Re: [Chicken-users] latency tests?

[Kon Lovett]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Jun 13, 2006, at 10:35 AM, felix winkelmann wrote:

> Hi!
>
> I was wondering how one could test GC "latency", i.e. tthe delays  
> caused
> by GC in high-performance applications, for example games.
> Can anybody come up with a good scenario, or even test code, that
> can be used to measure the impact of GC pauses? Or would it be
> a purely subjective measurement, like whether a graphical application
> that displays a sequence of frames shows noticable "jumps"?

The goal would be to simulate a heap mutator load that mirrors that  
of a game's "smooth" time.  Does Chicken have enough heap access  
profiling info to build such a test?

One idea is to use the srfi-27 distributions, probably normal, to  
construct a random set of opengl objects that might be used in a game  
scenario. Parameterized by a set of <object class mean variance>  
shouldn't be too big a job. ("What you mean 'we' Kimo-sabe?")

The "gears.scm" example in the opengl egg will show artifacts of gc  
latency (if I remember correctly). Putting a '(gc)' call in the  
'(run)' procedure would be an interesting test. One of the ways to  
handle gc in a game is to invoke it during idle times, like after a  
prompt while waiting for user input.

> (felix)
>
>
> _______________________________________________
> Chicken-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/chicken-users

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (Darwin)

iD8DBQFEjxC3JJNoeGe+5O4RAgCcAJ42NsSXoYObI2404aBg6BU11ssGawCdHWJu
yD2utpcpLufbb88/cCUe40Y=
=GywT
-----END PGP SIGNATURE-----