Re: I/O Latency (Was: Layman with a mission)
Re: I/O Latency (Was: Layman with a mission)
- Subject: Re: I/O Latency (Was: Layman with a mission)
- From: Dennis Gunn <email@hidden>
- Date: Wed, 13 Oct 2004 19:50:27 +0900
On Oct 13, 2004, at 5:56 PM, Brian Willoughby wrote:
Any time you take a general purpose computer processor and run an
audio stream
through it, you absolutely must have buffering. Because the software
is
configurable by the user, it cannot be precisely predicted, and the
buffer is
there to allow for variation in the timing of the delivery of samples
from the
software routines. The buffers are not there to reduce jitter, or to
adjust
for differing clock rates, although those factors can also be
addressed. The
crux of the matter is that digital software does not produce samples
in a
steady stream where the timing between each sample is precise. A DAW
does not
process a single sample at a time. Instead, software produces samples
in
chunks or groups. This fact means that any general purpose DAW must
utilize a
buffer between input and DAW and also between DAW and output. There
is no way
around this.
OS 9 Macs are General Purpose computers and running RME hardware they
can apparently accomplish that buffering in 6 samples each direction
for a total of 12 samples when a DAW application is run on them.
Jitter, or something like it, can be a problem even in hardware
samplers or
digital effects which have digital input (because the clocking of the
samples
is external). But that is not exactly the reason why software has
buffering
between the digital audio stages.
It may be true that OS9 has lower latency,
It is true and is very well known.
but it cannot be as low as a
hardware sampler or standalone digital effects.
Please don't attribute this idea to me I never suggested that.
I think your numbers may be
impossible, even on OS9, for some hardware, because how can you have
62 samples
total of latency when the MOTU 896 has 34 samples of latency in each
direction?
Well for one thing I was talking about an RME system not MOTU. I don't
currently own any MOTU hardware though I have in the past. I measured
in in out latency using their their ADDA and their "total mix" mixer.
I found it was 62 samples, John Pitcairn measured it on his and got the
same result and then I contacted the techs at RME they said that that
was the expected latency. What exactly is it that you are questioning?
Are you saying you don't believe that it is possible for any system to
be faster than the MOTU system or are you saying that you think John
and I and the Tech at RME are lying?
Now, even once you accept that a DAW must use buffering,
and that the minimum
is probably 64+64 or 32+32 samples,
Excuse me I did not mean to say that I thought that the safety buffer
could be *eliminated* just reduced to something like what it was in
OS9.
I really think you should read what has been written here again.
Nobody, certainly not me suggested that a DAW does not need buffering,
I have been talking about a specific spot in the signal path that could
probably be reduced since historically it was smaller. Specifically I
am talking about the safety offset.
I started to respond to the rest of your post but I think it would
probably be better if you just checked what has been posted in this
thread already because it somewhat appears that you think I said things
that I did not actually say.
Once again I am talking about the function of core audio to conduit the
audio stream between an audio sequencer and Audio hardware and the
safety offset buffer that it seems to need to accomplish that end. In
OS9 it was MUCH smaller and that is a documented fact that is very well
known in the audio community. One example of that are these numbers
recently obtained by John Pitcairn.
I am sorry I don't know exactly what apple machine he ran these tests
on but here they are again:
Parameters:
Buffer size (ASIO/CoreAudio) - 64 samples
Apps - Logic 6.3.2 (OS X), Logic 6.1.0 (OS 9)
Audio - RME Multiface/cardbus
OS X (CoreAudio):
TotalMix monitoring latency - 62 samples
Logic 6.3.2 monitoring latency - 322 samples
OS 9 (ASIO):
TotalMix monitoring latency - 62 samples
Logic 6.1.0 monitoring latency - 192 samples
I am sure he can fill in the details about what machine he used.
_______________________________________________
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