Second, I admit it, part of me is credit whore. I have invested thousands of dollars in hardware and pay over $400 a month for electricity to keep my farm running. I do this because I want to earn more credit. My farm is split working of different projects. Partially because I seen value in the work the other projects are doing and partially because they pay better. I even have some of my computing power focused on lower paying project because I value the work being performed. On the other hand some of my crunchers are working higher paying projects strictly for credit. If S@H paid more, then I would devote greater resources to it.

Carlos,

Are you familiar with the artist J. S. G. Boggs?

He'd be fascinated by the cobblestone economy.

Mr. Boggs is a performance artist, and to my knowledge, the only one whose art can be collected.

He draws money. His money is convincing enough that his art has been seized by the U.S. Treasury -- they have not chosen to prosecute him, or return his art.

He draws a $20, walks into a store, and says "I would like to by this product, and I would like to trade my art, at face value, for this item, and receive change."

If the clerk refuses, he asks how his "art" is different from the "art" printed by the treasury -- and why one is "money" and the other isn't.

It's kind of the same thing here: we think cobblestones are worth something because we think they are worth something.

... and we all have different "work ethics" when it comes to BOINC. Some will crunch anything for "money" and others have to be genuinely interested in the result.

-- Ned

---

I think I would rather talk about religion, politics or if O.J. was guilty or not than talk about credit reduction. BUT... a credit reduction does reduce the value of crunching for SETI vs. any other @home project in a team setting, like my Air Force team vs. Navy for example. I also recall the top of the BOINC food chain saying in other forums and posting here, SETI is the big gorilla and all other @home projects need to align their RAC to SETI. If SETI users did not use the AK-8 app. or any optimized app. for that matter, would a credit reduction be necessary? I am really convinced BOINC needs to get the other @home projects straightened out and credits more in line to the big gorilla before doing a credit reduction on SETI. As a side note, the wonderful job mslatter (Mark) done selling his time (and money) to the highest bidder being SETI USA, just apparently had the value of the efforts reduced.

This surprisingly is my position too! The "gold standard" relative to seti is being changed and the new work will not be comparable. The same computer will doing the same work will apparently be getting around a 15% less credit with this amount based on a floating snapshot "benchmark".

Chasing an average BOINC benchmark across across that mess of a previously referenced table, bearing in mind the wide variety of project calculation types and the not representative balance of the mythical cobblestone computer, it seems a lot like "tilting at windmills".

Cross comparison of BOINC projects is a great concept, but is it really attainable?

I wish to see the actual calculation basis of the credit change with that Cobblestone "Gold Standard" against seti. If in reality the change is chasing the multi-headed BOINC variable monster and then hoping other projects fall into line, which will again change the variables, it is misguided in a very basic way.

The current change isn't to get cross-project-comparison, but to be consistent within SETI@home. Basically, it boils-down to:

1; SETI@home/BOINC was launched in June 2004, and was relying on the BOINC-benchmark. While the benchmark was "all over the plase", on average, the "gold computer" running 24/7 got 100 Cobblestones/day, while a computer doing 10x the work per day got 1000 Cobblestones/day and so on.
2; After some time, various optimizations was added to SETI@home, and some users used these optimized applications, doing significantly more work. Not sure how much faster, but let's say 4x faster, leading to a computer getting 1000 Cobblestones/day with default application getting 4000 Cobblestones/day.
3; SETI_Enhanced was in May 2006 released, and this incorporated most of the optimizations. It uses "flops-counting", instead of relying on BOINC-bencmark. Also, to give roughly same credit as if used BOINC-bencmark, a credit-multiplier was introduced, but, it was set a little too low, so AFAIK the "gold computer" was only getting 90 Cobblestones/day instead of 100, if ran the default application.
4; Anyone running optimized, that had for some months got 4x more credit than other users, suddenly only got 1.5x or something more, since optimized SETI_Enhanced wasn't 4x faster than standard SETI_Enhanced, but only 1.5x faster.
5; Some more optimizations was added to SETI_Enhanced in August 2007, and a new version was released. Due to further optimizations included in default application, the credit-multiplier was decreased, but, the optimizations was better than expected, so the "gold computer" is AFAIK getting 117.6 Cobblestones/day instead of 100 Cobblestones/day as it should be getting. (the 85% that has been mentioned)
6; Now, over 11 months later, SETI@home is finally trying to correct the "gold computer" again, so it on average should be getting 100 Cobblestones/day as it was getting back in 2004, 2005 and until May 2006, as long as it runs 24/7 and uses the standard application.
Other computers will get more/less than the "gold computer", depending how much more/less work they are doing compared to the "gold computer". This is true, regardless of standard or optimized application.


For anyone that has kept running standard applications since 2004, the "low"-period in 2006/2007 was followed by the "high"-period in 2007/2008, so the differences has more or less evened-out. So, being back to same credit as from June 2004 to May 2006 shouldn't be a problem.

As for optimized, since the optimized application aren't 4x faster than standard application as it was back in late 2005, start 2006, they're not getting 4x more credit any longer. This shouldn't really be a problem either...


Now, some of users running optimized applications seems it would be better if the optimizers had continued getting the same credit as before back in May 2006, instead of the majority running standard should keep the same credit as before... But, if this method had been used, the optimizers would basically have devaluated their own previously crunching...

Because, let's say 2 users had the exact same computer, both running 24/7, there user A ran standard application and user B optimized application. Let's say A got 1000 Cobblestones/day.

For 100 days, user A got 100k Cobblestones.
If B was 4x faster, he got 400k Cobblestones for same 100 days.

So, switch to SETI_Enhanced, there let's say standard had same speed as optimized had before, while new optimed was 1.5x faster...

After a further 100 days, you would have:
A; 200k total Cobblestones, if standard same credit as before.
B; 550k total Cobblestones, if standard same credit as before.

or:
A; 366.67k total Cobblestones, if optimized same credit as before...
B; 800k total Cobblestones, if optimized same credit as before.

If standard application same credit as before, it would mean B had 2.75x more credit than A after 200 days.
If optimized application same credit as before, it would mean B had only 2.18x more credit than A after 200 days...

To me, 2.75x more correctly points-out that B for some time was 4x faster than A, so is in my opinion a better method.

---

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

...
1; SETI@home/BOINC was launched in June 2004, and was relying on the BOINC-benchmark. While the benchmark was "all over the plase", on average, the "gold computer" running 24/7 got 100 Cobblestones/day, while a computer doing 10x the work per day got 1000 Cobblestones/day and so on.
2; After some time, various optimizations was added to SETI@home, and some users used these optimized applications, doing significantly more work. Not sure how much faster, but let's say 4x faster, leading to a computer getting 1000 Cobblestones/day with default application getting 4000 Cobblestones/day.

Your memory has failed a bit here. The faster optimized app reported less time so the Scheduler calculated claimed credits were proportionally less. The optimized app did produce more work for the project but still got credits based on the host benchmarks. The workaround was to use a special version of BOINC which Trux built which adjusted values reported by the host to approximate a work based credit rate.

                                                              Joe

I think I would rather talk about religion, politics or if O.J. was guilty or not than talk about credit reduction. BUT... a credit reduction does reduce the value of crunching for SETI vs. any other @home project in a team setting, like my Air Force team vs. Navy for example. I also recall the top of the BOINC food chain saying in other forums and posting here, SETI is the big gorilla and all other @home projects need to align their RAC to SETI. If SETI users did not use the AK-8 app. or any optimized app. for that matter, would a credit reduction be necessary? I am really convinced BOINC needs to get the other @home projects straightened out and credits more in line to the big gorilla before doing a credit reduction on SETI. As a side note, the wonderful job mslatter (Mark) done selling his time (and money) to the highest bidder being SETI USA, just apparently had the value of the efforts reduced.

The rate of credit does not depend on CPU speed or optimized applications.

A work unit that runs for 2 hours, 24 minutes on the "100 cobblestone" computer is worth exactly 10 cobblestones.

The same WU running in 3 minutes on a mega-go-fast CPU with Alex's V8 is worth exactly 10 cobblestones.

The 100 cobblestone computer would do 10 of these in a day. The optimized mega-go-fast would do 480, and be granted 4800 credits.

If the 10 cobblestone work unit was only getting 8 cobblestones, then an adjustment (upward) is in order. If the 10 cobblestone work unit is granted 12 cobblestones, then an adjustment is also in order. That is what Eric Korpela said in his post in the Staff Blog some months ago.

Either a cobblestone is a cobblestone, or it isn't.

I'm tempted to start "Arithmetic@Home." The work units would be exceedingly short (much like the "uppercase" work units in BOINC Alpha) and credit would be granted at 20 times the rate SETI does. Results would be discarded without review, since we know what 2 plus 2 is.

---

Your memory has failed a bit here. The faster optimized app reported less time so the Scheduler calculated claimed credits were proportionally less. The optimized app did produce more work for the project but still got credits based on the host benchmarks. The workaround was to use a special version of BOINC which Trux built which adjusted values reported by the host to approximate a work based credit rate.

Since at the time min_quorum was 3, even if optimizers did give a low claim, they often still got fairly "normal" credit. Also, users quickly started using "optimized" BOINC-clients, majority given even higher claims even paired with optimized seti-application than "normal" clients with un-optimized applications... So, while claims was "all over the place", granted credit as average per day was still much higher for optimized applications.

---

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

3 of 4 PCs are now set to NNT, 4th PC(PC2) is out of action due to a dead motherboard(Which I'd like to replace asap, But right now the P5K Deluxe is locking up once a day from heat I think), I may be moving to Milkyway as they support the OS, Of course I don't know If they're supporting the 64bit Boinc for Windows XP x64 or not. I'll find out in about 2 days or so.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

3 of 4 PCs are now set to NNT, 4th PC(PC2) is out of action due to a dead motherboard(Which I'd like to replace asap, But right now the P5K Deluxe is locking up once a day from heat I think), I may be moving to Milkyway as they support the OS, Of course I don't know If they're supporting the 64bit Boinc for Windows XP x64 or not. I'll find out in about 2 days or so.

Sorry you feel you have to do that, Joker. But don't forget, the Welcome mat will be waiting if / when you feel moved to come back.

F.

---

Ow! Darn third rail keeps on electrocuting me!

I think there are some misunderstandings about how this is going to affect credit. This doesn't specifically change credit for optimized applications, so don't worry about that.

So, here's the story... Right SETI@home faces two problems. One technical and one political. The technical problem is that in the SETI@home code is a value called the "LOAD_STORE_ADJUSTMENT" or also called the "credit multiplier". This value sets the ratio of the number of floating point operations in a workunit to the number you get credit for. That value is currently 2.85 for SETI@home 5.27, meaning for every floating point operation you do for SETI@home, you get credit for 2.85. This value was chosen so that the number of credits you would get from SETI@home (using the stock application) would be about the number that you would get if we used CPU time to calculate credit, like other projects do.

Every time we release an application we need to recalculate this value, because every time we compile the application it gets a little bit slower or a little bit faster. But since this value is in the code, you can't decide what value it should have until after you've compiled the code. For SETI@home 5.27 the value should have been about 2.4. For SETI@home 6, the value should be about 1.7, but I won't know what the value should be until after the application is released. For Astropulse, who knows.. So we've added a correction factor that can be calculated after the release. That factor is the 30 day moving average of median granted credit for the application to the median benchmark score.

The political problem is related to this. SETI@home has been granting about 15% more credit per cpu second than comparable projects. Other projects have threatened to increase their own credit multipliers to compensate. The problem is that they all have different ideas about how much credit we should be granting. One project has threatened to give 50% more credit per second than he benchmarks would indicate they should. So to avoid the coming credit war, BOINC is implementing this credit multiplier BOINC wide. This will be an objective way to make sure that projects don't grant too much credit. In other words, this will (probably) be happening at most every cpu intensive BOINC project.

So what does it mean for your credit. Well, long term it means that on average, SETI@home will probably be granting about 15% less credit per cpu second than it currently does. That factor will apply whether you use an optimized app or not. If you currently earn credit twice as fast as everyone else, you will continue to do so. It also means that if you get an Astropulse work unit, you will probably get about the same amount of credit per second as you would with a stock SETI@home. It also means that every other cpu intensive BOINC project will grant about the same amount of credit per cpu second.

It would be a wonderful world if we could base credit upon the actual number of floating point operations done, especially since our friendly neighborhood optimizers have made sure that SETI@home is the most highly optimized BOINC application around. SETI@home would then legitimately be able to claim more credit per CPU second than everyone else. But that won't fly with the rest of the BOINC projects, especially the ones that don't make source available.

I'll try to post some interesting graphs I've made later on.....

Anyway, here's the start of an FAQ...

Q. Is this multiplier based upon a theoretical "golden" machine?
A. No. This is based upon the credit per unit CPU time that would be granted to a middle of the pack machine. The definition of "middle of the pack" changes every day. Today the "middle of the pack" is a dual processor 3 GFLOP machine with 1GB of RAM, not a 386 machine with 32M of memory.

Q. Does this multiplier fix the credit for that machine to a certain credit per day (say 100)?
No. The credit per CPU second for the middle of the road machine is its floating point benchmark plus its integer benchmark times 5.787037037037e-13. That is exactly what it would be if we calculated BOINC credit based upon work unit run times. The middle of the road machine six months from now will get more credit per cpu second than today's middle of the road machine.

Q. Will the credit I get from my current machine go down as the middle of the road machine gets faster?
A. If you use a stock app, 6 months from now the machine you are using now will still get about the same number of credits per second that it will a month from now. For optimized apps, the rate of credit granted depends upon future release of SETI@home. (This is the case under the old fixed credit per FLOP system as well. The credit optimized apps get depends upon the credit multiplier in the stock app, which changes with each release.)

Q. Does this multiplier penalize fast machines?
A. No. Every machine is affected by this multiplier equally, whether it is fast or slow, whether it runs the stock app or an optimized app.

Q. Does this mean SETI@home will grant my machine fewer credits than ?
A. Probably not. SETI@home currently grants more credit than most other BOINC projects. This multiplier will level the playing field somewhat. When other projects start to use this multiplier, the playing field will be very level.

Q. I'm mad and I'm going to quit.
A. That's not really a question, but if you are one of the people who crunch to get the highest RAC, then you probably will try to find the BOINC project that gives you the most credit for the least work. You could start your own BOINC project that grants you as much credit as you want for no work at all. It's not very satisfying though.

---

@SETIEric@qoto.org (Mastodon)

Thanks for jumping in and explaining that Eric. Hopefully that will help some folks in their decisions to stay or leave. Hopefully everyone stays, but sometimes you just can't make everyone happy.

---

Thanks for jumping in and explaining that Eric. Hopefully that will help some folks in their decisions to stay or leave. Hopefully everyone stays, but sometimes you just can't make everyone happy.

Ain't that the truth!

From Eric's email:

So to avoid the coming credit war, BOINC is implementing this credit multiplier BOINC wide. This will be an objective way to make sure that projects don't grant too much credit. In other words, this will (probably) be happening at most every cpu intensive BOINC project.

So any project using Boinc is going to have their credit levels forced down by David Anderson?!?!

That seems really uncool to me. Other projects should be able to grant whatever credit they want to grant.

---

From Eric's email:

So to avoid the coming credit war, BOINC is implementing this credit multiplier BOINC wide. This will be an objective way to make sure that projects don't grant too much credit. In other words, this will (probably) be happening at most every cpu intensive BOINC project.

So any project using Boinc is going to have their credit levels forced down by David Anderson?!?!

That seems really uncool to me. Other projects should be able to grant whatever credit they want to grant.

So they should be able to offer 5x the credit of SETI@Home? What about credit inflation and fighting over users? Why shouldn't all project offer the same credits per hour?

From Eric's email:

So to avoid the coming credit war, BOINC is implementing this credit multiplier BOINC wide. This will be an objective way to make sure that projects don't grant too much credit. In other words, this will (probably) be happening at most every cpu intensive BOINC project.

So any project using Boinc is going to have their credit levels forced down by David Anderson?!?!

That seems really uncool to me. Other projects should be able to grant whatever credit they want to grant.

No matter how far they drive the credits down, I don't think it will solve our problems with inflation or Fedral/US budget deficit;(

---

When we finally figure it all out, all the rules will change and we can start all over again.

the credit of SETI@Home? What about credit inflation and fighting over users? Why shouldn't all project offer the same credits per hour?

How is it David Anderson's job to decide what other projects offer for credits? That should be up to the project admins themselves

---

the credit of SETI@Home? What about credit inflation and fighting over users? Why shouldn't all project offer the same credits per hour?

How is it David Anderson's job to decide what other projects offer for credits? That should be up to the project admins themselves

If it were left up to the project admins, what's to keep them all in line with some sort of standard? Further, if they have no problem offering the same or similar amounts of credit, why is David Anderson the bad guy for trying to make it easier for them by removing the burden of responsibility? Why get in a huff and complain about something that may not be worth fighting for? Even if the other admins do have a problem with this, why is forcing a level playing field such a bad thing? Why should project admins be allowed to decide how much credit to give? What good would that do in the interest of fairness?

SETI was the first Distributed computing project around. Some crunch it solely because of the credits and some for the benefit of the project itself. It shouldn't however be up to the scientist here to set the credit standard across the board for all projects even though the BOINC program was developed here. That should be at the sole discretion of each indiviual project.So what if somebody else wants to increase their granted credit by 50%.That's their decision. The ones that really want to crunch for SETI will remain at SETI.

Are the BOINC programs in a competition amongst themselves? Not for cobblestones but for users? Even if this is so the SETI program wins hands down. Leave the credits where they were. If another program raises theirs , so what! They must really need the extra help and are willing to pay extra for it. I don't believe SETI would suffer much from it.

---

Eric,

Thanks for the explanation of the whys and wherefores.

In all honesty it doesn't make it that much more palatable, it pretty much confirms my speculation.

It might have been a bit easier to swallow if it was explained prior to implementation.

Myself I'm going to run down my work units, leave and observe what happens both in the Seti project and BOINC wide, I may be back.

Why I'm I leaving?

It is really two fold, one the introduction of another app into Seti that being Astropulse, IMO it should be another project as in "astropulse@home" after all that is one of the advantages of the BOINC concept. I see its introduction into seti@home as an attempt to guarantee its success by forcing it down the throats of a captive audience at the expense of another balancing act.

The other sore point is the ever moving credit value and the cross-project cobblestone dream. Seti@home was ground breaking for distributed computing and still is an undeniable leader. IMO Seti@home was better off without BOINC although it is a worthy concept and has proven beneficial to many projects.

I think BOINC should merely be a common interface between projects, forget the inter-project comparisons and let the projects stand on their own merits with credit competition merely within each project. I see a lot of sideways looks at so called "credit whores" chasing the best credit paying across projects, but BOINC created the animal and the "problem". The multi-project teams could still exist, with standing within each project being the measure of comparison.

Just my thoughts and reasons,

Marcus

---

SETI was the first Distributed computing project around. Some crunch it solely because of the credits and some for the benefit of the project itself. It shouldn't however be up to the scientist here to set the credit standard across the board for all projects even though the BOINC program was developed here. That should be at the sole discretion of each indiviual project.So what if somebody else wants to increase their granted credit by 50%.That's their decision. The ones that really want to crunch for SETI will remain at SETI.

Why should it be their decision? Why does everyone seem to have a voice in this without actually running a project yourself? Why should one project be able to offer 50% more credit and turn projects against one another over users who are admitted credit hounds?

Are the BOINC programs in a competition amongst themselves? Not for cobblestones but for users? Leave the credits where they were. If another program raises theirs , so what! They must really need the extra help and are willing to pay extra for it. I don't believe SETI would suffer much from it.

Every project is going to feel they need help. So project A increases their "pay" by 50% to attract the admitted credit hounds, benefiting Project A with more CPU power, but this steals CPU power away from Project B. Project B feels their project is very important and are upset that Project A raised their credit and took some of its CPU power (users) away, so they in turn (in an attempt to recover those users and perhaps gain new ones) raise their credit 50% more than Project A and regains their CPU power from Project A and some from Project C. Now Project C is upset with losing CPU power and wants to increase their "pay".

Do you see the developing problem with this scenario?

Projects should be selected on their scientific merit and not for the credits they offer, but some users will follow the credits wherever the pay is highest. Many of those crunchers have very fast machines, are very competitive and want to be number 1 (or as close to it as possible). What if one of these competitive users belongs to a small project with very few users? What if they achieve number 1 status simply because there's not too many people on the project and they like the credit "pay" they receive. Now what if another project decides they want to attract more users so they increase their pay beyond this small project, so this user decides they're going to leave and join that project to get the higher paying credit? The small project loses their biggest contributor over what? Credit? It shouldn't be that way.

No. Project should not have to enter a competitive state for users based upon pay. There is nothing wrong with making all projects "pay" the same amount. It should not be up to the individual Admins - and they shouldn't want the burden of paying credit, they should be focussed on their science instead.

Why should it be their decision? Why does everyone seem to have a voice in this without actually running a project yourself? Why should one project be able to offer 50% more credit and turn projects against one another over users who are admitted credit hounds?

Does David run these other programs??? So then why should he decide how they run their projects?

---