And that's what I tried to point out: You only get credits in the hundreds 1.) if you are using a Boinc client that is not intended to be used with seti_enhanced and 2.) if you are paired up with another user who uses a Boinc client that is not intended to be used with seti_enhanced ...

At angle ranges between 0.226 and perhaps 0.275, the FPOPS based credit claim would be above 100. Work in that range is very rare, so I'm not surprised nobody has seen it yet.

                                                            Joe

... and if someone has one of these, they may not have finished it yet.

---

Here is a WU I did that got over 100 credits it had a low AR http://setiathome.berkeley.edu/workunit.php?wuid=81235973

Hello!

That´s what I meant...
In my answer:

http://setiathome.berkeley.edu/forum_thread.php?id=31761#333946


The other PCs have Boinc V4.45:
PC 1382918 and 1156377.

You must have V5.2.6 and later to become correct Credit...


Greetings!

---

Here is a WU I did that got over 100 credits it had a low AR http://setiathome.berkeley.edu/workunit.php?wuid=81235973

---

I reject your reality, and substitute my own.

SETI@home classic workunits: 5,000
SETI@home classic CPU time: 15,019 hours

On a side note, I really think people should be able to "buy" credits at 1 penny (USD) per individual credit. I really think that would surpass donations. And to even drive things more crazy, credits should be able to buy some things at an online gift shop!!! LOL

Now there's an Idea, LOL

---

D.

And that's what I tried to point out: You only get credits in the hundreds 1.) if you are using a Boinc client that is not intended to be used with seti_enhanced and 2.) if you are paired up with another user who uses a Boinc client that is not intended to be used with seti_enhanced ...

At angle ranges between 0.226 and perhaps 0.275, the FPOPS based credit claim would be above 100. Work in that range is very rare, so I'm not surprised nobody has seen it yet.

                                                            Joe

The benchmark * time method is the original one, and every new method has to be similar in regard of credits, at least with stock app'n'client.

"Similar" but not necessarily the same.

Remember that for some machines, the benchmark ran fast, and for others it ran slow, just depending on differences between the benchmark (artificial "make work" routine to estimate machine speed).

FPOPS is exact.

So, averaged across all of the machines out there, the scores should be the same between 4.x and 5.x. On individual machines, averaged over a bunch of work units, claimed credit on 5.x should be close to granted credit on 4.x.

---

Anyone here with a recent completed work unit with over 100.00 Claimed or Granted credit

Hello!

Tony said it here... that´s the reason why somebody becomes incorrect "Claimed Credit"


http://setiathome.berkeley.edu/forum_thread.php?id=31572#327425


And he said what Berkeley will do to eliminate this...


Greetings!


! You must read every thread ! :-)

---

Hi Saenger,

I don't see any difference between project vs. project or inner-project comparsion. It's all just mathematical calculations, that should deliver the same amount for the same number of them. At least for the stock apps & client setup.
If you increase the efficiency compared to the stock app, you should get the same credits as stock would have, even for a physically lesser effort (i.e. you get a bonus), but once stock is as efficient as optimized (and that's the goal of Eric and Co), the bonus has to stop.

Actually, the question behind my statements ("Is the new credit system fair compared to the old system?") was a rhetorical question ...

Exactly what you describe/demand has happened during the transition from seti 4.x to seti_enhanced 5.x. If you are comparing a 4.18 standard application with a 5.15 standard application on the same computer you *do* get similar credits. And as the new application is more efficient than the old one the "bonus" for those users who use optimized applications isn't as high as it was before the transition.

Look at this user: http://www.boincstats.com/stats/user_graph.php?pr=sah&id=888 (I randomly picked out this user because he has a similar world position like me). He does not use optimized applications at the moment. Most likely he didn't use optimized applications during the time of seti 4.x. You do not see any change in his credit production during the transition of seti to seti_enhanced. So, the only conclusion you can draw is: Yes, the new credit system is fair.

The benchmark * time method is the original one, and every new method has to be similar in regard of credits, at least with stock app'n'client.

I hope that my explanations helped to realize that this *is* the case ...

But again, a thread is being hijaced with "Is the new credit system fair?" discussions ... ;-) It has been discussed many times. So, please let's stop it.

Regards,
Carsten

If you increase the efficiency compared to the stock app, you should get the same credits as stock would have, even for a physically lesser effort (i.e. you get a bonus), but once stock is as efficient as optimized (and that's the goal of Eric and Co), the bonus has to stop.

Agreed - if at all possible, it would be great if already enhanced applications went out by default to *everyone* via BOINC anyway.
That's the goal we should ultimately work towards - no need for optimized clients because the default would be efficient enough.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

I know, there had been some difficulties with the new WUs, so that they spend useless cycles on the CPU. That has been counted as time in the old method, and imho rightful so. It's not the users fault that the app (or the scheduler or whatever) was sloppy programmed, the CPU was used, so give it credit.

I'd have to agree with the last part...

the CPU was used, so give it credit.

As computer time (aka CPU time) was used, credit should be at least used to compensate for time spent whether it's good/clean software or buggy infested software. Electricity has went up in many places in the USA. ;)

I could probably type up a 3 to 4 page report on the issue of "CPU time spent"... but I won't bore people. lol The worst case would be that I indirectly cause a mass exodus. rofl

The benchmark * time method is history here at SETI@home. If you want to determine which credit claim is the correct one at the moment you have to look at the results with fpops counting and not at those results still relying on the old method.

The benchmark * time method is the original one, and every new method has to be similar in regard of credits, at least with stock app'n'client.
We don't live in a Seti-only world, and even here it would be unfair to decrease the credit for the same effort over time.
I know, the old benchmark system didn't work well, it underscored especially on non-Win machines, and I've heard it was "good" for slow, and "bad" for fast crunchers, so it had to be optimized. Certainly the fpops-method is better, but to be comparable it has to be calibrated against the old method. You can't change the lenght of the meter to 1.5m, but keep the records of the long jump with the old values.

I know, there had been some difficulties with the new WUs, so that they spend useless cycles on the CPU. That has been counted as time in the old method, and imho rightful so. It's not the users fault that the app (or the scheduler or whatever) was sloppy programmed, the CPU was used, so give it credit.

Well, ruffle me feathers. Hmmm... I wonder if there's something I can post revolving around SETI@Home/Boinc where "Cruncher" does not get mentioned at all. LOL Just a scientific/psychological type of curiousity. ROFL!

Threat? I hope not :)
It only partially plays into that thread, by the way. That one's not about optimized vs. stock but project vs. project in credit granting per hour of cpu time. Eric has asked for data from people who crunch more than one project so he can try and see whether S@H generally claims different credits/hour than other projects.
[snip]

It seems to be a threat to some ;)

But on topic:
I don't see any difference between project vs. project or inner-project comparsion. It's all just mathematical calculations, that should deliver the same amount for the same number of them. At least for the stock apps & client setup.
If you increase the efficiency compared to the stock app, you should get the same credits as stock would have, even for a physically lesser effort (i.e. you get a bonus), but once stock is as efficient as optimized (and that's the goal of Eric and Co), the bonus has to stop.

Hi,

What's interesting imho is, that the claim by Crunch3rs 5.12 was by far the lowest compared with the stock apps of the other two.

It's exactly the other way round. The other two results are far the highest compared to the (nearly) correct claim of 58.87.

And that's what I tried to point out: You only get credits in the hundreds 1.) if you are using a Boinc client that is not intended to be used with seti_enhanced and 2.) if you are paired up with another user who uses a Boinc client that is not intended to be used with seti_enhanced ...

The benchmark * time method is history here at SETI@home. If you want to determine which credit claim is the correct one at the moment you have to look at the results with fpops counting and not at those results still relying on the old method.

If the new method is fair is another story, and we try to find out in the "Cross Project Credit Equalization and Adjustment" thread. At the moment, the result is that the credit claims will be raised by 5 % in future applications.

Regards,
Carsten

What's interesting imho is, that the claim by Crunch3rs 5.12 was by far the lowest compared with the stock apps of the other two.

That’s because the v4.x BOINC clients on the others ignored the Flop count and claimed by the old time-&-benchmarks method instead.

But that old benchmark is the measure for the fairness of the new count. Unless you inflated it by using "optimized" clients, it should deliver the same value for new and old one on stock apps. Imho that's what the Cross Project Credit Equalization and Adjustment threat is about.

Threat? I hope not :)
It only partially plays into that thread, by the way. That one's not about optimized vs. stock but project vs. project in credit granting per hour of cpu time. Eric has asked for data from people who crunch more than one project so he can try and see whether S@H generally claims different credits/hour than other projects.

The credit claiming methods of optimized clients have (for me) been at least as reliable as stock ones (i.e. claimed / granted having less to no delta).

Optimized clients do not inflate credit per WU, they simply crunch more data in the same time, therefore leading to higher credits/hour. If you check out the other thread, you will see Eric specifically asked for data from people running stock clients for that reason.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

What's interesting imho is, that the claim by Crunch3rs 5.12 was by far the lowest compared with the stock apps of the other two.

That’s because the v4.x BOINC clients on the others ignored the Flop count and claimed by the old time-&-benchmarks method instead.

But that old benchmark is the measure for the fairness of the new count. Unless you inflated it by using "optimized" clients, it should deliver the same value for new and old one on stock apps. Imho that's what the Cross Project Credit Equalization and Adjustment threat is about.

What's interesting imho is, that the claim by Crunch3rs 5.12 was by far the lowest compared with the stock apps of the other two.

That’s because the v4.x BOINC clients on the others ignored the Flop count and claimed by the old time-&-benchmarks method instead.

---

Hi,

It claimed 119.53 and got granted 94.88.
Link to Result stats

And if you look closer you will notice that you should get 58.87 for this result.

Regards,
Carsten

According to the old formula he should have claimed
(976.56 + 2244.42) * 51276.477571 / 1728000 = 95.5789981056942,
as he's using stock client and stock application.

What's interesting imho is, that the claim by Crunch3rs 5.12 was by far the lowest compared with the stock apps of the other two.

Hi,

It claimed 119.53 and got granted 94.88.
Link to Result stats

And if you look closer you will notice that you should get 58.87 for this result.

Regards,
Carsten

Yes, I have had my MacMini claiming too much for a specific long-running WU. It claimed 119.53 and got granted 94.88. This overclaim was due to me forgetting to update the BOINC client there, it was running with a 4.44 one still (i.e. old BOINC client, new SETI default cruncher).

Ran 51276 seconds, which would be 14 hours and 15 minutes, or thereabouts.

Link to Result stats

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information