On Mon, Jul 18, 2011 at 12:52 PM, Erik de Castro Lopo wrote:
Thanks for clarification. One question remains: Is it then up
to the Jack clients to decide whether to do internal processing
with single or double precision, or is 64 bit mandatory/imposed
>> 32 bit floating point gives a dynamic range of ~192dB, well above the
OK, probably a good reason to use doubles for internal math
> This is probably the best known paper on the issues surrounding
Stuff to read, thanks!
> However, the problems of floating point are almost non-existant
>> All that
Hm. Yes. it depends on the type of operations done, and there are (many)
cases where fixed point math actually is inferior.
I mainly assumed the PT guys do all this "faffing about" to make sure
those shortcomings don't degrade audio quality, but maybe they cannot
guarantee this for any plugin.
Wha'ever. I'm not into PT and this will most likely never change.
Linux-audio-user mailing list