I don't understand....

wijata.com
wijata.com
Joined: 11 Feb 05
Posts: 113
Credit: 25495895
RAC: 0

I did unlocked viewing of my

I did unlocked viewing of my CPUs.
If somebody cares to look at it...
These I was refering to (added same time about):
http://einsteinathome.org/host/51601
http://einsteinathome.org/host/51689
http://einsteinathome.org/host/51568

And maybe somebody will try explain me what is RAC, like to child, with simply words.
Thanx for patient ;)

Paul D. Buck
Paul D. Buck
Joined: 17 Jan 05
Posts: 754
Credit: 5385205
RAC: 0

> I did unlocked viewing of

Message 6804 in response to message 6803

> I did unlocked viewing of my CPUs.
> If somebody cares to look at it...
> These I was refering to (added same time about):
> http://einsteinathome.org/host/51601
> http://einsteinathome.org/host/51689
> http://einsteinathome.org/host/51568
>
> And maybe somebody will try explain me what is RAC, like to child, with simply
> words.

I thought I had ... :)

> Thanx for patient ;)

why not ...

Ok, first... of the three computers you listed, one is a 2.8 GHz machine, other is 2.0 GHz, and the last is REAL old ...

So, even if they started on the same day. RAC is computed for each computer independently of all other computers.

So, as you return results, and those results validate, and finally when the quorum is established credit is granted. At that point the RAC for that computer is calculated.

With those machines you are doing work, each at its own speed and the results are returned by each machine and each machine has its own RAC.

Like I said, think of it as a rate. If you drive to the store and it took you 30 minutes and it was 30 miles away, you did an average of one mile a minute ...

The Gas Giant
The Gas Giant
Joined: 18 Jan 05
Posts: 72
Credit: 3109569
RAC: 0

Why are we interested in RAC

Why are we interested in RAC v's total credit?

During the days of BOINC this was a debate that was had when the way the credits were issued came up. It was explained to me sort of like this....

"...the projects only really care about how much work is being crunched now. Not last week, last month or last year. They want the available work completed as quickly as possible."

Hence the idea of RAC was born. This was sort of being used in seti classic with some stats sites calculating the number of results returned per day, which was then used as bragging rights by the users. In reality credit is just about bragging rights anyway.

Paul

wijata.com
wijata.com
Joined: 11 Feb 05
Posts: 113
Credit: 25495895
RAC: 0

> > I did unlocked viewing of

Message 6806 in response to message 6804

> > I did unlocked viewing of my CPUs.
> > If somebody cares to look at it...
> > These I was refering to (added same time about):
> > http://einsteinathome.org/host/51601
> > http://einsteinathome.org/host/51689
> > http://einsteinathome.org/host/51568
> Ok, first... of the three computers you listed, one is a 2.8 GHz machine,
> other is 2.0 GHz, and the last is REAL old ...

But still , they were joined same time (28th feb). And the real old has contributed biggest amount (6wu, 505credits, 82rac), the 2Ghz (5wu, 466c, 121rac), the 2.8Ghz (4wu, 390c, 218rac). And there is also machine joined 1st march(7wu, 586c, 115rac - http://einsteinathome.org/host/53378)

So I still don't understand, why machines that are faster delivering WU and obtaining credits have lower RAC. Is there any relation between credits obtained recently (28th feb i consider recently), and RAC ?
Ech, maybe I'm not supposed understand...

Paul D. Buck
Paul D. Buck
Joined: 17 Jan 05
Posts: 754
Credit: 5385205
RAC: 0

> But still , they were

Message 6807 in response to message 6806

> But still , they were joined same time (28th feb). And the real old has
> contributed biggest amount (6wu, 505credits, 82rac), the 2Ghz (5wu, 466c,
> 121rac), the 2.8Ghz (4wu, 390c, 218rac). And there is also machine joined 1st
> march(7wu, 586c, 115rac -
> http://einsteinathome.org/host/53378)
>
> So I still don't understand, why machines that are faster delivering WU and
> obtaining credits have lower RAC. Is there any relation between credits
> obtained recently (28th feb i consider recently), and RAC ?
> Ech, maybe I'm not supposed understand...

Ok, maybe we are getting some where after all. Ok, part one; RAC is an average over time. Because of that, there are cases, especially when you first start up, where the RAC numbers do not match reality well.

In the Glossary where we discuss RAC, there is this statement:

[pre]Right after joining a project and getting the first credits granted are working against a very short time, and slightly odd things can happen to the numbers. The odd numbers for RAC at the beginning of a users tenure at a project are strictly an artifact of the equation used to generate them and are the result of the extremely short time span and the lack of previous data.[/pre]

And this is what you are seeing, in part.

The other aspect is that the faster computers were added later, so, their production is faster and will/should create a higher RAC. And this is what I see with the systems.

Early on you said that they were all added at the same time, and were identical in speed. Yet, they are not the same speed, and they all did not start at the same time. So, the RAC will be different.

In theory RAC should be looking at about a 2 week window of time. But this is also dependent on delivering and obtaining credit from updates to the web site. If you stop a computer and do not let it return work its RAC will "freeze" at the last value and will not change ...

wijata.com
wijata.com
Joined: 11 Feb 05
Posts: 113
Credit: 25495895
RAC: 0

Well, I was just expecting,

Well, I was just expecting, that computer which is earning more credits in the same recent time than other, will have bigger RAC. Apparently this is not true.
That's it.

wijata.com
wijata.com
Joined: 11 Feb 05
Posts: 113
Credit: 25495895
RAC: 0

And again,

And again, compare
http://einsteinathome.org/host/51601
vs
http://einsteinathome.org/host/53378
fisrt joined 28 Feb has 490 total credits and 209 rac
second joined 1 Mar has 1360 credits and ONLY 162 rac.
It's obvious that the second machine is much faster, delivers more credits within shorter times. Hence it's rating in RAC is lover. Is it fair? Is RAC showing anything?
I don't understand and don't trust RAC.

Paul D. Buck
Paul D. Buck
Joined: 17 Jan 05
Posts: 754
Credit: 5385205
RAC: 0

> fisrt joined 28 Feb has 490

Message 6810 in response to message 6809

> fisrt joined 28 Feb has 490 total credits and 209 rac
> second joined 1 Mar has 1360 credits and ONLY 162 rac.
> It's obvious that the second machine is much faster, delivers more credits
> within shorter times. Hence it's rating in RAC is lover. Is it fair? Is RAC
> showing anything?

Yes, it is just not showing you something in a way that you think it should be showing you something.

RAC is an decaying average over time. It is calculated when results are validated. So, the older computer has not had a result validaded (at the time of this writing) since 2 March.

The other computer has results validated as soon as yesterday, 5 March. Because the time between results being validated, and the rate of validation is different, the RAC is different.

RAC is an average over time. Your mental model is that it is a speed. So you want to read it as a speedometer. It is not a speedometer. It does not give you an instantaneous indication that you can compare two machines at a point in time and say, this one is faster because its RAC is higher.

I will say it again, RAC is calculated over time. It uses a decaying function so that results more distant in the past have less influence over the current value. In time it will take on more of the characteristics of a general indication of the rate at which you are earning credit. BUt that is all that it is. A general indication. Changes in the rate in which you return results will, in time, change the RAC.

Short term calculations of RAC, especially over small data sets will have a poorer, and usually (but not always) a higher fluctuation in calculated values than those performed over longer periods of time and larger data sets.

The one computer, though it has returned about the same number of results, has only 4 validated results. So, RAC is calculate on ONLY those results. Since there has been none validated for nearly 4 days this RAC value also will not change.

> I don't understand and don't trust RAC.

No doubt about it. RAC is hard to understand. But that is why there is such a long entry in the Glossary about RAC. I did not include the actual computation formulas in that I did not feel that they would be useful.

Maybe I should see if I can come up with a way to make it so that you (and others) can better see how RAC works.

Part of the problem is that (as I understood you to say) English is not your primary language. But, it also seems to me that you are not being careful in reading what is being said to explain this to you. Please do not take offense, you asked us to be patient with you, and in turn you have to be patient with us too! :)

Paul D. Buck
Paul D. Buck
Joined: 17 Jan 05
Posts: 754
Credit: 5385205
RAC: 0

Ok, I made an example from

Ok,

I made an example from the data for your computer id 51601, where I took the data off of the page and entered it in and used the functions to calculate RAC.

Making this more interesting was the fact as I was working on this you got another Result validated making my numbers change, so I added a row.

THESE NUMBERS ARE NOT EXACT.

RAC is VERY dependent on the exact time that the credit was granted. So, my numbers are not exact as to what you see in your account. However, you can see the RAC and the change that occurs because of the addition of the records to the flow.

More precise, and better controlled values are actually found on the statistic sites. These get the output files from the project and maintain databases with the data in there sliced and diced, blended, and puree-ed to a fare-thee-well ...

Looking at a table similar to your computer list here shows more interesting looks at what is going on with your computers.

I don't know if any of this helps, but, I gave it my best shot ...

For any other lurkers out there, was the RAC Estimator of interest to anyone?

I don't want to spend a week making a tool that no one will use ...

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.