Menu
▾
▴
#18 Fuse emulation speed multiplies when sound is enabled.
On Fuse 0.6.2.1 on my machine, whenever sound is
enabled (and emulation speed is set in the preferences
at 100%), actual emulation speed goes to 200% or even
sometimes 400%. Setting any other emulation speed
doesn't have a multiplication problem (perhaps because
any other rate disables sound).
I'm on Linux 2.4.25 with an Athlon 2500+. My big
suspicion is that it might be something to do with my
sound card, which is integrated into my Gigabyte
7NX400Pro2 motherboard (Realtek ALC650/655 Audio AC'97
Codec, I think) driven by the NVidia provided driver
nvaudio 1.0-0261)
I also experienced the same problem with fuse 0.6.2
Logged In: YES
user_id=11017
It is almost certainly your soundcard driver (which Fuse relies on by
default to control speed).
Try compiling with the SDL UI and seeing if that makes your problem go
away...
Logged In: YES
user_id=29214
Hmmm.... you're not the first person to report this problem,
although its obviously not something I can reproduce :-(
Just to collect some more statistics here: what distribution
are you using, and do you know if you're using ALSA or OSS
for your sound?
If I get time (certainly not before the end of the weekend),
I'll try and write a small test program which outputs (say)
middle C to the sound card to see if we can track down
what's causing this problem. Of course, if anyone else wants
to do it before me :-)
Logged In: YES
user_id=160296
Compiling Fuse-0.6.2.1 with the SDL interface (--with-sdl)
does indeed solve the problem (emulation runs with sound at
correctly at 100%).
Logged In: YES
user_id=160296
Philip, my distribution is technically Redhat 7.2 but I only
ever install stuff built from tarballs so RH is now just the
compost at the bottom (it's kind of like a Linux From
Scratch distro, except not as good).
AFAIK, I'm using the kernel sound system (the OSS ish one)
with the nvaudio driver, which according to nvidia docs is
based off the kernel's i810_audio driver.
Logged In: YES
user_id=1135765
I have the same problem. I am using fuse 0.6.1 on Gentoo.
I have tried to compile both with and without SDL and the
problem does not go away.
I am using kernel 2.4.26 with the regular OSS drivers.
Logged In: YES
user_id=1135765
Another interesting note: spectemu has exactly the same problem.
Logged In: YES
user_id=1135765
More on this issue:
It turns out that with SDL the problem DOES go away. I had a
problem with Gentoo USES flags not enabling SDL correctly.
Also, I have tried version 0.7.0 and I have seen that the
problem is still there. However, if I check the sound option
"beeper pseudo stereo", then everything works fine (I get
sound at 100% speed). What is puzzling is that if I disable
this option, after having enabled it previously, everything
continues to works fine. In version 0.6.2, fiddling with
this option did not make any difference.
Logged In: YES
user_id=11017
As far as we know this problem is only relevant for people
using the OSS routines (mostly Linix + GTK+ user interface,
sound on). Could people with those problems tell us:
1) Their Linux distribution + kernel version
2) Their sound harware (e.g. Soundblaster Live!, NForce
motherboard etc.)
3) Their sound drivers in use (e.g. OSS vs. ALSA)
Also note: If you switch a build of Fuse from GTK+ to SDL,
you must make clean before make or you will have problems!
Logged In: NO
Gentoo Linux.
Linux 2.4.26 compiled with OSS drivers (NOT alsa).
This is a Dell Inspiron 8200, the sound chipset is integrated
in the Intel mother board (the kernel module is the i810_audio).
Logged In: YES
user_id=294680
Have you tried 2.4.27? There are some i810 OSS fixes in
there (see
<URL:http://www.kernel.org/pub/linux/kernel/v2.4/ChangeLog-2.4.27>).
Failing that, does ALSA's OSS emulation on that hardware
work well?
Logged In: YES
user_id=11017
Could people experiencing this problem please try the patch
below to sound/lowlevel.h?
Logged In: YES
user_id=1135765
I tried the patch. Here is what happens:
Now the speed is always 67% (with or without sound enabled:
it doesn't make a difference any more).
This effect scales: if I set the speed to "200%" in the
options menu, the resulting emulation speed is "133%".
Logged In: YES
user_id=1135765
Another note: I tried the patch in version 0.7.0
Logged In: YES
user_id=11017
Could you try adding patch 1050741 from the patch section of
the website as well?
Logged In: YES
user_id=1135765
I can't apply that patch to fuse 0.7.0 or 0.6.2. I get the
following error:
$ patch -p0 <../timer_fix.patch
patching file rzx.c
Hunk #1 succeeded at 90 with fuzz 2 (offset 3 lines).
Hunk #2 succeeded at 156 with fuzz 1 (offset -4 lines).
Hunk #3 succeeded at 354 (offset 16 lines).
can't find file to patch at input line 101
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|diff -u -r1.11 sdlsound.c
|--- sound/sdlsound.c 27 Jun 2004 01:11:34 -0000 1.11
|+++ sound/sdlsound.c 20 Oct 2004 13:05:15 -0000
--------------------------
If I try to give this file manually (and some others that
give the same error later), I get compile errors down the road.
Also, I think it might be better if we continue this over
email (febertra at indiana . edu).
Logged In: YES
user_id=29214
The patch is against CVS HEAD.
Logged In: YES
user_id=1135765
Partial success.
This is what I did: I tried to apply the timer patch to the
cvs tree and apparently it was already applied so I compiled
and ran without any patching. As expected, the speed problem
was there. On the side, I had a linking error about -lX11
not being found but I edited that by hand and linked ok.
Then I applied the oss patch below and compiled again. This
time things turned out more or less ok: sound worked and
speed was at a steady 100%. However, a new issue appeared:
at random intervals (~1 sec or less) the speakers would emit
a short cracking noise (similar to what you get when you
turn on and off a speaker).
Logged In: YES
user_id=11017
Thanks for trying that, the sound glitches you mention may
mean you need a bit more latency to cover for jitter in your
process scheduling (or machine load or something else),
could you let me know if your machine is unloaded?
As regards the linking problem, could you make a new bug
report with your config.log and original Makefile, so I can
see what went wrong?
In any case, try changing this line in timer.c:
if( difference < ( ( 2.0 * speccy_frame_time_usec /
1000000.0 ) - 0.005 ) ) {
to
if( difference < ( ( 3.0 * speccy_frame_time_usec /
1000000.0 ) - 0.005 ) ) {
Logged In: NO
All the tests were performed with the machine unloaded. I
have been timing the "glitches". When I leave the emulator
running without touching anything, they happen exactly every
two seconds. Making the modification that you suggest does
not make any difference. (I have even tried other values for
that constant and it still does not make any difference).
Changing the timer_sleep_ms(10) line to timer_sleep_ms(100)
does make a difference in the interval between glitches but
it does not make them go away.
Logged In: YES
user_id=11017
Originator: NO
Does anyone still see these issues with current Linux kernels?