Re: [Algorithms] Algorithm performance benchmarks and timings

[Jonathan B. <jo...@nu...>]

There is no good way to answer this question without trying the algorithm
in some full-scale games and seeing how it becomes a problem.  (I am saying
"how", not "if", because it's rare to come up with a new algorithm that
does something useful that doesn't become a performance problem in at
least some cases).

If you don't have 8 million dollars to spend on a game right now, just 
sit down and make as much of a real game as you can manage.  

  -J.

----- Original Message ----- 
From: "Arno Kamphuis" <ar...@se...>
To: <gda...@li...>
Sent: Sunday, April 04, 2004 3:09 PM
Subject: Re: [Algorithms] Algorithm performance benchmarks and timings


> I know all about Big-O notation, that is not the problem. I can profile 
> whatever I want. Then I know exactly how much time my algorithm takes to 
> perform its task. But it is the practical problem that bothers me. So my 
> problem is: How can I relate the performance of the algorithm with the 
> time a game engine schedules for the algorithm. For example if I would 
> have a algorithm to determine the direction a non-player-character 
> should take and this algorithm would require 1 second for 100% pure 
> processor power for 1 frame it is definitely not suitable. However, if 
> this algorithm would only need 1 millisecond of full processor time, it 
> might be a very usable algorithm.
> 
> I am looking for 'benchmarks' or criteria on how much processor 
> power/time an algorithm can take in games. Is there any meaningful thing 
> to say about that?
> 
> thanks,
> 
> Arno
> 
> 
> > Google on Big-O notation, if you don't know about it already, for the 
> > theoretical aspect.
> >
> > Profile, profile, profile, for practical measurements, with all caveats 
> > of incorporating perf measurements in your code
> > "Premature Optimization is the root of all evil" says D.Knuth (or is 
> > commonly quoted as saying ;))
> > HTH,
> > Have a nice day,
> > Nicolas
> > 
> > On Dimanche, avri 4, 2004, at 13:56 America/Montreal, Arno Kamphuis wrote:
> > 
> >> Hi All,
> >>
> >> First of all I would like to apologize if this message is outside the 
> >> scope of this mailing list. I'm new to it, but I think it is appropriate.
> >>
> >> I have the following problem/questions. When developing a new 
> >> algorithm to be used in computer games, one problem that comes to mind 
> >> is whether  the algorithms performance is good enough so that it can 
> >> be used in games. Are there any benchmarks for this? I could for 
> >> example measure the amount of time it takes to perform a number of 
> >> steps in the algorithm and compare this to the amount of time it takes 
> >> to 'play' the results of the algorithm in the game. One example of 
> >> this is path planning, how long a path can be planned in the time 
> >> between two frames? Are there any other performance benchmarks that 
> >> are used to determine if an algorithm is useful?
> >>
> >> thanks in advance,
> >>
> >> Arno Kamphuis
> >>
> >> -- 
> >>
> >> url:   www dot sevendarfs dot nl
> >> email: arno at sevendwarfs dot nl
> >>
> >>
> >>
> >>
> >> -------------------------------------------------------
> >> This SF.Net email is sponsored by: IBM Linux Tutorials
> >> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> >> GenToo technologies. Learn everything from fundamentals to system
> >> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> >> _______________________________________________
> >> GDAlgorithms-list mailing list
> >> GDA...@li...
> >> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> >> Archives:
> >> http://sourceforge.net/mailarchive/forum.php?forum_id=6188
> >>
> > 
> > 
> > 
> > -------------------------------------------------------
> > This SF.Net email is sponsored by: IBM Linux Tutorials
> > Free Linux tutorial presented by Daniel Robbins, President and CEO of
> > GenToo technologies. Learn everything from fundamentals to system
> > administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> > _______________________________________________
> > GDAlgorithms-list mailing list
> > GDA...@li...
> > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> > Archives:
> > http://sourceforge.net/mailarchive/forum.php?forum_id=6188
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: IBM Linux Tutorials
> Free Linux tutorial presented by Daniel Robbins, President and CEO of
> GenToo technologies. Learn everything from fundamentals to system
> administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_id=6188