Well i see some people have problems with the credit so i will put some info about that:
When your computer completes a result, BOINC determines an amount of claimed credit in one of two ways:
In general, the claimed credit is the result's CPU time multiplied by the CPU benchmarks as measured by the BOINC software. NOTE: the BOINC software is not optimized for specific processors. Its benchmark numbers may be lower than those produced by other programs.
Some applications determine claimed credit themselves, and report it to BOINC. This would be the case, for example, with applications that use graphics coprocessors or other non-CPU hardware.
Claimed credit is reported to a project when your computer communicates with its server. The granted credit that you receive may be different from the claimed credit, and there may be a delay of a few hours or days before it is granted. This is because some BOINC projects grant credit only after results have been validated.
Recent Average Credit
Projects maintains two counts of granted credit:
Total credit: The total number of Cobblestones performed and granted.
Recent average credit: The average number of Cobblestones per day granted recently. This average decreases by a factor of two every week, according to the algorithm given below.
Both quantities (total and recent average) are maintained for each user, host and team.
Credit
The project's server keeps track of how much work your computer has done; this is called credit. To ensure that credit is granted fairly, most BOINC projects work as follows:
Each work unit may be sent to several computers.
When a computer reports a result, it claims a certain amount of credit, based on how much CPU time was used.
When at least two results have been returned, the server compares them. If the results agree, then users are granted the smaller of the claimed credits.

I hope this help some people
GC


<img>
<A>Team Seti X Files[/url]
<A>Team Seti X Files Members[/url]

Do fast computers suffer a penalty with the way cobblestones are calculated? My fastest computer, which completes more WU/day than any of the other machines I use, has fewer credits granted to it than my slower machines which don't complete as many WUs.

I dont think the fasters PC suffer a penalty
GC

Team Seti X Files
Team Seti X Files Members

Sorry, but they are penalized:

From one of my previous posts:

Name:-------Whetones--Dhrystones---Cobblestones/Day--CPDN Work Perf.
P3 845 MHz--1033------1842.6-------190.34058---------11,078 Ts/Day
AMD 2200+---2119------3818---------809.0342----------24,570 Ts/Day
AMD 3000+---2678.1----4743.1------1270.46396---------27,809 Ts/Day

AMD 64 3K+--2499.1----4657--------1163.83------------38,987 Ts/Day

The 4th column is the interesting part. Cobblestones per Day. This is the amount of Credits each machine would get if it dedicated 24 solid CPU Hours.
(I know the BOINC 4.09 has reduced the figures, but I don't think the ratio is much changed. If You are interested I will check it.)

The 5th column is the work actually performed.

On my machines 28 TS equals 8 SETI-WU's give and take a grain.

Just a question for the Cobblestone-experts who correctly points out that no 2 WU's are equal, and therefore requires different Credits.
Do we have to 'invent' an entire new Credit system for that? I think not..

SetiSpy could tell You how many TeraFlops each WU needed, so why not just take the number of TeraFlops Required to complete a WU and give credit accordingly, and forget which machine actually does the processing?

ChrisD

> SetiSpy could tell You how many TeraFlops each WU needed, so why not just take
> the number of TeraFlops Required to complete a WU and give credit accordingly,
> and forget which machine actually does the processing?
>
> ChrisD

Yes, but, SETI Spy had to be "calibrated" for each and every processor. The details of that are off the operational page here and, in my experience was just as flukey and flaky as is an of the schemes used in BOINC over time.

There are any number of benchmarking papers out there to read and they all say the same things. It is an impossibility to craft a benchmark that will consistantly predict how fast a particular machine is when running its workload. They are good for a first order comparison and that is all.

Since credit is as usless as a screen door on a submarine (sorry, 20 years in the Navy does things to one's mind), there is no compelling reason that it has to be "accurate" whatever that means. To be honest, I wonder if losing credit is really going to cause that many people to leave projects. I know I did SETI@Home and it was cute to plot my contribution. But I was doing SAH CLassic because it was the only reliable distributed project I could find.

<p>
For BOINC Documentaion: Click Me!

> Sorry, but they are penalized:
>
> From one of my previous posts:
>
> Name:-------Whetones--Dhrystones---Cobblestones/Day--CPDN Work Perf.
> P3 845 MHz--1033------1842.6-------190.34058---------11,078 Ts/Day
> AMD 2200+---2119------3818---------809.0342----------24,570 Ts/Day
> AMD 3000+---2678.1----4743.1------1270.46396---------27,809 Ts/Day
>
> AMD 64 3K+--2499.1----4657--------1163.83------------38,987 Ts/Day
>

Uhm, according to this table, the A64 gets 6.11 times more credit than the p3, but does only 3.51 times as much work... In other words, the A64 gets 73% more credit than it should...

But looking on cobblestones/day, this doesn't make sence at all.
By re-doing the math, a more accurate table should AFAIK be:

P3 = 143.78 Cobblestones/day
amd2 = 296.85 CS/day
amd3 = 371.06 CS/day
A64 = 357.805 CS/day

The A64 is clearly not getting the correct benchmark, atleast not then compared to performance in CPDN. But the 3 other machines is within 10% of their actual performance, so there isn't a big problem here.

>
> SetiSpy could tell You how many TeraFlops each WU needed, so why not just take
> the number of TeraFlops Required to complete a WU and give credit accordingly,
> and forget which machine actually does the processing?
>

Just a slight problem here, that will you do with the 1-minute wu? ;)

Since the crediting in Seti is remove highest & lowest claimed and average the rest, if the variation for other machines except A64 & multi-cpu is below 10%, most of the variation is averaged out. Over time this will be even better, meaning a machine doing 2x the work as another machine will have roughly 2x as much credit.

Ingleside:

snip
> >
> > Name:-------Whetones--Dhrystones---Cobblestones/Day--CPDN Work Perf.
> > P3 845 MHz--1033------1842.6-------190.34058---------11,078 Ts/Day
> > AMD 2200+---2119------3818---------809.0342----------24,570 Ts/Day
> > AMD 3000+---2678.1----4743.1------1270.46396---------27,809 Ts/Day
> >
> > AMD 64 3K+--2499.1----4657--------1163.83------------38,987 Ts/Day
> >
>
> Uhm, according to this table, the A64 gets 6.11 times more credit than the p3,
> but does only 3.51 times as much work... In other words, the A64 gets 73% more
> credit than it should...

How about the AMD 3000+? :)

>
> But looking on cobblestones/day, this doesn't make sence at all.
> By re-doing the math, a more accurate table should AFAIK be:
>
> P3 = 143.78 Cobblestones/day
> amd2 = 296.85 CS/day
> amd3 = 371.06 CS/day
> A64 = 357.805 CS/day
>

Sorry, but You have to enlighten me on this one because:

snip from BOINC Docs.
----
Each project gives you credit for the computations your computers perform for it. BOINC's unit of credit, the Cobblestone 1, is 1/100 day of CPU time on a reference computer that does
1,000 double-precision MIPS based on the Whetstone benchmark.
1,000 VAX MIPS based on the Dhrystone benchmark.
----

So:

One Days worth of CPU time is 100*(Whetstones * Dhrystones)/(1,000 * 1,000)

or

A Computer doing just 1000 Whetstones and 1000 Dhrystones will earn 100 Cobblestones in 24 CPU Hours.

If the above assumption is correct, my table stands. Or where do I loose track??

ChrisD

> But I was doing SAH
> CLassic because it was the only reliable distributed project I could find.
>

Sorry, this is slightly off topic, but I found it interesting to see you say that Paul.

I felt the same way when I started doing S@H Classic. After doing distributed.net's RC5-64 project for years (and it was stable, reliable and well supported), I tried a myriad of other projects and ran into buggy, crashing clients, unreliable work, bad support, etc. etc. in nearly every project I tried that really interested me.

I might get lambasted a bit from the true SETI devotees here, but to be honest, I think the odds of S@H actually finding anything are (no pun intended) astronomical. Not because it's a flawed search, but because we're severely limited in our ablility conduct it with our present technology and knowledge. I'm a fairly serious amateur astronomer, and I suppose that once you've been slapped in the face good and hard with just how massive the universe is, you really get a feel for how difficult SETI may be. I simply ran the client because it intereseted me and it was a 'set it and forget it' level of stablility.

That's why the concept of BOINC excited me so much when I first heard about it (seems like it was nearly 2 years ago when the concept was first floated around). Being able to support a variety of scientific projects, from SETI to biology to weather to high-energy particle physics and hopefully soon even gravity waves is fabulous to me! :D

In any case, I realized that I can 'get behind' other BOINC projects far more than SETI and I'm happily backing my clients off to where I'll do a SETI WU every couple days (to keep a few fresh ones in the queue) and putting most of my effort into other projects.

Just wanted to make that comment... please resume your regularly scheduled topic of the credit system now. ;)

> > But I was doing SAH
> > CLassic because it was the only reliable distributed project I could
> find.
> >
>
> Sorry, this is slightly off topic, but I found it interesting to see you say
> that Paul.

I tried several all on separate computers so they would not step on each other. One was mersine (sp?) primes. Every time I came back to the machine it had stopped work ...

I think I also tried a cp.net but the fantastic run times then were just too much. With SETI@Home Classic running under SETI Spy with SETI Driver as a queue I could start them up and not look at them for weeks and all was well ...

So, that is where I stayed. I did a fair amount in Beta, but I never had problems so I was not much help. Health also played a part...

Now, I play on the systems when I can. Write as much as I can ... I was doing well till the last couple days, sigh ... the good news is that I am still getting in a couple to 12 hours worth of work ... well, real SLOW work ...

> I felt the same way when I started doing S@H Classic. After doing
> distributed.net's RC5-64 project for years (and it was stable, reliable and
> well supported), I tried a myriad of other projects and ran into buggy,
> crashing clients, unreliable work, bad support, etc. etc. in nearly every
> project I tried that really interested me.

That was my experience too ...

> I might get lambasted a bit from the true SETI devotees here, but to be
> honest, I think the odds of S@H actually finding anything are (no pun
> intended) astronomical. Not because it's a flawed search, but because we're
> severely limited in our ablility conduct it with our present technology and
> knowledge. I'm a fairly serious amateur astronomer, and I suppose that once
> you've been slapped in the face good and hard with just how massive the
> universe is, you really get a feel for how difficult SETI may be. I simply ran
> the client because it intereseted me and it was a 'set it and forget it' level
> of stablility.

It is almost a hopeless search. However, I like two things about the project. They have data, and now they are going to mine that data for new science. So, we took readings once, and now we are going to funnel the data through two sieves looking for what ever it is we will be looking for ...

> That's why the concept of BOINC excited me so much when I first heard about it
> (seems like it was nearly 2 years ago when the concept was first floated
> around). Being able to support a variety of scientific projects, from SETI to
> biology to weather to high-energy particle physics and hopefully soon even
> gravity waves is fabulous to me! :D

Yeah, me too ... like a kid in a candy store (well, it wasn't nautical this time ...)

If you have been following some of my posts you know I am doing all five because I was lucky enough to get signed up at the right times. And it is back to being fun again to look at what the computers are doing, what do the screen savers show (I am working on adding examples of the screen savers as I can right now ... though I am getting the international signal to quit ... :) I feel like I am going to fall out of my chair (for real) ... :)

> In any case, I realized that I can 'get behind' other BOINC projects far more
> than SETI and I'm happily backing my clients off to where I'll do a SETI WU
> every couple days (to keep a few fresh ones in the queue) and putting most of
> my effort into other projects.

I am kinda fascinated with the physics side of things which is why I like the LHC@Home and Einstein@Home (if they ever get it out of low gear). Predictor@Home, though I am not all that interested in the project is a nice one to have attached as they have good "filler" size work units.

> Just wanted to make that comment... please resume your regularly scheduled
> topic of the credit system now. ;)

Do we have to?
<p>
For BOINC Documentaion: Click Me!

> > Just wanted to make that comment... please resume your regularly
> scheduled
> > topic of the credit system now. ;)
>
> Do we have to?

At least in THIS post, yes! ;) We can have fun talking about gravity waves and high-energy particle physics somewhere else. :D

> So:
>
> One Days worth of CPU time is 100*(Whetstones * Dhrystones)/(1,000 * 1,000)
>
> or
>

For a day, the correct is: 100 * (Whetstone + Dhrystone) / (2 * 1000)

This is a linear formula, meaning a 486-33 MHz using 1000 hours on a wu and a 486-33 GHz using 1 hour on the same wu will claim the same credit. In this credit-system no machine has an advantage, except that the fast machine of course crunches 1000 times as much, and therefore gets 1000 times as much credit. :)

> For a day, the correct is: 100 * (Whetstone + Dhrystone) / (2 * 1000)
>
> This is a linear formula, meaning a 486-33 MHz using 1000 hours on a wu and a
> 486-33 GHz using 1 hour on the same wu will claim the same credit. In this
> credit-system no machine has an advantage, except that the fast machine of
> course crunches 1000 times as much, and therefore gets 1000 times as much
> credit. :)

Ingleside.

Thank You for telling me where I lost track. :)

This misunderstanding from my side may be due to the fact that English is not my native tounge, or the fact that the explanation from the Boinc Crew is not clear enough.

Maybe the Boinc Folks should consider using a few of their precious minutes to clarify this formula, so no ambiguity is present.

It would certainly have reduced my fustration quite a bit.

Thanks again.

ChrisD

> This misunderstanding from my side may be due to the fact that English is not
> my native tounge

Ironic isn't it that Ingleside is Norwegian, you're Danish and you're both having to work this stuff out in English. :P

You both have my respect though, since your English completely dwarfs my grasp of either Norwegian or Danish! :)

BTW, interesting reading your progress. I learned a bit from it!