Advances counted in the nuber of asteroids, not credits
log in

Advanced search

Message boards : Science : Advances counted in the nuber of asteroids, not credits

Author Message
Markus Sadeniemi
Send message
Joined: 24 Sep 12
Posts: 8
Credit: 4,388,040
RAC: 2,158
Message 522 - Posted: 27 Dec 2012, 8:09:28 UTC

Some questions out of curiosity. Not really science but not number crunching either:

1
It is said that it takes a lot of cpu to calculate the rotation and other properties of an asteroid. Huw much is a lot? Ten/ten thousand/ten million cpu hours?
2
With this capasity we participans now provide, how much do we manage to calculate (say) within a month?
3
I understood that you start by testing the system with artificial data. When do you expect to be able to move to the real thing?
4
Have more than the two asteroids mentioned in the web page been calculated? How many?

regards
Markus Sadeniemi

cykodennis
Send message
Joined: 10 Jul 12
Posts: 69
Credit: 9,086,498
RAC: 0
Message 523 - Posted: 27 Dec 2012, 18:39:22 UTC - in response to Message 522.

I'm interested in these questions, too.

Profile Kyong
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 9 Jun 12
Posts: 570
Credit: 52,629,744
RAC: 0
Message 527 - Posted: 30 Dec 2012, 10:02:24 UTC

Answers:
1. The approximate time about 1200 hours on 2.66 GHz Q9400. An asteroid is divided in 250 - 300 parts.
2. According to answer 1 of how many parts it is divided to, you can calculate it from data of server status. Performance is still increasing so it is better if you calculate it, the time to complete one asteroid is still reducing.
3. Strange question, we have been putting the real data there. The artifical data was only at very beggining of the project.
4. There are more asteroids then two calculated. I will write later how many I don't know the number by heart.

cykodennis
Send message
Joined: 10 Jul 12
Posts: 69
Credit: 9,086,498
RAC: 0
Message 530 - Posted: 30 Dec 2012, 12:41:16 UTC

Thank you Kyong!

Markus Sadeniemi
Send message
Joined: 24 Sep 12
Posts: 8
Credit: 4,388,040
RAC: 2,158
Message 531 - Posted: 30 Dec 2012, 12:54:55 UTC - in response to Message 527.
Last modified: 30 Dec 2012, 12:55:32 UTC

Thank you!

regards
Markus

RanmaCanada
Send message
Joined: 29 Dec 12
Posts: 1
Credit: 130,440
RAC: 0
Message 532 - Posted: 31 Dec 2012, 0:09:05 UTC
Last modified: 31 Dec 2012, 0:12:06 UTC

So if that is how long it takes on a Q9400, with our i7's that run considerably faster, is it possible that we have halved the time it takes to do one computation? And if and when you get a GPU client running, all in good time mind you as you have done so much already!, would we cut the computational time down to one fifth?

Yes yes I know I am dreaming here, but the faster we can find these things, the better? And I honestly have no idea how the scaling works in regards to work units and processing time, etc. Though I do know my i7 970 is clocking 12 WU at at a time in about 3.5 hours. Which means I am doing one asteroid in 87.5 hours?? Am I oversimplifying things haha.

Dagorath
Send message
Joined: 16 Aug 12
Posts: 293
Credit: 1,116,280
RAC: 0
Message 534 - Posted: 31 Dec 2012, 4:31:14 UTC - in response to Message 532.

Yes yes I know I am dreaming here, but the faster we can find these things, the better?


What do you think we are trying to find?


And I honestly have no idea how the scaling works in regards to work units and processing time, etc. Though I do know my i7 970 is clocking 12 WU at at a time in about 3.5 hours. Which means I am doing one asteroid in 87.5 hours?? Am I oversimplifying things haha.


I don't think you are oversimplifying. I think you don't have any idea what this project does.

Enric Surroca
Send message
Joined: 14 Dec 12
Posts: 43
Credit: 10,611,840
RAC: 4,772
Message 537 - Posted: 1 Jan 2013, 0:42:46 UTC

Hi all.
Kyong thanks for the answers. Really appreciate having feed back from admins of projects. Again thanks and good job.

Dagorath, don't be so tough. Most of us don't have big scientific knwoledge. In my case, I'm a lawyer being fascinated with science since my grand-father explained me his job and some science curiosities, and I like participating with this project in order to help scientists in their research. So You must understand sometimes we make silly questions (can be good in our jobs but not so much in science). Beg You, be patient with us.

But anyway, I've read some of your posts in the forum and always understood Your answers (You are good 'cause I'm no pro in science, just begginer).

Thank You all and wish You a Happy New Year.

Dagorath
Send message
Joined: 16 Aug 12
Posts: 293
Credit: 1,116,280
RAC: 0
Message 539 - Posted: 1 Jan 2013, 18:56:36 UTC - in response to Message 537.
Last modified: 1 Jan 2013, 18:57:41 UTC

Happy New Year to all!!

Enric,

As a lawyer you know how important words are. It's all in the meanings of the words we write and read. If RanmaCanada read something here or anywhere else that conveyed the idea that this project is looking for NEOs then I would like to see that information corrected, that's all. I think s/he thinks Asteroids@Home is looking for NEOs and if so then I think someone should clarify the matter for him. S/He and everyone who joins Asteroids@Home should know exactly what cause s/he is supporting with their resources. Unfortunately people spread false information around and others pick it up, believe it and make decisions they normally would not make.

[AF>Libristes] Dudumomo
Send message
Joined: 5 Aug 12
Posts: 3
Credit: 16,194,240
RAC: 146,398
Message 541 - Posted: 2 Jan 2013, 10:03:27 UTC - in response to Message 532.


Though I do know my i7 970 is clocking 12 WU at at a time in about 3.5 hours. Which means I am doing one asteroid in 87.5 hours?? Am I oversimplifying things haha.


Based on Kyong answer, my understanding is that:

A Q9400 (quad core) needs 1200h for an asteroid.
Each asteroid is divided in 250-300parts (Let's say 300) and thus each part needs 4h per unit (1200h/300parts) at full load. (4 cores)

You are doing 12WUs in 3.5h, meaning you are doing 3 times more WUs than a Quad core while calculating 14% faster the units.

So basically, you are analyzing a complete asteroid in:
300 parts x 3.5h / 3 times more cores
= 350h.

It is an approximation.

But I will be more interested to know how many asteroids we are analyzing, at what speed, etc... for curiosity

Profile rilian
Avatar
Send message
Joined: 18 Jun 12
Posts: 8
Credit: 83,844
RAC: 0
Message 1303 - Posted: 25 May 2013, 8:01:49 UTC - in response to Message 527.

Answers:
4. There are more asteroids then two calculated. I will write later how many I don't know the number by heart.


hi!

how often is results page updated http://asteroidsathome.net/scientific_results.html ?

how many WUs does it take to get to the list ? :)
____________

cykodennis
Send message
Joined: 10 Jul 12
Posts: 69
Credit: 9,086,498
RAC: 0
Message 1304 - Posted: 25 May 2013, 13:38:30 UTC - in response to Message 1303.


how many WUs does it take to get to the list ? :)


Looks like, it's completely random. *frustrated*

Josef Durech
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 18 Jun 12
Posts: 26
Credit: 4,851,687
RAC: 2,240
Message 1308 - Posted: 27 May 2013, 8:40:46 UTC - in response to Message 1303.

The "Scientific results" page is updated irregularly. Please be patient. The more WUs computed, the higher is the chance to get in the list - however, no amount of WUs can secure it.

magic8192
Send message
Joined: 9 Oct 12
Posts: 1
Credit: 30,128,040
RAC: 0
Message 3117 - Posted: 30 May 2014, 16:58:55 UTC - in response to Message 1308.

The "Scientific results" page is updated irregularly. Please be patient. The more WUs computed, the higher is the chance to get in the list - however, no amount of WUs can secure it.


That is pretty neat.


Post to thread

Message boards : Science : Advances counted in the nuber of asteroids, not credits


Main page · Your account · Message boards


Copyright © 2019 Asteroids@home