On 04/19/2014 08:21 AM, Len Ovens wrote:
On Sat, 19 Apr 2014, Jonathan E Brickman wrote:
That helps a lot. Had not heard that USB3 was broken for low-latency
audio, I'll have to remember that. The reason I'm suddenly interested
I should clarify. There are no (that I know of) USB3 Audio Interfaces
made at this time. The problem is with USB2 Audio Interfaces plugged
into USB3 ports. The work around is to remove the USB3 driver and just
use the USB2 driver. This is fine for audio use, but means if you wish
to use the USB3 port for a USB hard drive on the same machine, it will
be slow (usb2 speeds). I do not know if this affects the USB2 ports on a
machine with USB3 hw, but one of the musts with USB audio is finding a
USB port with a clear IRQ... which might happen to be one of the usb3
ports. Also there are some machines (mostly laptops) that have only USB3
ports available. There are apparently some scanners that have problems
too. Intel has called this a "non-problem" saying the software should
just have the unit "resend" any bad packets... not low latency words.
My USB2 Canoscan scanner doesn't work on my Intel laptop's USB ports at
all; either 2 or 3.
My USB 1.1 (UCA-202) audio card works fine on Intel USB3 ports.
One issue that I think may complicate audio things on modern laptops is
that the built-in webcams may also be connected via USB. (Such is true
on my laptop, which uses an Intel motherboard.)
If USB3.1 has a new plug format, it will be usb3 only and this may force
the audio interface producers to design a USB3 audio interface that may
not have these problems. They have not done so at this time because
there is no need to, the USB3 spec does not add anything for the audio
world. A laptop with no usb1/2 ports might change that. An audio IF that
plugged into a USB2 hub plugged into a USB3 port is probably not a good
low latency audio solution.
My card has never worked well when plugged into a USB2 hub that connects
to a USB2 port on the laptop. Even before USB3.
I think I saw a comment in the Intel USB3 driver code that the USB3
standard doesn't propagate device resets down the chain. So if you have
an external hub attached and the onboard USB3 hub is reset, the external
hub is not informed for the reset. So if your audio card is hanging off
an external hub, it sounds to me like it would never get a reset signal
unless you power cycle the hub it's attached to?
I had not heard of the insertion count thing before. That is good to
know. For a lot of people it will not matter as they only plug things in
once and leave them. But for mobile use it sound like spending the extra
for reliablility is worth while. I wonder what the numbers are for the
standard "D" conectors are (like serial or VGA).
I don't know about the insertion count thing. I don't know if Linux does
the same thing, but Windows is notorious for storing a configuration for
a removable device, and adamantly reusing that configuration even if the
configuration is broken. I've seen that mentioned as a frequent cause of
problems with HP USB multifunction devices and some USB NICs.
--
David W. Jones
gnome@xxxxxxxxxxxxx
authenticity, honesty, community
http://dancingtreefrog.com
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user@xxxxxxxxxxxxxxxxxxxx
http://lists.linuxaudio.org/listinfo/linux-audio-user
