Re: [yoshimi-user] progress report
- From: cal <cal@xxxxxxxxxxxx>
- To: yoshimi-user@xxxxxxxxxxxxxxxxxxxxx
- Date: Thu, 05 Aug 2010 07:14:43 +1000
On 05/08/10 04:21, Will J Godfrey wrote:
[ ... ]
I'm certainly up for some testing. Is there anything specific you want
checked out or just a general tweak everything?
P.S.
A link to where this beastie is might be a good idea :)
Thanks Will, (and Jeremy too!). I think I spoke perhaps just a moment
prematurely.
Yesterday was supposed to the crowning moment when I finally pulled it all
together,
but it didn't quite work out that way. But I now have a plan, where I didn't
have
one before.
The performance impact of breaking down the audio generation down into short
blocks,
with midi processing between the blocks has seriously astonished me. It's
similar in
scale to what I was seeing with the boost experiment. When you see the RT field
in
qjackctl running a steady 80%, peaking to 90%, and still holding it all
together,
the sense of "Am I completely deluded and deranged?" has to be considered
carefully.
I think my eager little blurt to the list came from a need for assurance that I
wasn't
(completely).
Yesterday was indeed a disaster. What I'm doing now is applying the "known good
bits"
from the experimental branch to the basic 0.058.1 codebase. A tedious business,
but
really the only way I'll ever get on top of the strange and mysterious
segfaults I've
introduced by poking too many things at once. I figure if I can pull it all
together
by the weekend I'll be doing well, and that's the plan.
cheers, and thanks!
oh ... and Unison is back in there too.
Other related posts: