Re: "philosophical" view of emulation vs real thing

[Ricardo Bittencourt Vidigal Leitao]

On Tue, 13 Jul 1999, Laurens Holst wrote:

> > Why not? T-states emulation is kind of easy.
> Oh, is it?

        Sure. All you have to do is compute how many t-cycles you have in
each time frame (16ms/20ms). If I recall correctly, the MSX-1 at 60Hz
should have more or less 59000 cycles. After executing each opcode, you
just subtract the t-value from the total, and issue an V-IRQ when the
counter reaches zero. You DO need an accurate t-cycles table, BrMSX uses
the table compiled by Cyberknight (adjusted to take into account the
extra clock in M1 cycle). Turbo-R emulation will be much harder than
that, since I'll have to check for ROM/DRAM modes...
 
> > Ask Adriano Cunha for the MSX implementation of the BogoMIPS
> > algorithm and check it with a real MSX, and then with BrMSX...
> But the BogoMIPS benchmark (it IS a benchmark, isn't it?) takes the average
> amount of MIPS, does it?
> 

        BogoMIPS is just an bogus measure, but the value returned by it
are EXACTLY the same in BrMSX and in a real Turbo-R running in Z80 mode.
Another test you can make is this one:

        TIME=0:FORI=0TO1000:NEXT:?TIME

        The value returned by an Expert 1.0 (model emulated by BrMSX) is
92. This is the same value returned by BrMSX. This value is not the same
in all MSX-models, due to differences in the bios irq handler, so if you
want to test BrMSX againts another msx-model, be sure to change the
MSX.ROM.

        I'm still looking for bugs in BrMSX. If you find ANY bug, even a
VERY SMALL bug, then please report it to me.

----------------------------------------------------------------
Ricardo Bittencourt               http://www.lsi.usp.br/~ricardo
[EMAIL PROTECTED]  "Ricardo is subtle, but malicious he is not"
------------ Say NO to Coca-Cola. Drink Tubaina. ---------------


****
MSX Mailinglist. To unsubscribe, send an email to [EMAIL PROTECTED] and put
in the body (not subject) "unsubscribe msx [EMAIL PROTECTED]" (without the
quotes :-) Problems? contact [EMAIL PROTECTED] (www.stack.nl/~wiebe/mailinglist/)
****