Re: [hatari-devel] EXA demo Entracte : Improvement in search

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

> To me this simply looks like a CPU speed test, counting the number of
loop iterations between two VBLs. I'd assume that the demo requires a
certain CPU speed to run correctly and so it wants to quit on (for
example) over-clocked CPUs.

> I think the 68030 cycle accuracy of Hatari is not good enough here for
this test to pass.

Thanks Christian, this confirms what I suspected.

So now, we've got at least 4 programs that rely on accuracy

- EXA demo "Evolution" for the CPU speed testing
- EXA demo "Entracte"  for the CPU speed testing
- DLM Badmood for the HUD at the bottom of the screen (rely on the HBLs 
to break the screen adress to another area)
- Pinball Dreams, again for HBL accuracy value to break the HUD pinball 
screen from the table screen area

Regards

Laurent




Le 05/02/2017 à 11:48, Christian Zietz a écrit :
> Laurent Sallafranque schrieb:
>
> > The problem is in the 5 last lines : the code test for a min and max
> > value for register DO.
> > if D0 is not between $4100 and $4200, the code goes to $4eb1c  and this
> > quit the demo immediatly.
> To me this simply looks like a CPU speed test, counting the number of
> loop iterations between two VBLs. I'd assume that the demo requires a
> certain CPU speed to run correctly and so it wants to quit on (for
> example) over-clocked CPUs.
>
> I think the 68030 cycle accuracy of Hatari is not good enough here for
> this test to pass.
>
> > ; Here is the code of the compute D0 subroutine
> >
> > $0004eb54 : 2f38 fa06                          move.l $fffffa06.w,-(sp)
> > $0004eb58 : 41f8 fa07                          lea $fffffa07.w,a0
> > $0004eb5c : 7000                               moveq     #0,d0
> > $0004eb5e : 0188 0000                          movep.w   d0,0(a0)
> This disables the MFP interrupts. Probably so they won't interfere with
> the measurement loop below.
>
> > $0004eb62 : 41f8 0080                          lea       $0080.w,a0
> > $0004eb66 : f010                               DC.W      $f010
> > $0004eb68 : 4200                               clr.b     d0
> Note that your disassembler is wrong here, the last two lines should be
> pmove tc,(a0), i.e. reading the MMU's Translation Control register.
>
> > $0004eb6a : 0210 007f                          andi.b    #$7f,(a0)
> > $0004eb6e : f010 4000                          pmove     (a0),tc
> This disables the MMU address translation, maybe because it also could
> influence the loop?
>
> > $0004eb72 : 4278 0468                          clr.w     $0468.w
> > $0004eb76 : 4a78 0468                          tst.w     $0468.w
> > $0004eb7a : 67fa                               beq.s     $4eb76
> Wait for a VBL to occur, as you already wrote.
>
> > $0004eb7c : 4278 0468                          clr.w     $0468.w
> > $0004eb80 : 7000                               moveq     #0,d0
> > $0004eb82 : 4a78 0468                          tst.w     $0468.w
> > $0004eb86 : 6600 0006                          bne       $4eb8e
> > $0004eb8a : 5240                               addq.w    #1,d0
> > $0004eb8c : 60f4                               bra.s     $4eb82
> Wait for the next VBL, while incrementing D0.
>
> > $0004eb8e : 41f8 0080                          lea       $0080.w,a0
> > $0004eb92 : 0010 0080                          ori.b     #$80,(a0)
> > $0004eb96 : f010 4000                          pmove     (a0),tc
> Restore address translation.
>
> > $0004eb9a : 21df fa06                          move.l (sp)+,$fffffa06.w
> Restore MFP ints.
>
> > $0004eb9e : 4e75                               rts
> Regards
> Christian