🚀 go-pugleaf

syd...@skypoint.com (Scott Butler) writes:
>IBM had a processor *in mind* when it approached Microsoft about doing
>a BASIC for the new machine - they HAD NOT made a decision on the
>processor yet.

Not the impression you would get from Intel's historical web site, but I'll
admit that's not surprising ;)

>Jack Sams of IBM discovered that "Gates and his crew
>obviously new quite a bit about hardware design as well", so Sams
>"gave him sort of an open invitation to tell us what was needed."
>(from "Gates", by Stephen Manes & Paul Andrews, page 152).

I'm sure Gates didn't exaggerate this in any way in his own biography.  I
find it unusual that monolithic IBM would solicit advice from what was
basically a pipsqueak software company.

>Gates
>pushed for many things, including "a 16-bit design instead of the
>outmoded 8-bit". 

The 8086 is 16-bit in some ways; settling on the 8088 just made the design
cheaper and quicker.

>Sams also contacted Gates regarding an OS, and Gates initially
>referred him to Gary Kildall of DRI, who was off flying his plane and
>missed the meeting - probably one of the most expensive airplane rides
>in history.

This is the anecdote to which I was referring.

>Maybe you should double check your *own* statements before you shoot
>your mouth off....

I still don't understand, though, how:

1) Intel's flawed architecture does a fine job of running UNIXes such as
Linux, FreeBSD, and the like

2) Windows 95 still has a lot of DOS internals

There are DOS-mode emulators for quite a few PC UNIXes, and they work
reasonably well, while allowing the UNIX kernel to still do things its own
efficient way.  Why couldn't this approach be taken for a new OS such as
Win95?  You still get better performance under OS/2 or UNIX, on the same
equipment, in general.

-- 
Kevin Martin                                                   si...@mcs.com
Pinball Archive: ftp://pinball.cc.cmu.edu/pinball http://pinball.cc.cmu.edu/
UFO's are for real, but the Air Force doesn't exist.