21 Aug
2003
21 Aug
'03
11:55 a.m.
Hello,
In the course of researching another Linux article for Sound on Sound,
I was having a discussion with the features editor about latency,
specifically Linux versus Windows and Mac. The only cross-platform
serious study I'm aware of is this one, which is a bit out of date:
http://gigue.peabody.jhu.edu/~mdboom/latency-icmc2001.pdf
Now the features editor looked it up for himself and found this page,
also out of date I think:
http://www.rme-audio.com/english/linux/alsa.htm
It quotes a latency under Linux for the Hammerfall of 20ms, where as
Windows can do 6ms with this card if you install the ASIO drivers. I
think he's now sceptical that Linux can compete with Windows/ASIO in
the area of low latency.
Can any RME (or other pro card) users on the list quote reliable
latency figures for their own systems? It would be very helpful.
Daniel,
I'm an RME user. My experience is that Windows and Linux are identical.
When using Jack I can set my RME cards up for the same latency values
(I.e. - buffer sizes) as I can in Windows. This should yield the same
latency I think.
Do we have any tools that will actually test latency? For instance, I
suppose I could send a signal output of an ADAT port and record it coming
back in on another ADAT port and measure by hand, but that's sort of labor
intensive. Any other ideas? I'd be happy to take some measurements if we
have an agreed upon method for doing it.
My personal experience is that you can get negligible latency on
Audacity overdubs with Mandrake and the 'multimedia' kernel, compared
to obvious latency on Windows 98. But I haven't set up any scientific
tests with matching hardware.
Cheers
Daniel