Eric added the adjustment mechanism to BOINC with changeset [trac]changeset:15661[/trac]. In a post to the boinc_dev mailing list thread on cross-project credits he said that after 2.8 days it had adjusted S@H credit multiplier from 1.0 down to 0.978. It keeps a 30 day history of the multiplier. Credit claims are calculated based on when the work was "sent" rather than when it is returned, so in most cases the two hosts with initial replication tasks will be claiming using the same multiplier.

The method does involve benchmarks, but uses statistics over the last 10000 results so individual benchmarks will have negligible effect. The goal is that overall project granted credits should be equivalent to Cobblestones.

Eric thinks the multiplier would get down to around 0.85 after 30 days with current apps, though he of course hopes to release the 6.0x app sooner than that. 6.02 is not much different in speed from 5.27 so the trend will probably not change much.

                                                                Joe

So the benefit here is that this will compare credit vs. time for Astropulse and Multibeam, and slowly adjust so that the two science applications will gain the same credit per hour -- on average across a whole stack of machines.

Good idea.

---

I say quit It or We leave. I'd rather go fold than to have the credit multiplier keep SINKING. As pretty soon there won't be any credits or a multiplier, I say We want 1.0 and nothing less, Or I with hold the data.

Actually, I'd be very much in favor of 1.0 instead of the 2.85 multiplier we have now.

---

I would still stump for a new stat.......let's call it temporary RAC or instant RAC.......
Would not replace the current stat, but would show a bit more current state of affairs......maybe a one week rolling average or even a 24 hour rolling average as opposed to 4 weeks......
Would be pretty nifty for us RAC obsessed folks.

And could also be very depressing if your machine is having problems. Ohhhh, that is getting similar to crunch3rs remark in the 'what is your longest uptime' thread ;-)

I agree that it is sometimes frustrating that some knifty new change to your computer doesn't show up immediately, especially because of all the trouble you have gone to to get the stupid thing to speed up. But it works both ways I guess. Some of my most silly mistakes were conveniently masked by the dampening effect of long-term rac calculation :-)

Regards,
John.

This project "benchmark" over the the last 10,000 units is going to be skewed by optimized apps and we are to be expecting a 15% drop in credit. I predict cross project BOINC crunchers will be jumping ship all over again.

Myself I only crunch seti@home and thought the project had reached some level of stability, but once again past work credits will not be comparable.

I too will wait to see how all this "pans out", but so far it looks to offer only disadvantages to participants.


UncleVom

---

I say quit It or We leave. I'd rather go fold than to have the credit multiplier keep SINKING. As pretty soon there won't be any credits or a multiplier, I say We want 1.0 and nothing less, Or I with hold the data.

Actually, I'd be very much in favor of 1.0 instead of the 2.85 multiplier we have now.

Then where did this 1.0 vs 0.85 come from and why would We readjust the credit multiplier down so drastically? To align again? That was their excuse last time, The slope is Slippery now, Seti is off here.

With all due respect, are we really that upset over something that is worth absolutely nothing?

Last time I checked, a million SETI credits wasn't worth a penny.

Maybe they should just give 100 billion credits per WU (of any length) and be done with it.

---

Are you messing with the kitties' credits??????
Eric....you should be ashamed...really now.
Is this for real or what?

Don't mess with the kitties......LOL.

---

"Time is simply the mechanism that keeps everything from happening all at once."

With all due respect, are we really that upset over something that is worth absolutely nothing?

Last time I checked, a million SETI credits wasn't worth a penny.

Maybe they should just give 100 billion credits per WU (of any length) and be done with it.

Credits are quite obviously worth something, they seem to be a common currency around here.

In recent history I remember a cyber-local cat owner auctioned of his kitties' output to the highest bidder.

People invest thousands of dollars to improve their hardware, not to mention the time and utilities involved, for science, for credits or both.

The science is probably of value even if spacemen are never found, but the credit system is the measurement of value and contribution amongst many of us, it has always been that way. Heck you can even get a certificate with the picture of the signature of David P. Anderson.

Devaluing the credit devalues the contribution, seems simple enough to me.


Marcus

---

With all due respect, are we really that upset over something that is worth absolutely nothing?

Last time I checked, a million SETI credits wasn't worth a penny.

Maybe they should just give 100 billion credits per WU (of any length) and be done with it.

Like you said here ( Message ID 784829) Ned, "Dennis, please don't take my posts too seriously -- I don't." Many of us put a lot of $$ and time in our rigs, squeezing a bit higher RAC. Decreasing the RAC devalues the $$ and efforts. Also, while I do not speak in any way for Alex, JD, Eric, Josef and all the awesome code makers that made the AK-8 app. decreasing the RAC seems to me to also be taking away from their efforts in so far as comparing RAC to other projects. The humanity in us begs for challenges, competition, and results, and fairness. Not many people like it when things are taken away, and when $$ and time are perceived to be diminished, negative emotions are an expected response. I am at a loss on my 1 year anniversary date as to what my future with SETI will be; my instinct is to say "ditto Joker, UncleVom and Mark".

This is quite concerning to me. I have asked Eric Korpela for a clear and concise confirmation answer---yes/no and if yes, why.

---

With all due respect, are we really that upset over something that is worth absolutely nothing?

Last time I checked, a million SETI credits wasn't worth a penny.

Maybe they should just give 100 billion credits per WU (of any length) and be done with it.

Credits are quite obviously worth something, they seem to be a common currency around here.

In recent history I remember a cyber-local cat owner auctioned of his kitties' output to the highest bidder.

People invest thousands of dollars to improve their hardware, not to mention the time and utilities involved, for science, for credits or both.

The science is probably of value even if spacemen are never found, but the credit system is the measurement of value and contribution amongst many of us, it has always been that way. Heck you can even get a certificate with the picture of the signature of David P. Anderson.

Devaluing the credit devalues the contribution, seems simple enough to me.


Marcus

Marcus,

Here's the thing. I have the exact same problem with "inflation" that you have with "devaluation."

I ultimately want a work unit to give the same number of credits no matter who does what. I want a work unit done today to be worth the same as a work unit done three years ago.

Trouble is, it just ain't that simple.

The concept is simple: a specific machine with specific characteristics is supposed to generate 100 cobblestones per day.

My understanding is that the 100-cobblestone computer would be generating closer to 280 cobblestones per day -- and would have been all along.

So we can't really get back to the 100 cobblestone standard.

Trouble is, if SETI can't adjust up and down, if they're only allowed to adjust upwards, then Cobblestones are an inflationary currency.

We all know what that does in real life: I'm making more in my 50's than I was in my 20's, but when I was 20 I could fill my gas tank for less than $10.

So, I'm advocating holding the line against inflation. I want a cobblestone today to be the same as a cobblestone from December 2005.

Make sense?

-- Ned

---

With all due respect, are we really that upset over something that is worth absolutely nothing?

Last time I checked, a million SETI credits wasn't worth a penny.

Maybe they should just give 100 billion credits per WU (of any length) and be done with it.

Like you said here ( Message ID 784829) Ned, "Dennis, please don't take my posts too seriously -- I don't." Many of us put a lot of $$ and time in our rigs, squeezing a bit higher RAC. Decreasing the RAC devalues the $$ and efforts. Also, while I do not speak in any way for Alex, JD, Eric, Josef and all the awesome code makers that made the AK-8 app. decreasing the RAC seems to me to also be taking away from their efforts in so far as comparing RAC to other projects. The humanity in us begs for challenges, competition, and results, and fairness. Not many people like it when things are taken away, and when $$ and time are perceived to be diminished, negative emotions are an expected response. I am at a loss on my 1 year anniversary date as to what my future with SETI will be; my instinct is to say "ditto Joker, UncleVom and Mark".

Dennis,

I would argue that allowing the rate of credit to rise actually decreases the value of your investment.

Why?

Let us take a hypothetical.

You've worked hard to improve your RAC. You've put time into optimized applications, and you've put money into new computers. Some have actually built machines for the sole purpose of producing "waste" clock cycles to give to SETI.

I don't spend money on SETI. If I do, it's in the form of a cash donation.

If your RAC is increasing with all this work, and my RAC is increasing because of inflation, then it is harder to know just what all of your hard work is worth.

My RAC at the moment is 155. Your RAC is 47,610. If credit is reduced by 10%, then my RAC will drop to 140, and yours will drop to 43,849.

Your RAC is still more than 300 times mine.

I looked in the BOINC TRAC to see what the change did. It does not differentiate between optimized and standard -- there are waves in the credit "sea" and we all rise and fall with them.

In the case of Mark and the kitties, the raw numbers may change, but the rankings won't change a bit. He's #46 on the rankings now, and unless there is a huge change in how credit is granted, he'll be somewhere around #46 when everything settles out.

If they decided tomorrow to raise credits 10%, I'd be unhappy because it makes it harder to measure today against past contributions, but I'd still crunch.

Now, if they somehow decided to identify a certain class of users (those whose computers are dedicated only to SETI, or those who run Optimized applications) then I could see a reason for all of the unhappiness.

... but when the tide comes and goes, we all float up and down together.

-- Ned

---

So, I'm thinking about this.....

Let's say that credit was "inflating" by 20% per year.

This year, you crank out 500,000 credits.

Next year, you crunch 600,000.

Next year, someone comes along, duplicates your farm exactly, just two of everything you have, and they'd crunch 1,200,000 credits compared to your 1,100,000.

Which is actually worth more, their one-year sprint, or your two years of dedication?

Let's say credit was deflating by 20% a year.

This year, you crank out 500,000 credits.

Next year, you get 400,000, for a total two year contribution of 900,000.

The same mega-cruncher comes along next year, duplicates your farm x2, and he gets 800,000 for the same number of FLOPs. Is he entitled to less for his efforts?

That's why I think the project needs to be able to adjust credit up or down to keep parity -- so that everything else being equal, a cruncher on the stock application should get pretty close to the exact same amount of credit per year.

Those on optimized will likely always get more than those on standard, and that's fine, but the advantage is not going to be constant -- it'll be better sometimes, not so much at other times.

---

Well...the kitties have reflected on things a bit and have come to the following conclusion........

It just doesn't matter.

I am not in this just for the credits shown under my name.....if I were, I would credit shop for the best credit/hour that I could find....and we all know there are some projects that issue credits like candy.

Eric....you do what you gotta do. This project has survived worse storms than this...and will continue to do so. I have faith in you to guide the project where it needs to go.

The kitties and I have had some stormy weather here and there, but I am still here, am I not?

It really does NOT matter if the credits are cut in half.....it's all subjective really....if my 40 some k RAC turned to 20k, what's the diff? All others on this project are given the same set of rules, and I can still compare what I do to anybody else on the same level playing field....my position among other participants will not change.....

So....
Take this from your A#1 Seti fanboy.......

Folks....forget about the smoke and mirrors...pay no attention to the man behind the curtain....and worry ye not about credit adjustments....

IT'S ABOUT THE SCIENCE..............

Glad I got that off my chest....the kitties shall sleep well tonight.
Now if I can just get a new hard drive and get my 8th rig crunching again...........

---

"Time is simply the mechanism that keeps everything from happening all at once."

If the guy has more PCs than you or Me, So what?(I would've gotten punished for saying that by My dad) There's nothing wrong with buying and possibly(or not) overclocking a PC for crunching, Don't like It, Too bad, So sad, My older brother would say(US Navy Retired 20yrs, Deceased-cancer), Bender also had a saying, But I not go there, Although I'm sorely tempted to. It's My money and I'll spend It how I like, If You can't keep up or don't want to then I can't do anything about that, I help the project by making up for those that login once and do a few units and then do nothing, The Seti Deadwood, How many users out there that never come to the forums do that or even contribute money to the project? I've done so in the past, If I could've I would've this Year too, But My house comes first for a while as I'm on a fixed income. Lowering the Multiplier says the project doesn't value My help as much and yet wants the same amount of work done and expects Me to keep paying about $150.00 a month for Electrical costs, If the Multiplier does go down I'll take My toys and any possible money elsewhere.

Joker....chill just a bit and think about it.....
My finances are in the tank because I am obsessed with this project. OCD to the max. And that's OCD, not overclocked. Oh well......I can live with that.

And I try to share what I can with those that I care about.......
400 pounds of kitty food to the shelter, a few sheckels to Seti....
And more advice shared with you fellow Setizens than some might like.

Stay here with me, come what may.....the amount of work that you and I contribute shall not change.......

As Chris Rea said in his song 'Road to H***'......it's all just bits of paper, flying away from you..........

---

"Time is simply the mechanism that keeps everything from happening all at once."

And all this time I was thinking of the other song, The Highway to H*** written by Bon Scott, Angus Young and Malcolm Young(AC/DC fame). ;)

Ahh....those were the days, eh? The band was never the same after Bon passed. Even thought they continued to have commercial success, the soul had died with Bon.

But......we be getting off topic now.

What you say to sticking around bro?

---

"Time is simply the mechanism that keeps everything from happening all at once."

So, I'm thinking about this.....

Let's say that credit was "inflating" by 20% per year.

This year, you crank out 500,000 credits.

Next year, you crunch 600,000.

Next year, someone comes along, duplicates your farm exactly, just two of everything you have, and they'd crunch 1,200,000 credits compared to your 1,100,000.

Which is actually worth more, their one-year sprint, or your two years of dedication?

Let's say credit was deflating by 20% a year.

This year, you crank out 500,000 credits.

Next year, you get 400,000, for a total two year contribution of 900,000.

The same mega-cruncher comes along next year, duplicates your farm x2, and he gets 800,000 for the same number of FLOPs. Is he entitled to less for his efforts?

That's why I think the project needs to be able to adjust credit up or down to keep parity -- so that everything else being equal, a cruncher on the stock application should get pretty close to the exact same amount of credit per year.

Those on optimized will likely always get more than those on standard, and that's fine, but the advantage is not going to be constant -- it'll be better sometimes, not so much at other times.

If the guy has more PCs than you or Me, So what?(I would've gotten punished for saying that by My dad) There's nothing wrong with buying and possibly(or not) overclocking a PC for crunching, Don't like It, Too bad, So sad, My older brother would say(US Navy Retired 20yrs, Deceased-cancer), Bender also had a saying(KMSMA). It's My money and I'll spend It how I like, If You can't keep up or don't want to then I can't do anything about that, I help the project by making up for those that login once and do a few units and then do nothing, The Seti Deadwood, How many users out there that never come to the forums do that or even contribute money to the project? I've done so in the past, If I could've I would've this Year too, But My house comes first for a while as I'm on a fixed income. Lowering the Multiplier says the project doesn't value My help as much and yet wants the same amount of work done and expects Me to keep paying about $150.00 a month for Electrical costs, If the Multiplier does go down I'll take My toys and any possible money elsewhere.

The essential point in my message is "same number of floating point operations, different credit scores" no matter which way things are adjusted.

My position is "same number of flops, same credit" regardless of your level of dedication or longevity.

---

Ned,

I'm sorry I don't see where the inflation is, other than it is cheaper and easier to get credits with today's hardware and apps.

My K6-2 350mhz capable of a RAC of about 15 and is parked for now, my q6600 which cost less, even not counting monetary inflation, with an optimized app runs a RAC of over 6000 and would probably be sitting around 3500 with the stock app. If that's your inflation yes it exists, its called progress.

There is no question that the Whetstone/Dhrystone balance of the mythical 100 cobblestone computer is a little skewed based on the performance balance inherent in most current xx86 computers.

Looking at Whetstone benchmarks on line it doesn't seem to me to be that far out of whack for seti@home with a stock app.

You said to Dennis

"If they decided tomorrow to raise credits 10%, I'd be unhappy because it makes it harder to measure today against past contributions, but I'd still crunch."

Well we're allegedly currently looking a variable credit value with a projected drop of 15% within 30 days. How easy will that measure up against past contributions? As for whether I will still crunch it remains to be seen,

I have been pretty loyal to the Seti@home project with over 8 1/2 years in, but right now there are a few things holding me back from clicking off, the team to which I belong, finding a replacement project that does not disadvantage linux boxes and still does some good or maybe just saying heck with it and remembering that those "free" CPU cycles are not free.


Marcus

---

Well.......if you folks do not agree with me, that is fine.
But I will tell you one thing....

When the sun rises and the kitties awake, I shall still be here.

Hope you all can see past your differences and stick around too.

---

"Time is simply the mechanism that keeps everything from happening all at once."

...
I ultimately want a work unit to give the same number of credits no matter who does what. I want a work unit done today to be worth the same as a work unit done three years ago.

Trouble is, it just ain't that simple.

The concept is simple: a specific machine with specific characteristics is supposed to generate 100 cobblestones per day.

My understanding is that the 100-cobblestone computer would be generating closer to 280 cobblestones per day -- and would have been all along.
...
-- Ned

Eric's observation was that the 2.85 multiplier used by the application expresses the amount by which the flop counting needs to be adjusted to account for loading and storing data. The Whetstone benchmark runs almost totally in L1 cache on modern hardware so measures only pure computational capability.

The 100 Cobblestone computer was supposed to have 1000 Whetstone MIPS and 1000 Dhrystone MIPS. My P4 system most recent benchmarks are 814 and 1599 so it's approximately in the same range. It did approach 200 RAC here with the AK_V8 SSE2 Windows app, with stock it might have been near 130 or so based on standalone comparative testing. It's working SETI Beta now, there have been too many changes to allow RAC to stabilize, but is at about 73. It's hardly a typical system, but still tends to indicate the flop counting method is producing credits not hugely different overall than pure Cobblestones.

=========================

General observations for the thread:

Months ago Eric noted, probably in one of his Staff Blog posts, that he considered the credit rate about 15% higher here than it should be. If the 5.27 apps had the capability to respond to credit rate adjustment through the WU header, I'm sure he would have made a reduction then. Those who are bemoaning the adjustment now should keep in mind that it's been considerably delayed.

The new adjustment method is an option in BOINC which Eric hopes all CPU intensive projects will adopt once it has been demonstrated here. If they do, it will be a useful step toward keeping a sensible balance. There will still be the issue of how to value other kinds of contributions, though.

In volunteer efforts, changes always risk some participants opting out. That's inevitable since each volunteer has unique reasons for participating.

                                                                Joe

...
I ultimately want a work unit to give the same number of credits no matter who does what. I want a work unit done today to be worth the same as a work unit done three years ago.

Trouble is, it just ain't that simple.

The concept is simple: a specific machine with specific characteristics is supposed to generate 100 cobblestones per day.

My understanding is that the 100-cobblestone computer would be generating closer to 280 cobblestones per day -- and would have been all along.
...
-- Ned

Eric's observation was that the 2.85 multiplier used by the application expresses the amount by which the flop counting needs to be adjusted to account for loading and storing data. The Whetstone benchmark runs almost totally in L1 cache on modern hardware so measures only pure computational capability.

The 100 Cobblestone computer was supposed to have 1000 Whetstone MIPS and 1000 Dhrystone MIPS. My P4 system most recent benchmarks are 814 and 1599 so it's approximately in the same range. It did approach 200 RAC here with the AK_V8 SSE2 Windows app, with stock it might have been near 130 or so based on standalone comparative testing. It's working SETI Beta now, there have been too many changes to allow RAC to stabilize, but is at about 73. It's hardly a typical system, but still tends to indicate the flop counting method is producing credits not hugely different overall than pure Cobblestones.

=========================

General observations for the thread:

Months ago Eric noted, probably in one of his Staff Blog posts, that he considered the credit rate about 15% higher here than it should be. If the 5.27 apps had the capability to respond to credit rate adjustment through the WU header, I'm sure he would have made a reduction then. Those who are bemoaning the adjustment now should keep in mind that it's been considerably delayed.

The new adjustment method is an option in BOINC which Eric hopes all CPU intensive projects will adopt once it has been demonstrated here. If they do, it will be a useful step toward keeping a sensible balance. There will still be the issue of how to value other kinds of contributions, though.

In volunteer efforts, changes always risk some participants opting out. That's inevitable since each volunteer has unique reasons for participating.

                                                                Joe

Good observations, Joe. As usual, you are one of the most level-headed among us. You always seem to be able to cut to the quick and avoid the emotionalism that goes with some of these issues.....

---

"Time is simply the mechanism that keeps everything from happening all at once."