[QLab] Major problems with a combination of AU Lab and a FW 410

[Jarrett Krauss]

I know this isn't Qlab specific, but everyone on the listserv has always
been really helpful, so I'll tell you about my situation.

I'm designing this show, and playback is with a Qlab system, with all cues
patched to soundflower 16ch. I created an aggregate device that utilizes the
inputs of soundflower 16ch and the outputs of the FW 410. Neither the fw 410
nor soundflower has the resample button checked, and the clock source is the
fw 410. In AU Lab there are 4 Audio Units in play and a few faders were
moved around. I have a grand total of 8 ins and outs in AU Lab, routed
one-to-one. All cues are sampled at 44.1K, the internal clock for the 410 is
set to 44.1 and the aggregate device is set to 44.1. I'm running all of this
on a Mac G4 with under 1 ghz processing and 512 mb of ram (not so good, I
know). Qlab is fully updated and the FW410 drivers are only one update
behind (but the new update doesn't fix anything for the FW 410). All seemed
well.

The first two days of tech go by flawlessly. I am intrigued by essentially
having a digital console within my computer that will execute parameter
changes via midi from Qlab.

Then 3 days into tech, in the middle of the show, a cue was executed and it
sounded like three terrible things at once, truncated samples, distortion
and a duplication of the signal. The temporary fix was to push the reset
button in the software part of the FW 410. I also noticed that in AU Lab
there was a warning at the bottom that said CPU overload.

After some experimentation I was able to reproduce the problem by taking
processing power away from the aggregate device in AU Labs settings. So I
increased the dedicated processing power for the aggregate device from 68%
to 84%.

Then the next night it happened again, on an entirely different cue. The
reset button once again solved my problem temporarily. I went back to the AU
Lab preferences and increased dedicated processing power on both the
soundflower 16ch and the firewire 410 separately. FW 410 had only 22%
dedicated processing power at that time, so I assumed that that was the
problem. While I dedicated plenty of processing power to the aggregate
device, I did not give enough to the individual pieces of the aggregate
device.  All is well... Hopefully.

Then the next night, final dress, it happened not once but TWICE, at two
different points in the show. Now the sequences that it is messing up on are
not particularly demanding, maybe 6 steps maximum, and on this same computer
I built a cue that had 67 steps all firing simultaneously, and after a brief
screen freeze, it worked fine. I feel like all my problems are either within
the aggregate device or AU Lab.

So tonight, I am doing away with AU Lab, doing my EQs on the channel strips,
installing a delay unit and a compressor.

Now while that may solve the problem, the mystery still remains... What is
going on here?

Thanks in advance, Qlab lovers.

Jarrett Krauss