On Thu, 2006-04-13 at 11:34 +0200, Marcus Lindblom wrote:
> Interesting. I will definitely look into this if these functions prove
> to be bottlenecks.
>
> Since I can't parse french, I can only guess that a large part of these
> improvements is due to the fact that the built-in functions are
> IEEE-compliant with respect to rounding, INF/NaN-handling and exception
> reporting.
I can't parse French either, but I can parse graphs. ;)
> Fast-functions usually do an equally good job for your average data, but
> might produce unexpected/undefined (or at least not-to-specification)
> results for more esoteric datasets.
Looking at the graphs the main improvements seem to be with sin, cos and
tan. Other functions (sqrt, 1/sqrt, cosh etc.) are pretty much the same.
Given that we're not using sin, cos a dn tan very much I wouldn't expect
noticeable differences for OpenSG.
> Most compliers have optimziation flags that affect this behaviour (VS
> 2005 for instance, has three fp-options: fast, precise and strict) with
> varying tradeoffs between speed and behaviour.)
Has anybody ever benchmarked those? Do they make a difference?
Dirk
-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users
