On 01/29/2013 02:59 PM, Aere Greenway wrote:
> On 01/29/2013 01:55 PM, D. Michael McIntyre wrote:
>> I don't have a clue why Rosegarden is causing this and MusE isn't. Upon
>> a superficial and quick comparison, they both appear to work the same
>> way in the case where QSynth is already running, and it's configured to
>> use pulseaudio: they both start with audio support disabled, due to
>> being unable to start jackd, because pulseaudio has control of the
> Michael, and all:
> Perhaps that is a significant clue to the problem.
> JACK seems to have made great strides lately. Where it used to not
> co-exist with anything, it now seems to get along with most everything.
> You can now play Internet videos with sound while JACK (QjackCtl) is
> running, and can run all sorts of things together with it. On the
> other hand, Audacity doesn't get-along with it at all.
> If you look in the QjackCtl "Connections" dialog, in the "Audio" tab,
> you now see a "PulseAudio JACK Sink" and a "PulseAudio JACK Source"
> device. And the ubiquitous hangs seem much harder to cause.
> I had concluded, based on my testing on several desktop computers in
> my test-bed, that they had finally solved all of the problems, and
> that JACK nicely gets along with everything. They seem to have even
> put those fixes in the Long-Term-Support release, 12.04, because it
> behaves just as well.
> But then I tried the same test on my Acer Aspire laptop, and my new
> MIDI device initially trying to use the (PulseAudio) Java Sound
> synthesizer (with QjackCtl already running), hung it tighter than a
> drum, and I had to power the machine down.
> So I guess all is not yet well on the JACK audio front.
> The significant clue in what you said, was that Rosegarden would fail
> to start jackd, and would thus start up without audio support.
> In doing the test, there was no warning about audio support being
> disabled (that I in the past always used to see in such situations).
> Where they now co-exist better, it probably succeeded in starting
> jackd. But perhaps, they don't successfully co-exist in other ways.
> As you pointed out, running with JACK is much more efficient. Where
> processor usage was around 75% using PulseAudio, the same piece played
> using JACK had a processor utilization of around 25%, which is no
> insignificant difference. To even get it to work without cutting-out
> on PulseAudio, I had to disable both the Reverb and the Chorus. I can
> use them when I use JACK in place of PulseAudio.
> By the way, most of my testing was done on an Intel Celeron CPU 2.53
> GHz desktop machine, running Lubuntu 12.10.
The information in the next-to-last paragraph above (about the processor
utilization using PulseAudio and JACK) was incorrect.
The numbers I remembered were correct, but the numbers were not from the
When I revisited this test, using the same piece, the processor
utilization was 75% (for PulseAudio), and 55% (for JACK). The
difference is still significant, but not as great as I formerly reported.