[hatari-devel] DSP togging

Nicolas Pomarède npomarede at corp.free.fr
Wed Jan 12 23:24:27 CET 2011


Le 12/01/2011 22:50, Eero Tamminen a écrit :
> Hi,
>
> On keskiviikko 12 tammikuu 2011, Laurent Sallafranque wrote:
>> Btw. Why toggling WinAUE CPU cycle accuracy would need reset:
>> #if ENABLE_WINUAE_CPU
>>           /* Did change CPU cycle exact? */
>>           if (changed->System.bCycleExactCpu != current-
>>
>>> System.bCycleExactCpu)
>>
>> I've implemented quickly this part of code with WinUae CPU.
>> There's still some work to do on this CPU.
>
> Yes, but why toggling cycle acccuracy needs emulation reboot with
> WinAUE core although it's not done with AUE core?
>
> Is there some extra state that's actually visible inside emulation?

This should be checked by carefully reviewing the code. With current 
uae, I think this more or less give a better prefetch for 68000 and some 
better timings for exceptions. So toggling accuracy should not give any 
problem and should not harm running programs.
WinUae has a very more precise accuracy mode, it's possible it needs to 
be reseted.

>
>
>> I wish we could all work on this for the next 2 month and remove
>> definitively the old uae core.
>
> If old UAE core is noticeably faster I think it makes sense to retain
> that as long as it works fine...
>
>

I also think current UAE core should remain, even when winuae is 
correctly completed. There will be a lot of regression test to do, and 
only if we release a version that can still have the 2 cores will we be 
able to compare difference.
Lots of things (stf/ste) are working with current core, as long as we 
can maintain a compilable version, I don't see a need to remove it.

Nicolas



More information about the hatari-devel mailing list