Re: AppleUSBAudio Driver Latency
Re: AppleUSBAudio Driver Latency
- Subject: Re: AppleUSBAudio Driver Latency
- From: Jeff Moore <email@hidden>
- Date: Tue, 13 Nov 2001 19:15:24 -0800
on 11/13/01 4:32 PM, Sean Morrell <email@hidden> wrote:
>
What is the round-trip latency of audio through OS X with the
>
AppleUSBAudio driver?
Not sure what you mean. Drivers don't have latency, unless they are doing
some signal processing that has latency.
The device might have some hardware latency and the transport might have
peculiar characteristics (like USB and its 1 millisecond time slices), but
the way the stack works on X makes it so that the driver doesn't add any
latency.
>
How how does this compare to audio devices on other
>
buses such as PCI?
The differences between transports are the same on X as they are on other
systems.
>
Is the latency known for the Audio HAL, the driver,
>
and other components separately?
Latency in the OS X stack doesn't work like it does on other platforms. This
is one of the big features of the stack, in fact.
The throughput latency for a client app of the HAL on a full duplex device
is going to always be ((2 * IOBufferSize) + InputSafetyOffset +
OutputSafetyOffset + InputHardwareLatency + OutputHardwareLatency) no matter
what the transport is or what the device is. This is due to the decoupling
of the clients from the hardware interrupt and using the time stamps to dead
reckon where the DMA is at any given time so as to properly place the data
in the DMA buffer to hear it at the appropriate time.
All these numbers are available through properties in the HAL, BTW.
--
Jeff Moore
Core Audio
Apple