[hatari-devel] Higher precision when generating samples -> use nanosleep

[Nicolas Pomarède]

Le 27/01/2011 09:16, David Savinkoff a écrit :
> Hi Nicolas,
>
> I've pondered the discussion and looked at the code and now
> I'm synchronized with what is going on.
>
> Here is a new patch that is intended to be incremental and
> supersede my previous patches.
>
> This patch addresses:
> - Required boundary checking in your vbl/sample-rate rounding.
> - Using the least common multiple (LCM) of 71 60 and 50 Hz.
>
> This patch does not address my continuous clock cycle
> correcting because I have yet put it into Main_WaitOnVbl().
>
> I've tested this patch and am confident that the LCM and
> VBL/sample-rate rounding is working well at 60Hz because
> the buffer drifts extremely slowly. About what you would
> expect with 2 crystals.
>
> David

Thanks for the patch ; I will try to do some experiments during next 
days with nanosleep and merge both approach to get the best possible 
results.


Nicolas