Re: [hatari-devel] TT emulation sound setting experiment

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 25/11/2017 à 23:40, Nicolas Pomarède a écrit :
> Le 25/11/2017 à 23:28, David Savinkoff a écrit :
> > ----- On Nov 25, 2017, at 1:30 PM, Troed Sångberg wrote:
> >
> > > Did you try it yourself (the game Murder)? The below makes no 
> > > difference when I
> > > test with my latest macOS dev builds. My i7 is not taxed by the 
> > > emulation. The
> > > sound does not strike me as what it would sound like if it was an 
> > > overloaded
> > > CPU, but I can't be absolutely certain.
> > >
> > > /Troed
> >
> > Hi Troed,
> > I didn't test the game because I've got a slow 32-bit intel core2 cpu.
> > Given your test result, you're probably right.
> > Maybe someone with a fast computer can test the TT and provide
> > reports.
>
> Hi
>
> I quickly tried some sample playing with TT mode and it seems indeed to 
> sound wrong, even on a PC with fast enough CPU.
>
> This is strange because if I play same sample in STF with 8, 16 or 32 
> MHz (which was my initial test some times ago to check cpu ratio over 
> MFP freq) sound is correct, so there might be sthg else.

Hi

this sound issue was fixed in latest commits from yesterday ; it was 
mainly due to how cpu freq ratio were computed to speed up x2 or x4 the 
base 8 MHz clock and get a 16 or 32 MHz emulated freq.

The values were wrong for machines where the original clock was not 8 
MHz (ie Falcon or TT) ; this resulted in wrong sound when playing sample 
with YM2149 (but sound was correct with chip music changing YM only once 
per VBL, which explains why it was never noticed before)

Nicolas