RAC ?


log in

Advanced search

Message boards : Number crunching : RAC ?

Previous · 1 . . . 3 · 4 · 5 · 6
Author Message
Profile Crunch3r
Volunteer tester
Avatar
Send message
Joined: 15 Apr 99
Posts: 1540
Credit: 3,314,460
RAC: 0
Germany
Message 1094975 - Posted: 9 Apr 2011, 0:10:14 UTC - in response to Message 1094973.



And the RAC thing is gonna always be off a bit, I suspect.

Might have to do with where a particular cruncher is on the RAC curve VS just the grand total. I don't think the two would ever be just toooooooo far apart.

I don't think the project RAC is calculated by adding up the total RACs of the individual rigs...each must be kept track of by themselves.




But that doesn't make any sense at all. Summing up all RAC over all machines attaches IS the user RAC. How come S@H is the only project forging that ?

____________

Join BOINC United now!
Auto eVB | Autoversicherung

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38338
Credit: 561,774,232
RAC: 643,635
United States
Message 1094978 - Posted: 9 Apr 2011, 0:15:55 UTC - in response to Message 1094975.



And the RAC thing is gonna always be off a bit, I suspect.

Might have to do with where a particular cruncher is on the RAC curve VS just the grand total. I don't think the two would ever be just toooooooo far apart.

I don't think the project RAC is calculated by adding up the total RACs of the individual rigs...each must be kept track of by themselves.




But that doesn't make any sense at all. Summing up all RAC over all machines attaches IS the user RAC. How come S@H is the only project forging that ?


Ahh...
Could there not be a difference (given the 4-5 week weighting formula) that there could be a difference between the RAC calculated for a given rig's production and the total RAC calculated on total credits awarded?

I am not a mathematician...I am just trying to reason why the two totals might be slightly different.
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

Profile Crunch3r
Volunteer tester
Avatar
Send message
Joined: 15 Apr 99
Posts: 1540
Credit: 3,314,460
RAC: 0
Germany
Message 1094982 - Posted: 9 Apr 2011, 0:18:31 UTC - in response to Message 1094978.



Ahh...
Could there not be a difference (given the 4-5 week weighting formula) that there could be a difference between the RAC calculated for a given rig's production and the total RAC calculated on total credits awarded?

I am not a mathematician...I am just trying to reason why the two totals might be slightly different.


Well, i'm not a mathematician either, but why would there be any difference ? It makes no sense at all.

____________

Join BOINC United now!
Auto eVB | Autoversicherung

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38338
Credit: 561,774,232
RAC: 643,635
United States
Message 1094983 - Posted: 9 Apr 2011, 0:20:39 UTC - in response to Message 1094982.



Ahh...
Could there not be a difference (given the 4-5 week weighting formula) that there could be a difference between the RAC calculated for a given rig's production and the total RAC calculated on total credits awarded?

I am not a mathematician...I am just trying to reason why the two totals might be slightly different.


Well, i'm not a mathematician either, but why would there be any difference ? It makes no sense at all.

I'll give you an example in about 5 minutes...
Let me total some numbers.
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38338
Credit: 561,774,232
RAC: 643,635
United States
Message 1094987 - Posted: 9 Apr 2011, 0:31:11 UTC
Last modified: 9 Apr 2011, 0:34:15 UTC

In my case, I think it's just too dynamic.
8 high production rigs reporting.

I loaded my user page with it's totals.
Showed 267,025.23 RAC.

Went to my computers....
Just long enough to jot down all the individual rig's RACs.

Went back to my user page and reloaded.

Here's what I got.

First total RAC was 267,025.23

The individual totals added up to 266,248.73..

And when I went back to my main page and reloaded, the total RAC had already changed to 267,006.53
In a matter of less than a minute or two.

And it will probably change in my sig every time you reload the page.....
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

jmerryman
Send message
Joined: 3 Apr 11
Posts: 10
Credit: 126,837
RAC: 0
United States
Message 1094996 - Posted: 9 Apr 2011, 1:00:34 UTC - in response to Message 1094987.

In my case, I think it's just too dynamic.
8 high production rigs reporting.

I loaded my user page with it's totals.
Showed 267,025.23 RAC.

Went to my computers....
Just long enough to jot down all the individual rig's RACs.

Went back to my user page and reloaded.

Here's what I got.

First total RAC was 267,025.23

The individual totals added up to 266,248.73..

And when I went back to my main page and reloaded, the total RAC had already changed to 267,006.53
In a matter of less than a minute or two.

And it will probably change in my sig every time you reload the page.....


I am guessing you measure your RAC by how fast your electric meter is spinning out of control!!! :)

Profile Crunch3r
Volunteer tester
Avatar
Send message
Joined: 15 Apr 99
Posts: 1540
Credit: 3,314,460
RAC: 0
Germany
Message 1094998 - Posted: 9 Apr 2011, 1:02:10 UTC - in response to Message 1094987.

In my case, I think it's just too dynamic.
8 high production rigs reporting.

I loaded my user page with it's totals.
Showed 267,025.23 RAC.

Went to my computers....
Just long enough to jot down all the individual rig's RACs.

Went back to my user page and reloaded.

Here's what I got.

First total RAC was 267,025.23

The individual totals added up to 266,248.73..

And when I went back to my main page and reloaded, the total RAC had already changed to 267,006.53
In a matter of less than a minute or two.

And it will probably change in my sig every time you reload the page.....


But that's not How user RAC is defined... anyway, strange to post here again after (a year++ ?? ) .. Lot's of good people seem to have left and there hasn't been any real progress at all as it seems, just mediocrity. So S@H seems to be like froozen to AKV8, no progress after all those years...

The only real work done seems to be Raistmers ATI port, which is sad to see, it's a waste of what a ATI/AMD card can do, when used properly on the right project. (No offense towards Raistmers affords! At least he's doing something and y'all should be very thankful towards him!!!).

Anyway, i've been going back into your past post to 'catch up' a bit since last time i posted here, you were only involved in the usefull threads, moving S@H forward.

However, it was sad to see that and i will not comment any further.(Think about that!)

____________

Join BOINC United now!
Auto eVB | Autoversicherung

B-Man
Volunteer tester
Send message
Joined: 11 Feb 01
Posts: 253
Credit: 147,366
RAC: 0
United States
Message 1095042 - Posted: 9 Apr 2011, 2:45:41 UTC - in response to Message 1094998.

In my case, I think it's just too dynamic.
8 high production rigs reporting.

I loaded my user page with it's totals.
Showed 267,025.23 RAC.

Went to my computers....
Just long enough to jot down all the individual rig's RACs.

Went back to my user page and reloaded.

Here's what I got.

First total RAC was 267,025.23

The individual totals added up to 266,248.73..

And when I went back to my main page and reloaded, the total RAC had already changed to 267,006.53
In a matter of less than a minute or two.

And it will probably change in my sig every time you reload the page.....


But that's not How user RAC is defined... anyway, strange to post here again after (a year++ ?? ) .. Lot's of good people seem to have left and there hasn't been any real progress at all as it seems, just mediocrity. So S@H seems to be like froozen to AKV8, no progress after all those years...

The only real work done seems to be Raistmers ATI port, which is sad to see, it's a waste of what a ATI/AMD card can do, when used properly on the right project. (No offense towards Raistmers affords! At least he's doing something and y'all should be very thankful towards him!!!).

Anyway, i've been going back into your past post to 'catch up' a bit since last time i posted here, you were only involved in the usefull threads, moving S@H forward.

However, it was sad to see that and i will not comment any further.(Think about that!)

Hey Cruncher some things are going to change they just started testing the new MB v7 apps over on the beta site. They are introducing an new test to the WU in addition to spikes, pulse, triplets, Gaussian etc. So give us a few months and we will be running a whole new science app.
____________

Profile Crunch3r
Volunteer tester
Avatar
Send message
Joined: 15 Apr 99
Posts: 1540
Credit: 3,314,460
RAC: 0
Germany
Message 1095059 - Posted: 9 Apr 2011, 2:57:07 UTC - in response to Message 1095042.

In my case, I think it's just too dynamic.
8 high production rigs reporting.

I loaded my user page with it's totals.
Showed 267,025.23 RAC.

Went to my computers....
Just long enough to jot down all the individual rig's RACs.

Went back to my user page and reloaded.

Here's what I got.

First total RAC was 267,025.23

The individual totals added up to 266,248.73..

And when I went back to my main page and reloaded, the total RAC had already changed to 267,006.53
In a matter of less than a minute or two.

And it will probably change in my sig every time you reload the page.....


But that's not How user RAC is defined... anyway, strange to post here again after (a year++ ?? ) .. Lot's of good people seem to have left and there hasn't been any real progress at all as it seems, just mediocrity. So S@H seems to be like froozen to AKV8, no progress after all those years...

The only real work done seems to be Raistmers ATI port, which is sad to see, it's a waste of what a ATI/AMD card can do, when used properly on the right project. (No offense towards Raistmers affords! At least he's doing something and y'all should be very thankful towards him!!!).

Anyway, i've been going back into your past post to 'catch up' a bit since last time i posted here, you were only involved in the usefull threads, moving S@H forward.

However, it was sad to see that and i will not comment any further.(Think about that!)

Hey Cruncher some things are going to change they just started testing the new MB v7 apps over on the beta site. They are introducing an new test to the WU in addition to spikes, pulse, triplets, Gaussian etc. So give us a few months and we will be running a whole new science app.


From what i read, this is only done, throttling the demand on the servers and forcing a credit reduction (done in the past many times) it's not a new sience app after all. It's a mere extension of an old lunatics app 2.4 that became the stock app at seti... so still no progrees after all.

Optimized apps are stuck with with AKv8 snd that one is about 2.5 years old (thank god it got ported by JDWhale to windows &linux from OSX!!!) and the stock app is even worse..

So basically nothing news here after all.
____________

Join BOINC United now!
Auto eVB | Autoversicherung

B-Man
Volunteer tester
Send message
Joined: 11 Feb 01
Posts: 253
Credit: 147,366
RAC: 0
United States
Message 1095124 - Posted: 9 Apr 2011, 4:11:05 UTC - in response to Message 1095059.
Last modified: 9 Apr 2011, 4:12:28 UTC

In my case, I think it's just too dynamic.
8 high production rigs reporting.

I loaded my user page with it's totals.
Showed 267,025.23 RAC.

Went to my computers....
Just long enough to jot down all the individual rig's RACs.

Went back to my user page and reloaded.

Here's what I got.

First total RAC was 267,025.23

The individual totals added up to 266,248.73..

And when I went back to my main page and reloaded, the total RAC had already changed to 267,006.53
In a matter of less than a minute or two.

And it will probably change in my sig every time you reload the page.....


But that's not How user RAC is defined... anyway, strange to post here again after (a year++ ?? ) .. Lot's of good people seem to have left and there hasn't been any real progress at all as it seems, just mediocrity. So S@H seems to be like froozen to AKV8, no progress after all those years...

The only real work done seems to be Raistmers ATI port, which is sad to see, it's a waste of what a ATI/AMD card can do, when used properly on the right project. (No offense towards Raistmers affords! At least he's doing something and y'all should be very thankful towards him!!!).

Anyway, i've been going back into your past post to 'catch up' a bit since last time i posted here, you were only involved in the usefull threads, moving S@H forward.

However, it was sad to see that and i will not comment any further.(Think about that!)

Hey Cruncher some things are going to change they just started testing the new MB v7 apps over on the beta site. They are introducing an new test to the WU in addition to spikes, pulse, triplets, Gaussian etc. So give us a few months and we will be running a whole new science app.


From what i read, this is only done, throttling the demand on the servers and forcing a credit reduction (done in the past many times) it's not a new sience app after all. It's a mere extension of an old lunatics app 2.4 that became the stock app at seti... so still no progrees after all.

Optimized apps are stuck with with AKv8 snd that one is about 2.5 years old (thank god it got ported by JDWhale to windows &linux from OSX!!!) and the stock app is even worse..

So basically nothing news here after all.

According to Eric this is a new 5th test type according to his post on SETI Beta http://setiweb.ssl.berkeley.edu/beta/forum_thread.php?id=1829&nowrap=true#40183 This is a new Autocorrelation test. This test has never been run here on SETI before. This is a big change for MB WUs.
____________

Profile Crunch3r
Volunteer tester
Avatar
Send message
Joined: 15 Apr 99
Posts: 1540
Credit: 3,314,460
RAC: 0
Germany
Message 1095129 - Posted: 9 Apr 2011, 4:17:57 UTC - in response to Message 1095124.


According to Eric this is a new 5th test type according to his post on SETI Beta http://setiweb.ssl.berkeley.edu/beta/forum_thread.php?id=1829&nowrap=true#40183 This is a new Autocorrelation test. This test has never been run here on SETI before. This is a big change for MB WUs.


Some SETI researchers have theorized that ET would send a signal waveform, and then send a copy of the same waveform, but slightly delayed (but before the first waveform had ended).


Who's "some" and what is the valid scientific basis to assume that it's a wave form (if it's at all a wave, which is wild guess and based on no valid reason at all...) ?

____________

Join BOINC United now!
Auto eVB | Autoversicherung

-BeNt-
Avatar
Send message
Joined: 17 Oct 99
Posts: 1234
Credit: 10,116,112
RAC: 0
United States
Message 1095320 - Posted: 9 Apr 2011, 17:32:20 UTC - in response to Message 1095129.
Last modified: 9 Apr 2011, 17:32:39 UTC

User RAC... according to the last time i had enough RAC to post here it was the sum of all machines attached under my account.

Obviously that changed and somehow it's not the case any more... my account page shows that i got a RAC of 44.52...

Now let's look at my host that are attached:

#1 RAC == 0.09 +
#2 RAC == 46.03
USER RAC == 46.12

Must be me since my math doesn't add up...


You complain about the RAC without understanding how it's calculated. It's not merely a sum of your credits.


Updating RAC:
So how is the Recent Average Credit computed? Each time new Credit granted, the following function is used to update the Recent Average Credit of the Computer, Participant, or Team:

RAC(new) = RAC(old)*d(t) + (1-d(t))*credit(new)

Where d(t) is the decay function, and t is the time (in seconds) since the last Recent Average Credit recalculation.

d(t) = e^(-ln(2)*t / 604800)

The Recent Average Credit for any Computer, Participant, or Team is calculated every time that Credit is added to that object.

Note:
Also, the Recent Average Credit number is updated when Credit is granted after the Result has passed Validation, not when a Result is returned. This is true for all current Projects with the exception of Climateprediction.net (CPDN) which has different Quorum Rules.

Decreasing the RAC
If an entire week passed since the last credit was granted, the Recent Average Credit for the object will be divided by 2. The actual calculation is exponential. If an object starts with a Recent Average Credit of 128, and no Credit is added, after a week, the RAC will be 64, after another week, the RAC will be 32. Continuing week by week 16, 8, 4, 2, 1, 1/2, 1/4, 1/8, 1/16 ... Eventually, it will be close enough to zero so as not to matter.

Note:
The original intent was to actually calculate the Recent Average Credit at least once a week. This provision was dropped because of the impact on the BOINC Database by the query that would be used to update the numbers. Most of the Third Party Statistics Sites still calculate the Recent Average Credit this way.

Important to Notice
On one point, there is an oddity, and it doesn't make perfect sense. Removing a component will have no bearing on the calculation of the parent's Recent Average Credit.

Since the calculation of the Recent Average Credit is sensitive to time that the Recent Average Credit was last calculated, the Recent Average Credit of the children may not add to the Recent Average Credit of the parent. For example it is possible to have 2 computers that have Recent Average Credits of 15 and 30, and still have the Participant's Recent Average Credit of 35, or 25, etc.

Right after joining a BOINC Powered 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 Recent Average Credit at the beginning of a Participants 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.


There is much more information at that link above.
____________
Traveling through space at ~67,000mph!

jmerryman
Send message
Joined: 3 Apr 11
Posts: 10
Credit: 126,837
RAC: 0
United States
Message 1095342 - Posted: 9 Apr 2011, 18:43:07 UTC - in response to Message 1095320.

User RAC... according to the last time i had enough RAC to post here it was the sum of all machines attached under my account.

Obviously that changed and somehow it's not the case any more... my account page shows that i got a RAC of 44.52...

Now let's look at my host that are attached:

#1 RAC == 0.09 +
#2 RAC == 46.03
USER RAC == 46.12

Must be me since my math doesn't add up...


You complain about the RAC without understanding how it's calculated. It's not merely a sum of your credits.


Updating RAC:
So how is the Recent Average Credit computed? Each time new Credit granted, the following function is used to update the Recent Average Credit of the Computer, Participant, or Team:

RAC(new) = RAC(old)*d(t) + (1-d(t))*credit(new)

Where d(t) is the decay function, and t is the time (in seconds) since the last Recent Average Credit recalculation.

d(t) = e^(-ln(2)*t / 604800)

The Recent Average Credit for any Computer, Participant, or Team is calculated every time that Credit is added to that object.

Note:
Also, the Recent Average Credit number is updated when Credit is granted after the Result has passed Validation, not when a Result is returned. This is true for all current Projects with the exception of Climateprediction.net (CPDN) which has different Quorum Rules.

Decreasing the RAC
If an entire week passed since the last credit was granted, the Recent Average Credit for the object will be divided by 2. The actual calculation is exponential. If an object starts with a Recent Average Credit of 128, and no Credit is added, after a week, the RAC will be 64, after another week, the RAC will be 32. Continuing week by week 16, 8, 4, 2, 1, 1/2, 1/4, 1/8, 1/16 ... Eventually, it will be close enough to zero so as not to matter.

Note:
The original intent was to actually calculate the Recent Average Credit at least once a week. This provision was dropped because of the impact on the BOINC Database by the query that would be used to update the numbers. Most of the Third Party Statistics Sites still calculate the Recent Average Credit this way.

Important to Notice
On one point, there is an oddity, and it doesn't make perfect sense. Removing a component will have no bearing on the calculation of the parent's Recent Average Credit.

Since the calculation of the Recent Average Credit is sensitive to time that the Recent Average Credit was last calculated, the Recent Average Credit of the children may not add to the Recent Average Credit of the parent. For example it is possible to have 2 computers that have Recent Average Credits of 15 and 30, and still have the Participant's Recent Average Credit of 35, or 25, etc.

Right after joining a BOINC Powered 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 Recent Average Credit at the beginning of a Participants 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.


There is much more information at that link above.



What happened to using small words and typing slow...ohh I thought it was hard to figure out how many apples sally would have was tough...my head hurts now! lol

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38338
Credit: 561,774,232
RAC: 643,635
United States
Message 1095372 - Posted: 9 Apr 2011, 19:21:33 UTC - in response to Message 1095342.
Last modified: 9 Apr 2011, 19:21:49 UTC


What happened to using small words and typing slow...ohh I thought it was hard to figure out how many apples sally would have was tough...my head hurts now! lol

Edit the post...LOL.
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

Previous · 1 . . . 3 · 4 · 5 · 6

Message boards : Number crunching : RAC ?

Copyright © 2014 University of California