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

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

I don't understand something :

According to hatari debug mode, I can read :

cpu video_cyc=  1034  10@  1 : 0004EB82 4a78 0468                TST.W 
$00000468
cpu video_cyc=  1038  14@  1 : 0004EB86 6600 0006                BNE.W 
#$0006 == $0004eb8e (F)
cpu video_cyc=  1046  22@  1 : 0004EB8A 5240                     ADD.W 
#$00000001,D0
cpu video_cyc=  1048  24@  1 : 0004EB8C 60f4                     BT .B 
#$fffffff4 == $0004eb82 (T)

If(s the CPU speed loop of EXA demos.
The code runs into the instr cache.

I've taken the 4th iteration (they all take the same time except the 
first one, which seems OK to me).

If I can rely on the debugger, I read 14 cycles for this iteration.


If I compute the loop by hand, I get :

tst.w     $0468.w        {2,    2,     8,1,0,0,    12,1,2,0}, // TST.W 
(xxx).W
bne       $1cd2a         {6,    0,     6,0,0,0,  8,0,1,0},    // Bcc.W 
(not taken)
addq.w    #1,d0          {2,    0,     2,0,0,0,  4,0,1,0},    // ADDQ.W 
#<data>,Dn
bra.s     $1cd1e         {6,    0,     6,0,0,0, 12,0,2,0},    // Bcc (taken)

Which is 8+6+2+6-2 cycles = 20 cycles.

(-2 cycles because of the Head/tail -2 between the first and second 
instructions)


So, it seems that the loop run faster under hatari than it should, but 
D0 value is lower than $4100 (it should be higher, no ?)

Laurent







Le 05/02/2017 à 20:14, Laurent Sallafranque a écrit :
> Oups, right ;)
>
>
>
> Le 05/02/2017 à 19:56, Nicolas Pomarède a écrit :
> > Le 05/02/2017 à 19:43, Laurent Sallafranque a écrit :
> > > I've got the same value for D0, ie $3e65.
> > >
> > > What I find strange is that D0 increase in a loop that waits the 
> > > time of
> > > a VBL, and it only increase 15 973 times during a VBL ($3e65) ?
> > > I would have expected a much higher value.
> > > At least, the add is a word add, so it may overflow many times during
> > > the test, which also means that the accuracy will have to be very
> > > precise if we must reach this $4100 to $4200 area after many "overflow
> > > loops"
> >
> > it's normal to have sthg around $4000 : VBL has ~320000s cycle at 16 
> > MHz and the loop takes between 18-20 cycles (it's not just a addq, 
> > you have 4 instructions in the loop)
> >
> > 320000/20 = 160000 = $3e80, so $4100-$4200 is correct, there's no 
> > overflow