2 Jan
1999
2 Jan
'99
2:38 a.m.
<If that's true, I'd really like a PDP-11! ;-)
I have a few they are fun.
<> It was however, slow!
<
<Compared to what? Clock per clock, it was the fastest thing for quite a
<long time. And remember, the 6809 was *not* the fastest of the fast
Compared by instruction cycle time.
<8-bitters... ever heard of a *6309*??? Pretty rare critter there! Made by
<Hitachi as an OEM to Motorola, CMOS, 3MHz (most overclockable to 4MHz) and
<would kick a '286 in the backside easier than anything 8-bit (with 6309
<hand-optimized code) -- it did have twice the registers of the 6809 and ha
<-- no typo here -- 32 bit math capabilities.
Well I am familiar and the 6309 was still not so fast. It was however
programtically efficient. Well written programs used less code and
therefore executed quickly. Also the extra registers and long math made
it particulary nice for compiler target for C language.
<>Comparing it to a 286/10mhz, sorry, no way.
<
<I did it. Admittedly, it wasn't *purely* apples to apples comparison, but
<took totally portable M$Basic programs doing integer & real math, and
<integer, real & string sorting, and ran them in RS-DOS on a CoCo2 and in
<Basic & BasicA on a True-Blue Bummer/AT '286-10. At .89 Mhz, the 6809 was
<roughly .7x in everything except integer-related items - remember, RS-DOS
<doesn't have integers... everything's a five-byte real.
Your not comparing CPUs, you compared BASICs. Interger math is faster
as iwould be five byte reals compared to the floating ppoint used on the
PC/AT. A more reasonable would have been a collection of sorts and math
tests that are in asm for the target machines. The 286 would have looked
far better and the 287 would have made the case stronger.
<vs. optimized assembly? IIRC, most (if not all) instructions execute in
<fewer clock cycles on the 6809. But also, the addressing capabilities of
Fewer slow cycles.
<the 6809 far outweigh anything 86ish - I've tried learning x86 assembly -
The '09 had better addressing modes and if you were not used to the way
things are done in the z80/x86 world it would have been a hard translation.
<ran screaming into the night... took my wife a week to find me! Any assy.
<job I've ever seen can be done in a *lot* fewer instructions on a 6809 tha
<an x86.
maybe.
<In my book, faster _and_ fewer gives a pretty decent improvement - sure, a
<couple loops would run faster on a '286, but a regular proggie is more tha
<just counting loops, and all of the branching & indexing capabilities of
<the 6809 really improve the odds.
X86 instuction set is strange but it is efficient. The problem is
programming philosophies and methods are very differnt between 68xx
and x86 due to type and use of registers and addressing modes.
Then again the PDP-11 blows both away as it's not a primary accumulator
machine also its regular in addressing modes and a two address machine.
even the slow LSI-11 beats the 09.
<version of Xenix available for an IBM / Clone? I know there was one for
<the Tandy 2000, but that's the only version I've ever seen. How far did
<M$'s licensing go with that OS?
Yes there was, not cheap either.
Rogue, is a good example of how code went from ASM to C, PASCAL, or whatever
and grew huge in the process.
Allison