Stats for SETI@home only

Message boards : Number crunching : Stats for SETI@home only
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 3 · 4 · 5 · 6 · 7 · 8 · Next

AuthorMessage
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1017030 - Posted: 18 Jul 2010, 3:50:58 UTC - in response to Message 1017009.  
Last modified: 18 Jul 2010, 3:55:50 UTC

Re-coded in 100% Reverse Polish Notation - I haven't had time to fine tooth comb the results, so let me know if something is munged.
RAC / User Charts


first glance looks a lot closer, thank you scarecrow

toto too

Besides refining the "daily" part of it, I also am throwing out users with a RAC of true 0. However I think that some users that have stopped crunching long ago still have a very small (even fractional) RAC "on the books". Not sure how/what/when they get to a true 0.

* Rather than Toto, thanks go to Miep for various mathematical and sanity checks.
ID: 1017030 · Report as offensive
Profile Uli
Volunteer tester
Avatar

Send message
Joined: 6 Feb 00
Posts: 10923
Credit: 5,996,015
RAC: 1
Germany
Message 1017033 - Posted: 18 Jul 2010, 3:56:50 UTC - in response to Message 1017023.  

Thanks
Question: Do I now have to chuck the old link bookmark?

Nope, it's all in the same place, just bigger, faster, better than before.

Thanks
Pluto will always be a planet to me.

Seti Ambassador
Not to late to order an Anni Shirt
ID: 1017033 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1017087 - Posted: 18 Jul 2010, 5:26:47 UTC - in response to Message 1017030.  


Besides refining the "daily" part of it, I also am throwing out users with a RAC of true 0. However I think that some users that have stopped crunching long ago still have a very small (even fractional) RAC "on the books". Not sure how/what/when they get to a true 0.
...

The only users with true 0 RAC are ones which have never been granted any credit. Either they never crunched a WU or they only got errors.

For users who quit, RAC declines by half each week there's no credit granted. A RAC of 524288 (2^19) would decline to the smallest number which a double can represent in just about 20 years. A RAC of 0.0078125 (1/128 or 2^-7) would get to the smallest representable number 26 weeks earlier. Disclaimer: The BOINC RAC calculation does use doubles, but what happens when the number is a denormal after 18 years or so may not conform to the expected decline.
                                                               Joe
ID: 1017087 · Report as offensive
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1017090 - Posted: 18 Jul 2010, 5:38:49 UTC - in response to Message 1017087.  


The only users with true 0 RAC are ones which have never been granted any credit. Either they never crunched a WU or they only got errors.


Thanks, Josef. I'll add a label similar to those found on some food packages..."may add an insignificant amount of RAC". ;)

ID: 1017090 · Report as offensive
Profile Uli
Volunteer tester
Avatar

Send message
Joined: 6 Feb 00
Posts: 10923
Credit: 5,996,015
RAC: 1
Germany
Message 1017097 - Posted: 18 Jul 2010, 5:53:09 UTC - in response to Message 1017090.  
Last modified: 18 Jul 2010, 5:54:48 UTC


The only users with true 0 RAC are ones which have never been granted any credit. Either they never crunched a WU or they only got errors.


I think you deserve some chocolate...

Pluto will always be a planet to me.

Seti Ambassador
Not to late to order an Anni Shirt
ID: 1017097 · Report as offensive
Profile soft^spirit
Avatar

Send message
Joined: 18 May 99
Posts: 6497
Credit: 34,134,168
RAC: 0
United States
Message 1019378 - Posted: 25 Jul 2010, 3:47:12 UTC

Total Active
Users 1,108,774 163,223
Hosts 2,687,068 259,726
Teams 57,998 15,448
Countries 234 203

Total Credit 78,676,584,637
Recent average credit RAC 69,421,934
Average floating point operations per second 694,219.3 GigaFLOPS / 694.219 TeraFLOPS

Snapshot taken 7/24/2010
Janice
ID: 1019378 · Report as offensive
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1019403 - Posted: 25 Jul 2010, 8:34:08 UTC

Here's a question for y'all. With the charts ideally showing a "daily" picture of things, would the user count be better if it was only users that reported credit that day? For example, on Jul 24, there are 1,109,270 users included in the RAC groups, but looking at the change in credit from the 23rd to the 24th, only 101,195 users had an increase in their credits.

# Users / RAC Group With Increase In Credit In Past 24 Hours
>0 <1               192
1-29              10045 
30-49              7869
50-99             14658 
100-299           30395
300-499           12321
500-999           12342
1000-2999          9685
3000-4999          1794
5000-9999          1124
10000-29999         610
30000-49999          78
50000-99999          66
100000-299999        14
300000+               2


Should this be the chart's criteria? An additional chart based on those numbers? Leave everything as it is?
ID: 1019403 · Report as offensive
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1019407 - Posted: 25 Jul 2010, 9:47:27 UTC
Last modified: 25 Jul 2010, 9:48:07 UTC

ID: 1019407 · Report as offensive
Robert Ribbeck
Avatar

Send message
Joined: 7 Jun 02
Posts: 644
Credit: 5,283,174
RAC: 0
United States
Message 1019412 - Posted: 25 Jul 2010, 10:45:19 UTC - in response to Message 1019407.  

Examples of my previous babbling....

07/24/10 chart with total # users/group

07/24/10 chart with only # of users/group that reported credit


I vote for # of users/group that reported credit
ID: 1019412 · Report as offensive
Profile soft^spirit
Avatar

Send message
Joined: 18 May 99
Posts: 6497
Credit: 34,134,168
RAC: 0
United States
Message 1019413 - Posted: 25 Jul 2010, 10:48:53 UTC - in response to Message 1019407.  

Examples of my previous babbling....

07/24/10 chart with total # users/group

07/24/10 chart with only # of users/group that reported credit


By using Scarecrows top chart, and a top secret program entitled "Calculator" (ok maybe not so secret) and punching some of the numbers in...

on 07/24/2010 there were 3688 people with an RAC of 3000+.
The amount of credit crunched by those was 51%.

To me, this supports my theory that.. the large crunchers and GPU are indespensible. As is the lower numbers portion.

It also kills my theory of a possibility that it could all be done in house.




Janice
ID: 1019413 · Report as offensive
Profile Dirk Villarreal Wittich
Avatar

Send message
Joined: 25 Apr 00
Posts: 2098
Credit: 434,834
RAC: 0
Holy See (Vatican City)
Message 1019414 - Posted: 25 Jul 2010, 10:52:07 UTC


Scarecrow:

People/members with 300.000 credits are mathematically (for a mathematician!) not included in your list.
You started with the "<" and ">" sign at the top of it but for the last result you use the "+" sign.
To include the 300.000 a ">" and a "=" sign should be used :
>300.000
=300.000

You can [must] use both signs at the same time.I cannot show it here because my keyboard is limited.

A trick: use the BBCode underline tag---> >

ID: 1019414 · Report as offensive
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1019424 - Posted: 25 Jul 2010, 11:20:30 UTC - in response to Message 1019414.  
Last modified: 25 Jul 2010, 11:21:36 UTC


Scarecrow:

People/members with 300.000 credits are mathematically (for a mathematician!) not included in your list.
You started with the "<" and ">" sign at the top of it but for the last result you use the "+" sign.
To include the 300.000 a ">" and a "=" sign should be used :
>300.000
=300.000

You can [must] use both signs at the same time.I cannot show it here because my keyboard is limited.

A trick: use the BBCode underline tag---> >

The sql queries used to get the numbers do follow your syntax, Dirk. The 300000+ is selecting >= 300000 from rac. I just wasn't awake enough to put >= in the original post.
ID: 1019424 · Report as offensive
Profile soft^spirit
Avatar

Send message
Joined: 18 May 99
Posts: 6497
Credit: 34,134,168
RAC: 0
United States
Message 1019426 - Posted: 25 Jul 2010, 11:41:06 UTC - in response to Message 1019424.  

if someone can hit exactly 300,000.. Well I bet a dime to a donut they do not hit it twice in a row ;)

Janice
ID: 1019426 · Report as offensive
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1019427 - Posted: 25 Jul 2010, 11:45:14 UTC - in response to Message 1019426.  

if someone can hit exactly 300,000.. Well I bet a dime to a donut they do not hit it twice in a row ;)

We aim to please, you aim too, please.
ID: 1019427 · Report as offensive
Profile Miep
Volunteer moderator
Avatar

Send message
Joined: 23 Jul 99
Posts: 2412
Credit: 351,996
RAC: 0
Message 1019429 - Posted: 25 Jul 2010, 12:00:01 UTC

@Dirk 'Korintenkacker' ;) No offence meant. While mathematically incorrect I doubt the 'normal person from the street' has any problems with a 300000 + tag. If you argue like that, he also doesn't need the >0 at the top because of the 0 < 1 after it ( not 0 <= 1). After applying sufficiently pure mathematics, you'll often find you've left behind your common audience. Or, as my supervisor put it : always tailor your presentation to your audience.

@Scarecrow
Keep the total # users/group for the bars.
You are talking of a group of averages, so it shouldn't matter whether individual group members reported on that day.
If you really want to show the figures of users with credit on that day I suggest an additional line.
While you are at it, a total daily credits column at the end of daily credit/group would be nice.

Now, I somehow would expect a more or less gaussian distribution. For some reason the 300-499 group is exceptionally small. There must be system based reason why less users fall into that category.

And all of that on only two coffees...
Carola
-------
I'm multilingual - I can misunderstand people in several languages!
ID: 1019429 · Report as offensive
Scarecrow

Send message
Joined: 15 Jul 00
Posts: 4520
Credit: 486,601
RAC: 0
United States
Message 1019436 - Posted: 25 Jul 2010, 12:29:49 UTC - in response to Message 1019429.  

While you are at it, a total daily credits column at the end of daily credit/group would be nice

Like this?
ID: 1019436 · Report as offensive
Profile Miep
Volunteer moderator
Avatar

Send message
Joined: 23 Jul 99
Posts: 2412
Credit: 351,996
RAC: 0
Message 1019437 - Posted: 25 Jul 2010, 12:32:35 UTC - in response to Message 1019436.  

While you are at it, a total daily credits column at the end of daily credit/group would be nice

Like this?


Yes :)
Thanks!
Carola
-------
I'm multilingual - I can misunderstand people in several languages!
ID: 1019437 · Report as offensive
Profile John Neale
Volunteer tester
Avatar

Send message
Joined: 16 Mar 00
Posts: 634
Credit: 7,246,513
RAC: 9
South Africa
Message 1019478 - Posted: 25 Jul 2010, 15:14:17 UTC - in response to Message 1019429.  

Scarecrow asked:

With the charts ideally showing a "daily" picture of things, would the user count be better if it was only users that reported credit that day?

-[snip]-

Should this be the chart's criteria (sic)? An additional chart based on those numbers? Leave everything as it is?

Miep replied:

Keep the total # users/group for the bars.
You are talking of a group of averages, so it shouldn't matter whether individual group members reported on that day.
If you really want to show the figures of users with credit on that day I suggest an additional line.

I agree with Miep, although I can't think of any compelling reason to show the data for users reporting on that day.
ID: 1019478 · Report as offensive
Profile Bill Walker
Avatar

Send message
Joined: 4 Sep 99
Posts: 3868
Credit: 2,697,267
RAC: 0
Canada
Message 1019509 - Posted: 25 Jul 2010, 17:42:47 UTC - in response to Message 1019478.  

I don't think users reporting on a given day tells us much. Many of the smaller crunchers may only get credit every second or third day, at random. And don't we all get zero credit three days a week now?

ID: 1019509 · Report as offensive
Robert Ribbeck
Avatar

Send message
Joined: 7 Jun 02
Posts: 644
Credit: 5,283,174
RAC: 0
United States
Message 1019514 - Posted: 25 Jul 2010, 18:06:43 UTC - in response to Message 1019509.  
Last modified: 25 Jul 2010, 18:09:28 UTC

I don't think users reporting on a given day tells us much. Many of the smaller crunchers may only get credit every second or third day, at random. And don't we all get zero credit three days a week now?


I disagree it's showing that on any given day the users with the lower Rac
though there are many users in the lower rac groups are actually not turning
in many WU. and after a few days the trend is apparent
ID: 1019514 · Report as offensive
Previous · 1 . . . 3 · 4 · 5 · 6 · 7 · 8 · Next

Message boards : Number crunching : Stats for SETI@home only


 
©2022 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.