Re: [FFADO-user] Echo AudioFire 4 stability issues

[Jonathan W. <jw...@ju...>]

Hi Brian

On Tue, Feb 09, 2021 at 08:53:28AM -0500, Brian Hechinger wrote:
> On 2/8/21 9:44 PM, Jonathan Woithe wrote:
> > Firmware version 4.8 is prevalent when AudioFire (AF) devices are used with
> > FFADO.
> 
> Yeah, I kept seeing "use an older firmware" but never anything about 4.8
> specifically. When I decided "screw it I'll try them all" is when I found my
> choices were pretty much 5.8 or 4.8.
> 
> The AF4 somewhat worked with 5.8 and FFADO. But it was far from stable and
> couldn't run low latencies. Using the the 4.8 firmware I'm now running 48khz
> 16/2 rock solid.

Right.  So if I understand you correctly, you are able to run 16/2 with
FFADO if your AF4 is loaded with firmware v4.8.  Right?  OOI what sample
rate are you using?

> I'm familiar with xruns. This is completely different behavior though. In
> this case the xrun wasn't being gracefully handled would crash jack.
> Probably FFADO not able to do something due to the newer firmware is my
> guess.

While FFADO attempts to handle xruns gracefully it doesn't manage to do so
in every situation.  Thus the xruns are real.  What the underlying cause
might be is not, especially since firmware appears to be partially involved.

Looking back at some historical posts it appears the AF firmware issue is a
bit of a mixed bag.  The discussion in ticket 360[1] documents some work
which was done in relation to the AF firmware 5.8 in 2013.  The result of
testing done at the time indicated that AF8 and AF12 failed with firmware
5.8 while the other AFs (including the AF4 and pre8 were fine. 
Investigation of the initial problem indicated that in affected devices
there were some issues with some of the FCP commands (used during device
probing and initialisation) which prevented FFADO from starting the audio
streaming.  In this case streaming was never started and therefore xruns
could not happen.  Subsequently some commits went in to FFADO (first
available in version 2.2) which worked around the configuration issues.  At
the time users reported success running the AF8 and AF12 with firmware 5.8.

It should be noted that with firmware 5.8, Echo disabled the 4x sampling
rates (176.4 kHz and 192 kHz) due to unspecified reasons.

Since you are clearly able to start audio streaming under firmware 5.8 and
in any case have an AF4, you do not suffer from the same issue described by
ticket 360[1].  Clearly there's some other issue at play with the new
firmware which affects the interface's stability under particular
circumstances.  Reports we've had in the past suggest that people were able
to achieve stable running of the AF family with firmware 5.8 using FFADO 2.2
or later.  However, your experience shows that there are additional (as yet
unknown) issues which can affect some people with this firmware.

If firmware 4.8 works for you then I'd say stick with that.  I am not aware
of any particular reason to use firmware 5.8 instead of 4.8.  We could
perhaps inspect a verbose error dump from FFADO to determine whether there
was anything obvious happening, but unless someone with an affected system
had the motivation to work through the issues I'm not sure there's much
point.

> I'd have to pull the card to look at the chip, but StarTech says it's a B. 
> I can believe that.

Agreed - if your card was claimed to have a XIO2213B I have no reason not to
believe that.

> That all being said even without theaded irq handling I'm still seeing rock
> solid low latency. This is amazing.

It's a good outcome.

> Thanks for all the help guys!

Not a problem.  I'm pleased you have a configuration which works solidly for
you.

Regards
  jonathan