Open GoogleCodeExporter opened 8 years ago
From the description it sounds like I've got this same problem on 1.5. It
*used* to work for me without a problem about 18-24 months back on slower
hardware than I have now, so maybe a newer GCC/glibc/kernel/whatever broke it?
I'm using an au8830 sound card if that's significant, nothing else gives me
sound problems though.
No combination of audio plugin settings helps, I've tried changing the system
clocksource from tsc to hpet, turned off cpufreq, set the process affinity to a
single core, even tried changing the CFLAGS to something more conservative.
Original comment by flussence
on 4 Jul 2011 at 9:53
Just a thought: could this be caused by the glibc memcpy problem?
(https://bugzilla.redhat.com/638477)
Original comment by flussence
on 18 Jul 2011 at 8:45
I would doubt that it is this problem - at least not in mupen64plus. But I
could easily be wrong. You can test it by using memcheck from valgrind and
search for the failure as explained in
http://valgrind.org/docs/manual/mc-manual.html#mc-manual.overlap
You should enable debug symbols and disable strip when you build mupen64plus:
$ ./m64p_get.sh
$ ./m64p_build.sh DEBUG=1
$ cd test
$ valgrind --tool=memcheck ./mupen64plus /PATH/TO/TEST/ROM
Original comment by s...@narfation.org
on 19 Jul 2011 at 7:10
You could also test to enable performance governor and to pin m64p to a core:
$ sudo sh -c "echo performance >
/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor"
$ numactl -C 0 -- ./mupen64plus /PATH/TO/TEST/ROM
Original comment by s...@narfation.org
on 19 Jul 2011 at 7:20
valgrind doesn't find any problems in sound code. Complains about things inside
GTK and SDL-input/gfx, but that's all.
Setting cpufreq to performance does make the emulator run a lot better, but
like I mentioned already it doesn't make a difference with this specific
problem.
It's almost like the emulated CPU is having these sound dropouts, because they
sync up with how busy the screen is...
Original comment by flussence
on 19 Jul 2011 at 9:26
Can you try to remove the new_vi delay and try again (be aware that some stuff
will now ran faster then expected)?
Original comment by s...@narfation.org
on 30 Jul 2011 at 11:24
Attachments:
No difference there either :(
Original comment by flussence
on 30 Jul 2011 at 12:24
Issue 433 has been merged into this issue.
Original comment by s...@narfation.org
on 9 Sep 2011 at 11:20
Please try again with the code from
https://bitbucket.org/ecsv/mupen64plus-audio-sdl (don't forget to build
against libsamplerate) and change the RESAMPLE setting to "src-sinc-fastest"
Original comment by s...@narfation.org
on 24 Oct 2012 at 6:24
Original issue reported on code.google.com by
vash...@gmail.com
on 12 Jun 2011 at 9:51