2 May
2005
2 May
'05
1:33 p.m.
Tony Duell wrote:
Yes, but not sure what point you're making. Memory accesses to ROM were faster
than RAM, so the ROM could get away with this code to put a character+attribute
onscreen without "snow" during horizontal retrace:
MOV AX,BX
STOSW
...but doing so from RAM wouldn't work because the MOV AX,BX is a 2-byte opcode
that took 8 cycles to fetch, so it wouldn't be fast enough and you'd see snow.
This drove me quite insane until I figured it out; I modified my own code to
use a 1-byte opcode:
XCHG AX,BX
STOSW
...since I didn't need to preserve either AX or BX after the STOSW. This did
the trick.
I don't
have that either, but I've disassembled my own BIOS and found that it
is nearly useless to me -- for example, there are instructions that take longer
to execute out of RAM (4 cycles per byte opcode) than ROM (2 cycles) -- so it's
Rememebr the true-blue IBMs didn't shaddow the ROM. The BIOS ran from ROM
all the time. When I got
my TechRefs, I went through all the schematics, and exclaimed at
approximately 2 minute intervals 'They did WAHT???'
I can just imagine you doing this, LOL :-) The BIOS CGA code for putting a
pixel element on the screen is particularly inefficient, for example, which is
why it is so damn slow.
--
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Want to help an ambitious games project? http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/