Re: Question on Audio Built-in driver on latest Powerbook G4
Re: Question on Audio Built-in driver on latest Powerbook G4
- Subject: Re: Question on Audio Built-in driver on latest Powerbook G4
- From: Jeff Moore <email@hidden>
- Date: Mon, 12 Dec 2005 12:08:58 -0800
On Dec 12, 2005, at 3:51 AM, Stéphane Letz wrote:
I'm testing our code on the latest Powerbook G4 1.67 / 512 Mo (OSX
10.4.3) and see some strange behaviour in the built-in audio driver:
- the driver does not work reliably with buffers of 64 frames: it
triggers regular kAudioDeviceProcessorOverload. This behaviour is
also seen using the MillionMonkeys test application and other
coreaudio applications. The same machine using the MOTU 828
FireWire driver doe not shows the problem: it works OK at 64 frames
and even 32 frames.
While undesirable, there's never any guarantees about being able to
run at any particular cycle size on any hardware. Some hardware has
lower thresholds than others. Apps need to be prepared and have
suitable fallback plans.
That said, have you looked at the IO cycle telemetry to see exactly
why you are getting the overload? You may find that your IOProc has
some other issues on this hardware.
- running our code at 128 frames, our process consume something
like 13% CPU. Using Shark to profile the code, it appears that the
Built-in driver consume a lot of time in
AppleDBDMAAudio::clipOutputSample and
IOAudioStream::readInputSamples and in particular in functions like
"eqalizer, bassEnhance, and dynamicRangeControl". Moreover these
functions seems not to be Altivec accelerated. What is the purpose
of these functions and why are they consuming so much CPU?
These functions are signal processing the driver does to protect the
speakers of the PowerBook.
--
Jeff Moore
Core Audio
Apple
_______________________________________________
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