Thanks to Christian and Aere. Here are some more results on a file nightsin.kar of various choices using pidstat:
Version
|
flags
|
sound font
|
cpu %
|
memory %
|
comments
|
1.1.5
|
none
|
FluidR3_GM
|
75%
|
43%
|
unlistenable
|
1.1.5
|
-z 4096
|
FluidR3_GM
|
70%
|
43%
|
unlistenable
|
1.1.5
|
-z 4096
|
GeneralUser
|
76%
|
20%
|
unlistenable
|
1.1.6
|
none
|
FluidR3_GM
|
80%
|
40%
|
still distorting
|
1.1.6
|
-z 4096
|
FluidR3_GM
|
75%
|
40%
|
distorts when usage > 90%
|
1.1.6
|
-z 4096
|
GeneralUser
|
65%
|
none shown
|
distorts when usage > 90%
|
I haven't tried Aere's suggestions yet. Overall, 1.1.6 sounded better than 1.1.5, but that may be the result of the options compiled in.
Using pidstat with a resolution of 1 second I got the following histogram (using octave) of the CPU usage for the last case, 1.1.6 with GeneralUser soundfont. I know CPU usage of 300% is probably a bit exaggerated but the file didn't sound too good at that point! Are there standard MIDI test files for isolating what factors cause what effects?

Cheers
Jan
--
On Sat, 2012-11-17 at 08:09 -0600, S. Christian Collins wrote:
Try GeneralUser GS FluidSynth version from this site. It is much smaller than FluidR3, so it should hopefully use less CPU.
-~Chris
On 11/17/2012 04:25 AM, Jan Newmarch wrote:
Recently there was a thread about 1.1.5 on an RPi. I'm picking that up
for 1.1.6. I have a Model B RPi with 512M RAM. I have the Raspbian hard
float image installed rather than the Debian soft float (which behaves
worse). I have applied the security fixes suggested by Aere Greenway.
I'm running it with period set by -z 4096 as suggested by David
Henningsson. I'm observing CPU usage by running top in a separate
window:
fluidsynth -a alsa -l -z 4096 /usr/share/sounds/sf2/FluidR3_GM.sf2 \
youvemad.kar
Some simple MIDI files play okay. Others hit CPU usage between 40-99%,
varying throughout the song. At 99% CPU the sound is very broken up, too
bad to be really playable.
It has been suggested that smaller sound fonts will help. What smaller
sound fonts are there?
I can do more experiments if desired.
Jan
_______________________________________________
fluid-dev mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/fluid-dev
|