Re: Question about 64 bit Audio Units and CoreAudio's datatypes
Re: Question about 64 bit Audio Units and CoreAudio's datatypes
- Subject: Re: Question about 64 bit Audio Units and CoreAudio's datatypes
- From: Brian Willoughby <email@hidden>
- Date: Wed, 10 Nov 2010 04:00:45 -0800
On Nov 10, 2010, at 03:27, Olivier Tristan wrote:
On 11/10/2010 12:22 PM, Stefan Huber wrote:
I have been trying to understand what is needed be done in order
to create a 64bit Audio Unit. However I got to a point in the SDK,
that doesn't seem to make sense to me. In a true 64 bit Host the
Audio Graph should be able to be completely 64 bit. When I look at
the SDK I found out that audio data is of type
AudioUnitSampleType. An AudioUnitSampleType is a Float32, which
seems to be defined as float in MacTypes.h. As far as I know float
only uses 32bit, whereas double would use 64bit.
Even though the host application is a 64bit binary, it still uses
an SDK, that is processing with 32bit precision. Could anybody
please tell me what's the point in doing this or if I am making
any mayor mistakes here.
Major mistake.
64bits AU are NOT 64 bits audio processor just like 64 bits hosts
are NOT processing double(64bits) data instead of float(32bits)
64 bits only means 64 bits integer and pointers.
Another consideration is that a 32-bit OS running a 32-bit AudioUnit
can still do processing with 64-bit floats, at least between the
input and output. The only thing that the 64-bit OS and host enables
that isn't already possible in a 32-bit OS is more memory - i.e., 64-
bit address pointers. About the only AudioUnits that should benefit
from 64-bit are sampling instruments with more than 4 GB of sample
memory.
Brian Willoughby
Sound Consulting
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden