Re: laptop audio I/O hardware

[Brian Cass <b@the0verclock.com>]

my 2 cents,

beware of the occasional driver that displays 'roundtrip' latency. or  
maybe it's better to be aware that most do not show roundtrip. there  
are buffers on the way in and the way out, so if you are processing  
live audio you get twice the latency.
some driver setting windows will show one or the other, if you are  
setting a third party device from within the settings for a host app,  
who knows if you can trust the numbers it displays. also higher  
sampling rates give you significantly less latency, albeit at the  
cost of CPU.

- b



On Jun 7, 2006, at 4:10 PM, jeff larson wrote:

>> From: Jeff Kaiser
>> The buffers are set in the DSP options of the application itself.  
>> They
>
>> remained the same.
>>
>> Could the driver itself change the CPU load? Would that
>> affect latency?
>
> A driver can cause more CPU load, but that does not necessarily
> affect latency.  Latency is defined by the buffer size.  If your
> buffer size is 256 samples, then your latency is 5.8 milliseconds,
> period.  It doesn't matter what hardware interface you are using,
> if the buffer size is 256 your latency is 5.8 milliseconds, always
> and forever.
>
> What increased CPU load can do is cause you to "miss interrupts"
> because the computer can't keep up with requests from the audio
> interface.  This will sound like clicks or noise.  So while a
> particular driver may require more CPU, if you are not hearing any
> clicks it doesn't matter, it is still operating within the latency
> defined by the buffer size.
>
> Jeff
>