Re: [hatari-devel] UAE core regressions in Falcon demos

[laurent.sallafranque@xxxxxxx]

Hi,

According to a previous thread I exchanged with Anders Eriksson : [hatari-devel] First 68060 demo running from beginning to the end

According to Anders, The demo Eskimau uses intensively the FPU. 
This demo works with hatari (slowly of course and with a distorded sound), but it works from beginning to the end.
There are some graphical glitches in the tridi mapped screens that may be due to the FPU code in hatari.


Another point: When I was working on Magic emulation, it seemed that there are some registers specifics to the 68881/68882 that are not present into iomemtabfalcon.c
Maybe we could add them, like for the DSP, by testing if the FPU is present or not, and adding/removing them in the iomemtab table ?


Regards
Laurent


----- Mail Original -----
De: "Eero Tamminen" <oak@xxxxxxxxxxxxxx>
À: hatari-devel@xxxxxxxxxxxxxxxxxxx
Envoyé: Lundi 27 Février 2012 22h59:34 GMT +01:00 Amsterdam / Berlin / Berne / Rome / Stockholm / Vienne
Objet: Re: [hatari-devel] UAE core regressions in Falcon demos

Hi,

On perjantai 24 helmikuu 2012, Laurent Sallafranque wrote:
>  > centurbo benchmark:
> It gives good results for me for DSP (it detects 32 Mhz) and 68030 (it
> detects 14 Mhz).
> For the FPU, it detects 190% (it the second benchmark that gives this
> result).
> It's normal, I haven't yet added the correct cycles for the FPU and the
> MMU.

Btw. According to this mail:
	http://lists.debian.org/debian-68k/2011/04/msg00021.html

At least in Aranym, the FPU emulation is "broken", not just
non-cycle accurate.  I wonder is WinUAE emulation any better...


	- Eero