Re: [hatari-devel] Falcon emulation accuracy

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

Hi,
> > - DSP/host port timing is unclear but it seems like the Hatari port is
> > slower, perhaps as a compatibility helper to make up for the faster CPU
> > execution for uncached code. (R:TexStream is solid contiguous writes to
> > the DSP). The DSP timing is accurate, but timing of exchanges over the
> > host are not (R:VisPlanes also does a lot of host port read access, and
> > seems much quicker on HW).
> Laurent?

From memory, I think I've added a timing of 4 cycles per byte access to 
the dsp hostport.
This is probably not completly accurate and could be easily tuned, but 
to tune it, the rest of hatari should be better tuned before, else it 
won't change many things.

> > - FPU timing doesn't seem emulated at all (but I could be wrong about
> > that) so Hatari is quicker.
> That's also (very) visible in Atari benchmarks.  Laurent?

That's true. MMU timings either are not emulated. They both eat 2 cycles 
per instruction (if this could be the case of the real machine... ;)
there's just to add the timings into the falcon_cycles.h file in the cpu 
repository.

All of this is probably better emulated in the new core.
I'm waiting for this before changing anything.

Regards

Laurent