[geeks] Photo "sharing" site recemondations please.

[Geoffrey S. Mendelson]

On Tue, Apr 29, 2008 at 06:52:18AM -0500, Lionel Peterson wrote:

> I am 99 44/100th percent positive the 8088 was compatible with the 8080 
> instruction set - it could also run 8008 code (if you had any) if I remember 
> correctly. This was Intel showing they learned the lesson of the IBM 360 
> series and the value of a consistent instruction set.

Not that I know of. In fact there was a CPM for the 8088/8086 which was
NOT binary compatible with the CPM for the 8080. The instruction set itself
was a superset of the 8080 instruction set and the addressing modes
of the 8080 were supported.

Intel sold (gave away?) an 8080 to 8086 assembly languange conversion 
program. You ran your SOURCE code through it and got an 8086/8088 assembly
language program out of it. It then had to be assembled by the 8086
assembler.

I have (and it still runs under Windows XP) WordStar 3.3 for MS/DOS
that was created that way. It still has all the bugs of the 8080 version,
the bugs added by the converter programs, the bugs added by the assembler
and of course the bugs because CPM system calls were not perfectly 
replaced by the equivalent DOS ones.

Somewhere you can probably find the complete list of bugs in it, I've
never bothered, but I've seen them over the years. 

As far as I can remember, the only real difference between the 360
instruction set and the 370 implentation of those instructions from a
user point of view was the difference in virtual memory. The 360 used 32
bit addresses, the 370 used 31.

Of course the 370 added a whole lot of new instructions, my favorite
being MVCL. I probably used it to clear memory more than I used it
to actually move data. 

You can still see it in my C code, I tend to do lots of
strncpy(array,"",length(array)); which is probably one of the slowest
ways to do it on an Intel processor, but the fastest way to code it. :-)

Geoff.


-- 
Geoffrey S. Mendelson, Jerusalem, Israel gsm at mendelson.com  N3OWJ/4X1GM