Re: [LAU] Audio interface latency measurements

On 06/21/2014 07:18 PM, Len Ovens wrote:
> On Sat, 21 Jun 2014, Robert Jonsson wrote:
>
>> Follow up question: from a theoretical perspective, is it likely a usb
>> 2.0 interface would have similar transport latency as firewire? Usb 1
>> I suppose would be worse due to lower clockspeed.
>
> As was already stated, clock speed of the interface is not really
> relevant. It seems in fact that no one is really interested in USB3
> because it does not have any improvement for audio, USB2 is enough. The
> limitation with USB1 is bit depth, bit rate and channel count.
>
> In general, throughput and latency are two different things. Larger
> packets mean better throughput, but smaller packets mean lower latency.
>
> I am not sure, but it seems to me the USB1.1 audio standard effectively
> means that the lowest latency for USB1 is jackd set to 64/2.

With my Edirol UA-25 lowest possible setting is 48/2 @ 48kHz.

 This is the
> smallest buffer size supported. I do not know, but it seems that fire
> wire audio is about the same from what I have read (I don't have one of
> my own to confirm).

With the FireWire interfaces I've owned I could go as low as 16/3 @
48kHz. But settings like this are unusable, DSP load qickly rises as
soon as you start doing something serious. Nice to brag about but that's
about it ;)

>
> The main trouble with USB is on the MB. Finding a USB port that is not
> shared with something else via an internal hub. I think adding a USB
> card would make things better, but trying different ports on a laptop
> gives good results too. With any audio interface, having it's own irq is
> important, I have moved PCI cards to different slots with a big
> difference. It shouldn't be, but it seems tunning a computer for audio
> is a must still for low latency. Audio is very definately _not_ plug and
> play for (semi)pro audio work. There is no silver bullet kernel or
> distro that just makes everything work. On my laptop, there is one USB
> port that gives good audio... so long as the port next to it is empty...
> and the wireless kernel module is unloaded and .... you get the picture :)
>

It's indeed a matter of finding a free USB port. If you don't have any
you have to resort to unloading kernel modules or even unbinding drivers.

Jeremy

> --
> Len Ovens
> www.ovenwerks.net
>
> _______________________________________________
> Linux-audio-user mailing list
> Linux-audio-user@email-addr-hidden
> http://lists.linuxaudio.org/listinfo/linux-audio-user

_______________________________________________
Linux-audio-user mailing list
Linux-audio-user@email-addr-hidden
http://lists.linuxaudio.org/listinfo/linux-audio-user