Brian Beesley wrote:
> On Friday 18 May 2007 18:54, Coran Fisher wrote:
>
>   
>> be no decrease in accuracy.  Also boinc projects are sometimes serious
>> scientific endeavors so accuracy is extremely important, do you have any
>> reason to suspect a boinc project is returning inaccurate results
>> because it's used by boinc?
>>     
>
> I don't see much evidence of cross checking. I do see lots of evidence of 
> massive timewasting e.g. SETI having a few blocks of data run many, many 
> times. Also the presentational work involved in being run as a screensaver 
> wastes loads of CPU cycles per client. OK there may be other ways of running 
> BOINC clients but most people just don't bother. FYI most of the systems I 
> have running mprime don't run a GUI at all. Command line works just fine, 
> uses a lot less less resources... and lets mprime run a percent or two 
> faster, even on state of the art systems running GUIs that are a lot more 
> efficient than windoze, even when no foreground applications are running.
>   
All the processing routines and work assignment for BOINC projects are 
done by the project; i.e. SETI in this case.  They design the apps, 
assign the work, and do any cross-checking they feel is necessary.  All 
BOINC does is provide the platform for distributing the project's 
application and data, and allowing a user to distribute cpu cycles among 
various projects.  Even BOINC's routine communications are with the 
project's web site, not the BOINC servers. 

I don't run the graphical display; in fact, I didn't know there was one 
until yesterday.  Everything runs in the background (as a windoze 
service) on my system, with just a management app that I can bring up if 
I feel like checking the status of stuff.

D

>   
>> I also suspect you could continue to run 
>> the stand alone client as a few projects did that for a while so
>> basically you could choose the current style of client or boinc.
>>     
>
> For a while? Long enough to enable me to find something else?
>
>   
>> So I 
>> see no reason for people to leave if boinc was used.  The (minimal) gains
>> vs the work involved is the only show stopper I see.
>>     
>
> Minimal gains could well be net losses.
>
> Depends whether you want to see numbers of users or numbers of _useful_ CPU 
> cycles contributed.
>
> Regards
> Brian Beesley
> _______________________________________________
> Prime mailing list
> [email protected]
> http://hogranch.com/mailman/listinfo/prime
>
>
>   


_______________________________________________
Prime mailing list
[email protected]
http://hogranch.com/mailman/listinfo/prime

Reply via email to