• Please review our updated Terms and Rules here

Switch setting for an 8087 in a 5150 – info in Tech Ref is wrong

Lorne

Veteran Member
Joined
Jul 27, 2008
Messages
1,164
Location
Chandler, AZ, USA
I spent quite a bit of time trying to figure out why the Diagnostic disk wasn't listing the 8087 Math Coprocessor in one of my 5150s, and I think I’ve got it.

At first I thought maybe I had a unique 5150 where a switch was built or wired incorrectly (I even removed the mobo to check continuity on the switch itself) but I checked three 5150s, and it’s the same thing with all of them.
I double checked the switch settings listed in the IBM Technical Reference (page 6-3) which indicates that Switch 1, Position 2, should be:
“On if coprocessor is installed (must of OFF if coprocessor is not installed)”

Turns out the problem was the Technical Reference – it’s wrong !

I have a copy of the Option Instructions for the IBM Math Coprocessor Option, and on page 18 of those instructions, it states that if a coprocessor is installed, switch position 2 needs to be in the OFF position (ie: not the ON position that the Tech Ref dictates).

Hopefully, anyone else who comes across this anomaly will find this post, and not have to waste their time trying to figure it out.

PS: has anyone else noticed a similar thing on their 5150s, or am I the lucky first one to have gone through this?
 
The problem was the Technical Reference – it’s wrong !
PS: has anyone else noticed a similar thing on their 5150s, or am I the lucky first one to have gone through this?
Are you asking specifically about the maths coprocessor, or in general?
If in general, then I can tell you that you, from experience, that you can expect to find errors in IBM documentation just as frequently as in other company's documentation.
 
I don't have a copy of the Technical Reference so I don't know about that. However, I have a copy of Pocket PCRef by Glover and Young, from 1993. It has the switch setting as OFF for coprocessor present. I wonder where they got their information from.
 
I have the Operations Manual and it agrees with the 8087 Option Instructions; OFF = has 8087, ON = no 8087.
 
Note that the switch setting is for the benefit of the BIOS and any other program that wants to query it--IIRC, it doesn't actually disable the 8087 electrically.
 
So, what you are saying is that if a particular software is hard coded to use the 8087, the switch setting will not matter; the software will use the 8087 if it is present and possibly crash if it is not. Then relating to my other topic, it should POST regardless of the coprocessor switch setting. Is the 8087 tested during POST and is there such thing as a POST error for the 8087?
 
The switch only gates the interrupt from the 8087. It's perfectly possible to write 8087 code without using the 8087 NMI, in which case the switch setting doesn't matter.
 
I have the Operations Manual and it agrees with the 8087 Option Instructions; OFF = has 8087, ON = no 8087.

That's the Guide to Operations, right? Which version do you have? I have an electronic/scanned version of the GTO from April 1983, and it has the incorrect information, the same as the TechRef.
 
In the days when the IBM PC was extant, people would have used the addendum to the Guide to Operation that came with the Option for switch settings, rather than the Tech. Ref., which was really for hardware engineers designing options and programmers and the like.

I must check my copy sometime. :)
 
Back
Top