[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[iiwusynth-devel] Optimization in plain text
From: |
markus.nentwig |
Subject: |
[iiwusynth-devel] Optimization in plain text |
Date: |
Wed, 20 Feb 2002 13:14:39 +0200 |
(I'm sending the same mail again in plain text format here - sorry for the spam)
Hello,
first of all, thanks for iiwusynth - it works very well. I found a 12 MB Rhodes
EP-soundfont, which just blows away any synthesizer-fake-Rhodes I have ...
Some notes (should be 'some patches', maybe after the weekend :-)
- Compiler optimization: After upgrading to the latest GCC version, I managed
to compile with -O3 -funroll-loops -mcpu=i686 -march=i686.
At least the -O3 is needed for handling full polyphony (on a P III 700). Not
sure about the i686 options, though.
With previous compiler versions, strange behaviour resulted from -O3 (the root
key of each sample was missing, but that's a compiler bug).
- Bug: There should be a pitchbend default value for each Midi channel.
Currently pitch bend starts with +/- 0 for each generator, instead of the real
position of the pitch bend wheel.
- Filter: The filter -is- commented out at the moment? I'm not sure, I didn't
manage to set the filter. I'll try to find some time in the weekend to
understand the code...
- Equalizer: I have implemented a 10 band equalizer in C, that works just fine
(http://millisampler.sourceforge.net). It's based on a 'recipe' from a DSP
application note.
The CPU load with all 10 bands on is around 10 % on my machine. Turning off
bands (gain = 0 dB) reduces the load proportionally.
While I don't think that it could be included in iiwysynth in a way compatible
with the standard, It might be useful as a 'master effect', which is
only controlled via command line parameters. Any opinions?
-Sysex: Iiwusynth locks up, when sysex messages are transmitted.
Regards
Markus
address@hidden
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [iiwusynth-devel] Optimization in plain text,
markus.nentwig <=