14 Feb
2013
14 Feb
'13
3:27 p.m.
On 02/14/2013 01:46 PM, John Wilson wrote:
In the early days, some companies were afraid of
getting sued by IBM
(while most of them were totally fearless), so they'd go out of their way
to add stupid incompatibilities. I put together a machine (on a rackmount
panel) based on an Eagle 1600 motherboard, back when they were dumping those
cheap, and it was close to useless. I had to patch the BIOS to work with
a normal FDC board (IIRC they'd pointlessly changed the DOR layout), and
while it was nice that using an 8086 instead of an 8088 bought some speed,
the fact that they didn't account for that in the (8-bit-only) ISA bus
interface broke most software that touched the screen etc. ... since the
high halves of all word accesses were lost, so you had to special-case
your code accordingly.
Wait a second--the 1600 preceded the Eagle PC by about a year. Lots of
companies came out with 16-bit 8086-based boxes before it was apparent
that compatibility with the IBM iron was a necessity--remember that for
the first year or two, there wasn't that much in the way of IBM-specific
hardware. Lotus pretty much changed all of that.
According to my notes, the Eagle 1600 was probably a better system than
the 5150. 8086 CPU, optional hard drive, 96 tpi (800K) drives. It was
marketed with CP/M 86 initially. Eagle might have survived if the
founder hadn't gone out celebrating the IPO.
As another example, there was the Columbia 960 (IIRC) that had an 8086,
but was followed by the MPC that was actually an IBM compatible but
inferior to the 960.
How much compatibility was another question. Initially, it was assumed
that being able to run MS-DOS and early software, such as Multiplan was
sufficient. I participated (against my better judgement) in an effort
that inolved a 186 feeding a VT220-type terminals was sufficient--it ran
MS-DOS.
There's always the Tandy 2000 and the Stearns PC. Compatible, but not
sufficiently so to acquire any market share.
Remmber that the 8086 was sampling in 1978, so there was a lot of time
before the 5150 came on-scene. Heck, the IBM Displaywriter precedes the
5150 and could run CP/M-86.
Bill Godbout had cards out using both the 8086 and 8088 CPUs. I recall
that initially, the licensing for small quantities of MS-DOS was so
hostile, that Godbout's people worked out a way to patch a new IO.SYS
into a retail copy of PC-DOS 1.x to run on the S100 box.
Of course the NEC PC98 (APC, 9801) was completely incompatible with the
5150, but hung on for many years, due to its almost total domination of
the Japanese market.
--Chuck