Re: [fluid-dev] Major degradation in sound quality & cpu usage going from Ubuntu 11.04 to 11.10

[Aere Greenway]

David:

I've converted my machine to xubuntu (xfce desktop), which is like the ultimate low-fat option.  Everything runs fast and lean.  Some things (like applying system updates) seem to run visibly faster.  

I had no problems memory-wise with your PPA version of Fluidsynth.  I can run the two Qsynth 'engines' (with the large soundfont), as well as loading the same soundfont into the Soundblaster card, and running Evolution mail at the same time as playing using Qsynth via Rosegarden.  

Not a hint of memory problem, and even Evolution mail fetching e-mail while the music is going on didn't yield a single under-run in Jack.  In Ubuntu, the fetching of mail would usually cause an underrun, so I had to take down Evolution mail while I do music.  

No such problem with xubuntu.  

Everything works just fine - even my two Java development environments, and Virtual Box (for when I use Windows) seems to work better too.  

I think Xubuntu is the cure for successfully avoiding the Unity desktop (A.K.A. Linux Vista).  

- Aere


On Mon, 2011-11-07 at 11:43 +0100, David Henningsson wrote:
> On 11/05/2011 09:12 PM, Aere Greenway wrote:
> > David, et al:
> >
> > I have not heard back since I reported on performing the test you asked
> > me to run.
> >
> > I noticed on the website that there are "0 new bugs".
> >
> > What process do I need to go through to have this problem evaluated as
> > to whether it is officially a 'bug' or not?
> >
> > To me, this problem kills my future prospects. The music education
> > package I am poised to put out depends on FluidSynth (actually Qsynth,
> > which uses it).
> >
> > I could distribute my package, and people could get all excited about
> > it, but when they installed it themselves (instead of using what comes
> > with the package), they would find that the installed version won't even
> > play the demo-music, even on a fast machine.
> >
> > What I had to go through to compile the older version on the current
> > level is not something I could expect a new user of Linux to do.
> >
> > There is no path forward if this cannot be fixed. But of course, those
> > are the chances I take when I attempt to develop something.
> >
> > It would be very helpful if I had some indication of whether or not this
> > problem will be addressed, or even looked at.
> >
> > I am willing to supply MIDI and audio files you could test it with, and
> > verify any fixes.
> >
> > Please consider the idea that what I have been doing in the past to
> > avoid under-runs, by configuring fewer simultaneous voices (setting the
> > polyphony parameter to 64, or even 48 on a slower machine), may be a
> > more elegant solution than what has apparently been done in the latest
> > release (probably to address that same problem of under-runs).
> >
> > I think my planned product could make a big difference in the way people
> > learn to play (and compose) music, and I am really hoping FluidSynth can
> > be a part of it. It certainly has been a rock-solid, dependable
> > component up to now.
> >
> >
> > Sincerely,
> >
> > Aere Greenway
>
> Hi Aere,
>
> I've been on travel with limited time/possibility to help you. In 
> addition, the rebuild I asked you to do was harder than I expected.
>
> But now I'm home again, and here's what I've done now:
>
> I've taken the 1.1.5 package, and added the patch I committed as r435. 
> This is because you said something earlier about sound working up to the 
> polyphony limit.
> I then uploaded the source package to a ppa: 
> https://launchpad.net/~diwic/+archive/fluidsynth-test
>
> The ppa probably has finished building when you read this, so can you 
> download/install the packages from there and see if it resolves your 
> problem? Thanks!
>
> // David

-- 

Sincerely,
Aere