IBM 5160 math copro switch

[Bill Degnan]

A good game theory word problem.. The rationale causing manufacturers to
set the switch to "is installed" given they know that if there is no
co-processor installed the system will still work.  It would only be an
issue if one had the co-precessor installed and the switch was set to "none"

So my conclusion would be instructing the customer to insert the chip and
verify the switch was "on", rather than have them also flip switches,
possibly making an error is best.  I could see the conclusion would be to
ship them in the "on" position, despite possible confusion by the very
small percentage of owners who open the cover and inspect their switches vs
the risk of damaging the board or causing misconfigurstion of something
else as a result of changing switches incorrectly.  I was a support line
manager in the early 90s, I remember how simple decisions like that would
change call volume, effect labor requirements, etc.  Brings back memories.
Bill

On Mon, Sep 7, 2020, 9:50 AM Jules Richardson via cctalk <
cctalk at classiccmp.org> wrote:

>
> Sifting through my XTs and clones... I've had three so far with no math
> copro (8087) fitted, but switch 2 of SW1 is set to off.
>
> IBM docs say off is "math copro installed" and on is the "installed"
> position, i.e. the reverse of what I'm seeing (the minuszerodegrees site
> repeats this, although I expect that just copied what's in the manual,
> too).
>
> These three machines (two are 5160's, the other a clone) came from
> completely different sources - it seems unlikely that all three had 8087's
> fitted which were pulled at some point. Googling a few more system board
> images, ones with 8087's have that switch on - but for boards without, the
> setting seems to be pretty much 50/50.
>
> Did the meaning of the switch perhaps change at some point (both of my
> 5160's are late model 256-640 boards), and rather than being a simple
> installed / not installed it was more along the lines of "software should
> use it if present / software should never use it"? After all, it's not
> like
> the BIOS will do anything with an 8087; it's only there for software
> specifically coded to make use of it.
>
> I suppose it's also possible that users were in the habit of first setting
> all the switches to off when configuring a machine, then setting the
> relevant ones to on according to their memory/floppy/video config - and
> the
> copro setting just got overlooked.
>
> Weird, anyway.
>
> Jules
>
>