Re: [hatari-devel] WinUAE and 030 cache hits/misses?

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

Hi eero,

Le 26/01/2013 20:26, Eero Tamminen a écrit :

Which of the newcpu.c *run() "main loop" functions are used when
selecting cycle-exact and non-cycle-exact CPU mode with WinUAE?

This is encoded into : m68k_go() in the following code (that I really don't like this way)


        set_x_funcs ();
        if (mmu_enabled && !currprefs.cachesize) {
            run_func = m68k_run_mmu;
        } else {
            run_func = currprefs.cpu_cycle_exact && currprefs.cpu_model == 68000 ? m68k_run_1_ce :
                currprefs.cpu_compatible && currprefs.cpu_model == 68000 ? m68k_run_1 :
#ifdef JIT
                currprefs.cpu_model >= 68020 && currprefs.cachesize ? m68k_run_jit :
#endif
                currprefs.cpu_model >= 68030 && currprefs.mmu_model ? m68k_run_mmu040 :
                currprefs.cpu_model >= 68020 && currprefs.cpu_cycle_exact ? m68k_run_2ce :
                currprefs.cpu_compatible ? m68k_run_2p : m68k_run_2;
        }
        run_func ();



I did code the cycles values for cycle exact only (and as Thomas has added a new CPU for the MMU, my cycles are not taken into account there).

> I noticed that per-instruction cycle information is zero
just with the WinUAE "exact" CPU model.  If I use
"--machine falcon --cpu-exact false", I do get cycle
information.

 
Do you mean there's something missing ?
Just point me to the clue, I'm interrested.

Laurent