Old Man (like me) Bill Fairchild noted:
> you were supposedly able to configure a model 30 with only 8K.
True, and IBM took a boatload of first-day orders for 8KB 360/30
boxes. Before any of them shipped, it was clear that nothing at
all useful could be done with them. I don't know if any machines
actually shipped with 8KB, but I do know that by the time I took
a look at the green book (the IBM Sales Manual) -- while we were
still using a 64KB 360/30 -- it was not possible to order such a
core configuration: D [16KB] was the smallest one then available.
I have read that expectant 8KB IBM 360/30 customers for the most
part did not bat an eye at the requirement to order a bigger box.
I can believe that, but what I still don't know is what the sales
droids had to do/offer to make that be the case.
Regardless, with OS/360 obviously coming in at twice its design
point (effectively needing 64KB minimum) the world soon divided
itself into small (DOS/360) and large (OS/360) customers -- and
boxes.
> Dave Freeman, who had managed the development of the DOS/360
> supervisor for IBM, told me that its design point was 6K so
> that there would be 10K for application programs on a 16K
> machine.
That 16KB design point got set after it was clear (and precisely
because it became clear) that that nothing could effectively get
done in any 8KB Model 360/30 that lots of customers had ordered.
DOS/360 __HAD__ to run in a 16KB machine, period. That was what
was going to "save" the System/360. If IBM now had to go back to
the customers and tell them they needed 32KB machines, the sales
droids felt that customers would cancel orders in droves. But
things would work out differently than anybody, including those
in IBM, expected. The same factors that bloated the supervisors
(including DOS/360) also bloated customer applications. Thus,
larger core memories became the norm, not the exception. Another
consequence of this was that the original manufacturing estimate
of the amount of "core" that needed to be built was now realized
to be way low, so IBM had to quickly retool some existing sites
and build (two, I believe) new memory plants to be able to have
enough for surprising large number of existing & expected orders.
Finally, another consequence of this was that it focused IBM's
attention on the royalties they had to pay for every byte of
core memory BUILT to a specific patent holder. This inflamed
the effort to find a low-cost memory technology -- other than
ferrite cores and the very expensive monolithic memory that
IBM already (then) had plans to use. An IBM Fellow told me in
1971 that that patent holder had seen the handwriting on the
wall and offered to reduce the royalty. But by then it was too
late: it would be clear later that the 370/145 design had been
set in stone by then.
--
WB
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
