Re: Convert to float and more
Re: Convert to float and more
- Subject: Re: Convert to float and more
- From: Urs Heckmann <email@hidden>
- Date: Thu, 5 Jun 2003 18:14:23 +0200
Am Donnerstag, 05.06.03, um 16:20 Uhr (Europe/Berlin) schrieb Darrin
Cardani:
At 9:35 AM +0200 6/5/03, Urs Heckmann wrote:
Wouldn't it be nice to publish a small tutorial with ultra-optimized
example code and tricks for common problems? I.e.:
sin-cos approximation (getting both values at once)
Go to: http://www.shellandslate.com/ and click on "Research" at the
top, then click on "Computer Math 101" on the left side. I'd love to
see some of these written up as AltiVec routines that can do an
array's worth at a time. That would probably offer some significant
speedups for certain types of processing.
Hehe, I have found this page somewhere else before. The algo's are very
"general", so not as fast as stuff you code for your special purpose,
i.e. when you know the ranges of your values...
pow( 2, x )
You mean just shifting, or do you mean fractional powers, too?
fractional, of course...
(float) int
Well, we just discussed this, didn't we? :-)
(int) float
I am interested in how this is generally done when the processor
doesn't have a nice instruction for it.
There are instructions, but they do usually cost more than rolling your
own bitwise manipulations...
(Like transforming your float value to be inside a single mantissa
range, reference casting to int and throwing the mantissa out of it..)
However, these examples have been meant as, uhm, examples. - I assume
that the people at Apple have plenty of tricks which they'd love to
share with us audio developers so we can spend more time on supporting
the platform, hehe 8-)
Cheers,
;) Urs
_______________________________________________
coreaudio-api mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/coreaudio-api
Do not post admin requests to the list. They will be ignored.