Re: [hatari-devel] DMA sound and Falcon

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

Hi,

Thanks Nicolas, That's fixed now.

I've noticed that sometimes sound unsynchronize and then later, 
resynchronize (this generates noise and cracles in the sound).
With 1.7, if the sound was correct at the beginning, it stayed correct, 
and if it was noisy at the beginning, it stayed noisy.


For your info, some warnings :

[ 87%] Building C object src/CMakeFiles/hatari.dir/video.c.o
/home/laurent/Atari/hatari/src/video.c: In function ‘Video_InitTimings’:
/home/laurent/Atari/hatari/src/video.c:818:7: warning: unused variable 
‘i’ [-Wunused-variable]
  int  i;
       ^
/home/laurent/Atari/hatari/src/video.c: At top level:
/home/laurent/Atari/hatari/src/video.c:990:13: warning: 
‘Video_InitTimings_Round’ defined but not used [-Wunused-function]
 static void Video_InitTimings_Round ( VIDEO_TIMING *pSrc )
             ^
/home/laurent/Atari/hatari/src/video.c:658:13: warning: 
‘Video_Sync_SetDefaultStartEnd’ declared ‘static’ but never defined 
[-Wunused-function]
 static void Video_Sync_SetDefaultStartEnd ( Uint8 Freq , int 
HblCounterVideo ,
             ^
[ 88%] Building C object src/CMakeFiles/hatari.dir/wavFormat.c.o


Regards

Laurent




Le 11/09/2016 à 14:09, Nicolas Pomarède a écrit :
> Le 09/09/2016 à 00:12, Nicolas Pomarède a écrit :
> > Le 09/09/2016 à 00:07, Laurent Sallafranque a écrit :
> > > Hi,
> > >
> > > I think the Falcon DMA sound runs twice as fast as it should.
> > >
> > > I've downloaded the latest source and recompiled it from scratch.
> > >
> > > I've tested with 2 programs on both the latest version and the 1.8
> > > version.
> > >
> > > The sound plays 2x the normal speed.
> > >
> > > I've tested with mouse.prg (a little animation demo with a cartoon 
> > > mouse
> > > and a looping sample played with the DMA).
> > >
> > > Regards
> > >
> > > Laurent
> >
> > Hi
> >
> > do you have a link to download this small demo ?
>
> Falcon sound playing twice too fast has been fixed (this part was not 
> correctly updated after the big commit to count cycles in rev 6305)
>
> Nicolas