[hatari-devel] Little question with relative_interrupts
Laurent Sallafranque
laurent.sallafranque at free.fr
Wed Sep 16 23:07:08 CEST 2009
OK, thanks
Thomas Huth a écrit :
> On Wed, 16 Sep 2009 00:35:23 +0200 (CEST)
> npomarede at corp.free.fr wrote:
>
>
>> On Wed, 16 Sep 2009, Laurent Sallafranque wrote:
>>
>>
>>> Hello,
>>>
>>> While reading code in int.c and int.h (for the crossbar), I asked
>>> me the following question :
>>>
>>> Extract from int.c
>>>
>>> Thanks to Arnaud Carre (Leonard / Oxygene) for sharing this method
>>> used in Saint (and also
>>> used in sc68).
>>>
>>> Conversions are based on these values :
>>> real MFP frequency is 2457600 Hz
>>> real CPU frequency is 8021247 Hz (PAL european STF), which we
>>> round to 8021248.
>>>
>>> Then :
>>> 8021248 = ( 2^8 * 31333 )
>>> 2457600 = ( 2^15 * 3 * 5^2 )
>>>
>>> So, the ratio 8021248 / 2457600 can be expressed as 31333 / 9600
>>>
>>>
>>> Are these values still OK for a falcon (where the CPU is ~16 000
>>> 000 and not 8 021 248 ?
>>>
>>> If no, would this explain that some demos/games need to run at 32
>>> Mhz ?
>>>
>> yes, the current MFP code is adapted to a 68000 cpu and an MFP as
>> found in an STF/STE.
>>
>> Some factors should be applied when the cpu freq is different from
>> 8021247 Hz.
>>
>
> int.c always thinks that the CPU runs at 8 MHz. To emulate higher CPU
> frequencies, we simply divide the CPU cycles by two or four in
> M68000_AddCycles(). This should have the same effect as multiplying the
> frequencies by two or four.
>
> Thomas
> _______________________________________________
> hatari-devel mailing list
> hatari-devel at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/hatari-devel
>
>
>
More information about the hatari-devel
mailing list