Unit Speed

Message boards : Number crunching : Unit Speed
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile [B^S] madmac
Volunteer tester
Avatar

Send message
Joined: 9 Feb 04
Posts: 1175
Credit: 4,754,897
RAC: 0
United Kingdom
Message 94456 - Posted: 3 Apr 2005, 18:09:28 UTC

I have an old machine using only 850 Mhz, it takes about 10 and one half hours to do a unit. So how do I increase my speed so that it takes less. My classic took on averahe around 34 hours so I am quicker using Boinc but I want to do more. So thanks for any help
ID: 94456 · Report as offensive
Profile MikeSW17
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 1603
Credit: 2,700,523
RAC: 0
United Kingdom
Message 94473 - Posted: 3 Apr 2005, 19:25:53 UTC

madmac, you don't provide any information about your hardware or software that anyone can give advice about. What type of CPU, what memory, operating system etc.

That said, it is hightly unlikely that you can significantly improve an 850Mhz system. The effort would not justify any gain you might make.

If your system is something very unusual, there might be a possibility that someone has compiled a project client optimized for that hardware, but I would not expect it.



ID: 94473 · Report as offensive
Ron
Volunteer tester
Avatar

Send message
Joined: 16 Sep 99
Posts: 47
Credit: 3,900,704
RAC: 0
United States
Message 94474 - Posted: 3 Apr 2005, 19:30:15 UTC - in response to Message 94456.  
Last modified: 3 Apr 2005, 19:32:19 UTC

> I have an old machine using only 850 Mhz, it takes about 10 and one half hours
> to do a unit. So how do I increase my speed so that it takes less. My
> classic took on averahe around 34 hours so I am quicker using Boinc but I want
> to do more. So thanks for any help
>

Ya, an older machine is just slow. Nature of the beast. I have a 800 Mhz machine that turns out about two and a half work units per day. On the other hand, my server which has two 2.8ghz xeon processors cruches about 28 work units per day. 10 and a half hours per work unit on your 850mhz sounds about right.

Oh, and as far as I understand (I'm fairly new to BOINC), BOINC work units are far different than SETI Classic work units, so comparing them is kind of like comparing apples to oranges.
ID: 94474 · Report as offensive
Profile MikeSW17
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 1603
Credit: 2,700,523
RAC: 0
United Kingdom
Message 94535 - Posted: 3 Apr 2005, 22:31:50 UTC - in response to Message 94474.  
Last modified: 3 Apr 2005, 22:32:22 UTC

>
> Oh, and as far as I understand (I'm fairly new to BOINC), BOINC work units are
> far different than SETI Classic work units, so comparing them is kind of like
> comparing apples to oranges.
>

Remember that BOINC is just a Network Computing framework, to support Distributed Computing projects. BOINC doesn't process any data. BOINC does manage the uploading/downloading of projects and the data for those projects.

SETI/Classic had Workunits, SETI/BOINC has workunits (actually they're called Results - same difference). To you/me/us as participants, the Classic and BOINC SETI data sent to our systems to process are the same (may be identical?). So, a given unit of work given to either SETI/Classic or SETI/BOINC will take almost the same amount CPU cycles to run.

The big difference is, SETI Classic used to award 1 point for each work unit proceesed regardless of how much CPU it took. BOINC (whatever project SETI/Climate/Einstein etc) awards Cobblestones or CS. CS is awarded based on the amount of work done to complete 1 work unit (result).
This system is considered fairer as a unit that runs in 2hrs on one system and 5hrs on another will receive similar credit.

ID: 94535 · Report as offensive
Ron
Volunteer tester
Avatar

Send message
Joined: 16 Sep 99
Posts: 47
Credit: 3,900,704
RAC: 0
United States
Message 94567 - Posted: 4 Apr 2005, 0:00:41 UTC - in response to Message 94535.  

Ah, thanks for clearing that up Mike. I had misread something I think (in thinking that the wu's were different).

> >
> > Oh, and as far as I understand (I'm fairly new to BOINC), BOINC work
> units are
> > far different than SETI Classic work units, so comparing them is kind of
> like
> > comparing apples to oranges.
> >
>
> Remember that BOINC is just a Network Computing framework, to support
> Distributed Computing projects. BOINC doesn't process any data. BOINC does
> manage the uploading/downloading of projects and the data for those projects.
>
> SETI/Classic had Workunits, SETI/BOINC has workunits (actually they're called
> Results - same difference). To you/me/us as participants, the Classic and
> BOINC SETI data sent to our systems to process are the same (may be
> identical?). So, a given unit of work given to either SETI/Classic or
> SETI/BOINC will take almost the same amount CPU cycles to run.
>
> The big difference is, SETI Classic used to award 1 point for each work unit
> proceesed regardless of how much CPU it took. BOINC (whatever project
> SETI/Climate/Einstein etc) awards Cobblestones or CS. CS is awarded based on
> the amount of work done to complete 1 work unit (result).
> This system is considered fairer as a unit that runs in 2hrs on one system and
> 5hrs on another will receive similar credit.
>
>
<br><br><img src='http://www.boincsynergy.com/images/stats/1185sah.png'> <br>
Thanks to Tim (Captain Avatar) for my groovy Red Foxx avatar!
ID: 94567 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 94833 - Posted: 4 Apr 2005, 17:31:44 UTC - in response to Message 94567.  

> Ah, thanks for clearing that up Mike. I had misread something I think (in
> thinking that the wu's were different).

THe work Units are essentially the same.

The Science Applications are not the same. BOINC Has a number of improvements and additional processing changes. The good news is that the run times are not that different at this time.

However, I have seen hints that there are possible changes comming for the current Science Application to include optimized binaries for various processor families, additional science routines, and later additional Science Applications for other data, Astropulse, etc.

So, like CPDN is in the process of adding more to the models, so will SETI@Home. Even better, they can add new Science Applications or major modifications and run the two S-Apps in parallel ...
ID: 94833 · Report as offensive

Message boards : Number crunching : Unit Speed


 
©2024 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.