Hi Jonathan -
I look forward to the kernel improvements. The DLLs make the ffado code quite complex.
Something I noticed when I was digging through the code is that I couldn't find anyexplanation about what the "cycle timer" and "watchdog" are for. It would be great ifsomeone who knew added a description to these parts of the code.
> Date: Wed, 2 May 2012 10:28:20 +0930
> From: jwoithe@...
> To: stefanr@...
> CC: ffado-user@...
> Subject: Re: [FFADO-user] Warning messages, more on variable latency
> On May 01 Kristian Amlie wrote:
> > On 05/01/12 03:13, Jonathan Woithe wrote:
> > >> On rare occasions the latency on my machine still changes by 16 frames or
> > >> a factor of 16, but its pretty rare and seems to be more likely on my
> > >> machine when it is cold.
> > >
> > > That temperature dependency is strange. Although it's possible not
> > > feasible, it would be interesting to discover the cause.
> > I have seen that temperature dependency on my system too, although for
> > me it crashes. I have both an Edirol FA-101 and an AudioFire2, and they
> > have both been affected. Like Ben, it's quite rare, even when it's cold,
> > but when it's warm it never ever happens. Just saying, so you know that
> > Ben is not the only one.
> Thanks for the information - it's good to know that the problem isn't
> isolated to a single PC.
> On Tue, May 01, 2012 at 08:21:07PM +0200, Stefan Richter wrote:
> > Maybe the function of one or another oscillator in the PC is influenced a
> > lot by temperature. ...
> It's possible. Certainly most electronic components have some degree of
> temperature dependency, but obviously in most cases the effects of this
> aren't visible to the user. As you stated, perhaps there is an oscillator
> that's not quite stable until it warms up - it wouldn't be impossible, and
> admittedly FFADO is probably one of few applications where you'd see the
> effects (mainly because FFADO effectively pits the computer's local
> oscillators against those in the audio devices).
> > Whether that then just influences roundtrip latency or causes ffado's
> > streaming engine to go down entirely might among else depend on how well
> > the engine and the FireWire link layer controller work together.
> Yes, either that or the magnitude of the temperature effect.
> I'm not sure there's a whole lot more we can do with this issue right now
> unless Ben or Kristian decide to launch a low-level exploration of their
> systems. It seems we may have to accept this temperature effect on some
> systems for the moment.
> As an aside, I suspect that when the streaming system moves to the kernel (a
> long term work in progress) it wouldn't surprise me if this problem goes
> away. I think that if there is an oscillator temperate dependency then that
> would be messing with FFADO's DLLs. The kernel engine won't use DLLs to
> anywhere near the same extent as to what they're used now, and this would
> tend to insulate FFADO from the subtle effects we're hypothesising. It will
> be interesting to see.
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> FFADO-user mailing list