[Fwd: 8086 Segmentation (was 360 Architecture, Multics, ...)]

[John Ahlstrom]

Steve Morse wrote:

> The choice of 1MB had nothing at all to do with Intel being a memory
> company. That issue was never raised.
>
> It had everything to do with the fact that we needed something only
> slightly larger than the current limit since one large customer
> (Tektronix I think but I'm not sure) had a program that was pushing the
> 64K limit. So 128K was thought to be enough. I went to 1M but didn't
> want to go any further because of the dead space it would create -- the
> bigger the paragraph, the more unwasted space at the end of each last
> paragraph. With hindsight you can argue that that was false economy.
>
> John Ahlstrom wrote:
> >
> > Landon wrote:
> >
> > > The face of the micro industry would be dramatically
> > > different if Intel had picked a 256 byte paragraph size,
> > > rather than 16 bytes. This would have given the
> > > *architecture* 16M of address space, instead of 1M. Intel
> > > could have done something other than the (shudder) 286.
> > >
> > >
> >
> > Is it true - not just speculation or a good (or bad) guess -
> > that the 1 MByte total address space was due at least in
> > part to Intel being primarily (largely?) a memory company?
> > Who in a memory company would want RAM prices to come
> > down so far that a micro could afford more than a megabyte?
> >
> > Where is Steve Morse now that we need him??
> >
> > JKA
> > --
> > It's a good thing the iAPX432 never caught on -
> > otherwise a truly horrible Intel architecture
> > might have taken over the world.