do know that it was reporting a similar FP Speed to mine. I also know that that one comparision is a VERY poor basis for any statistical evaluation. It's all I had the time to work with.

I don't know if my credits are "fair" or not. (The standard application doesn't report a "flopcounter" like Crunch3r's application does.) I do know that I crunch more WUs in the same amount of time and my claimed credits are on a par with the standard application results.


____________

I don't know if my credits are "fair" or not. (The standard application doesn't report a "flopcounter" like Crunch3r's application does.)

I don't know if my credits are "fair" or not. (The standard application doesn't report a "flopcounter" like Crunch3r's application does.)

The stock app does report how many FFTs it’s performed, though, which I believe is directly proportional to the Flop count.

This NEW credit system is BS!

It takes 3 times as long to do a WU and you only get 15 Percent more credit, what the hell is that all about???

Ok here is my 2 cents on all this.

First I have used the optimized apps for some time. At first I used just the app not a optimized client as well. What I saw was a penelty on the crunchers useing the standard app. I would turn a 7 to 15 credit WU in . My low number caused a less than the so called 32 credits per WU to come about. So I switched to truxs client. After a short while of the client bumping numbers to hit the golden 32 per wu it stopped bumping numbers and held it there. Did my RAC raise doing this? Yes, but very little over not using truxs client. Everybody seems to be forgeting the 3 WU average that you actaully get. So did I do myself a favor by useing truxs client, yes I could feel better about the people using the standard client not getting tanked for their hard work.

Now as far as the new credit layout. It does suck, and yes it is tanking my RAC. But not as much on my fast machines as on my slower machines. The ones that under the optimized layout was producing about 120 RAC , are now down around 40 RAC. With crunch times going from 3 and a half hours to 15 plus hours. Now tell me that is fair at all to any of us.

The Farm owners like myself expect something in return for the thousands spent on machinery and the hundreds to thousands spent every month on electricty to run the farms produceing science for the project as effecient as possible.

I have seen my RAC fall from 4k to around 1.5k.

Now if you look to the left you will see my RAC is under 1k now. Well that is until the credit problem is fished out. I have pulled almost all my machines into other projects. I have sent a letter to my team on this, and told them to do as they see fit. That it is not a team pullout, but a personal pullout. I may come back full force, when things get straightened out. But for now I have put the weight of most of my machines into endevors that can help the human race on this planet. I have left a couple machine in the project to monitor how things are doing. I wish all well, no hard feeling, but I am stepping back to see how things are going to end up.


____________

Beach Bums Current Stats:


Come Join us at Hawaiian Beach Bums

I started out crunching using the standard application and when I started reading about the optimzed applications and how much faster they were, I switched over to them.

That does not answer my question. If you get 30 credits for 45 minutes work on Einstein and way less than that here, then how has enhanced made the credit system of Seti equal to that of other projects? Weren't we told that was the purpose of changing the way credits were granted? Sounds to me like they should have gone up and not down.

At the moment on Eistein you are able to take advantage of optimised applications. So bearing that in mind, Einstein, if a host is using the opptimised app, should be taken out of the comparison. All other projects will give approx the same credits/time.
If you can identify were the developers started from it sometimes gives a hint as to which cpu/OS combination works best. For instance the CPDN app is an adaption of an older program written for mainframes etc using Fortran.
So it has to be recognised that for some projects their apps are more suited to some cpu/os combinations than others. Some projects like cpu speed, some a large L2 cache (Seti) and some give better performance with large amounts of fast memory.
So if you want to chase for the maximum credits, do your research and put your computer(s) where they will gain most credits.

Andy

Maybe my question is unanswerable, I don't know, but everybody is just dancing all around it. If there are still MAJOR differences between each projects credit granting, then what has the change here accomplished? If the intent was to equalize the projects it appears to have failed. Otherwise there would not be projects with higher granted credit for me to move to.

I brought this over from the other thread in the hopes that maybe someone from Berkeley can answer my question.

... The Farm owners like myself expect something in return for the thousands spent on machinery and the hundreds to thousands spent every month on electricty to run the farms produceing science for the project as effecient as possible. ...

Maybe my question is unanswerable, I don't know, but everybody is just dancing all around it. If there are still MAJOR differences between each projects credit granting, then what has the change here accomplished? If the intent was to equalize the projects it appears to have failed. Otherwise there would not be projects with higher granted credit for me to move to.

I brought this over from the other thread in the hopes that maybe someone from Berkeley can answer my question.

Because I don't do all projects I cannot answer your question but on fairly good authority I am told using standard apps and client the credits are approx equal across the projects.


The person I think that does most projects is John Mcleod VII (aka JM7), he designed, coded and updates the scheduler. He could possibly answer your questions.

While I was crunching on the Beta trial for enhanced, I posted several posts about credits - angle_range and credits/hr two examples are msg # 2063 and msg # 1989 The figures for Einstein in this one was before the optimised apps were available over there.

Andy

... go to another project that gives more credits, when everyone is saying that there should be no such thing.

Maybe my question is unanswerable, I don't know, but everybody is just dancing all around it. If there are still MAJOR differences between each projects credit granting, then what has the change here accomplished? If the intent was to equalize the projects it appears to have failed. Otherwise there would not be projects with higher granted credit for me to move to.

I brought this over from the other thread in the hopes that maybe someone from Berkeley can answer my question.

Because I don't do all projects I cannot answer your question but on fairly good authority I am told using standard apps and client the credits are approx equal across the projects.

The person I think that does most projects is John Mcleod VII (aka JM7), he designed, coded and updates the scheduler. He could possibly answer your questions.

While I was crunching on the Beta trial for enhanced, I posted several posts about credits - angle_range and credits/hr two examples are msg # 2063 and msg # 1989 The figures for Einstein in this one was before the optimised apps were available over there.

Andy

Yet at every turn you are encouraging me to go to another project that gives more credits, when everyone is saying that there should be no such thing. Apparently you are giving advice without knowing the facts. Please have your facts straight before trying to answer my questions in the future. I still await someone who can answer it legitimately.

Maybe my question is unanswerable, I don't know, but everybody is just dancing all around it. If there are still MAJOR differences between each projects credit granting, then what has the change here accomplished? If the intent was to equalize the projects it appears to have failed. Otherwise there would not be projects with higher granted credit for me to move to.

I brought this over from the other thread in the hopes that maybe someone from Berkeley can answer my question.

Because I don't do all projects I cannot answer your question but on fairly good authority I am told using standard apps and client the credits are approx equal across the projects.

The person I think that does most projects is John Mcleod VII (aka JM7), he designed, coded and updates the scheduler. He could possibly answer your questions.

While I was crunching on the Beta trial for enhanced, I posted several posts about credits - angle_range and credits/hr two examples are msg # 2063 and msg # 1989 The figures for Einstein in this one was before the optimised apps were available over there.

Andy

Yet at every turn you are encouraging me to go to another project that gives more credits, when everyone is saying that there should be no such thing. Apparently you are giving advice without knowing the facts. Please have your facts straight before trying to answer my questions in the future. I still await someone who can answer it legitimately.

I'm not encouraging anybody to move, all I said was that those who want to crunch for the credits, can at the moment, go and do some crunching for Einstein until the end of S4 as they have optimised apps. From what I have seen their next project will probably be optimised from the beginning and possibly use Fpops or similar. So that sourse of extra credits/time will dry up.

And if they want to go anywhere else then to get the most credits/time then they need to do some research on which project will give them the most for their combination of cpu/OS.

So to be honest, in the long term, Seti enahanced with Crunch3r's apps is probably the best bet for credit hunters.

Andy