I'd start by taking the heat sink off and see what's under it.

Why would there be a giant heatsink for that single mosfet? There must be something under it.

Only a few resistors were under the heatsink, as in the middle of the existing socket.

I was partly wrong about the single/dual jumper as it is soldered to the single cpu selection, but that is easily fixed.

There is another jumper missing though.

According to the manual there should be a JP31 for enabling/disabling address pipeline feature. This should default to off for dual cpu and on for single cpu. Since it doesn't exist on my board, it is possibly enabled by default. If this is the case, do you think this would be a problem with dual cpu's?

If someone has the board it could be usesful with a closeup photo of the JP31 area (above the left socket) to see if it is possible to modify for a jumper socket.

Also, the revision of my board is 1.2. The only image I found online for this revision was also single socket with the heatsink. Images of boards with dual sockets installed seem to be revision 1.5, and they also have some electrolytic caps around the sockets that is missing on rev 1.2, and they lack the volatage regulator.

Well, when looking at the image on the manual, it is a rev 1.2 board with two sockets and the voltage regulator removed. 😀
Interestingly it has the JP19 still soldered to the single cpu position so it could be some kind of arranged picture. (There is also another manual with the heatsink on the image)
But otherwise it indicates that it should be possible to just solder a second socket in place.

And... problem solved. Unfortunately the ISA bus controller needs to support APIC, which it doesn't on this board revision.
It has 82378ZB for single cpu and needs 82379AB for dual cpu.

http://www.cebix.net/downloads/bebox/82378zb.pdf

The image with the 82378ZB and dual sockets must be lying! 😐

Then when I don't have a dual socket board, another question on opinion.

For my one spare AT case, which board do you think I should use?
This one or an Intel Premiere II, Plato, which I also got in the same lot?

Hello,

I don't know whether it helps, but I made a picture of the clock on the Asus PCI/I-P54NP4 I have:

mightylaocrahcot wrote:

Unfortunately the ISA bus controller needs to support APIC, which it doesn't on this board revision.
It has 82378ZB for single cpu and needs 82379AB for dual cpu.
The image with the 82378ZB and dual sockets must be lying!

I also see 378ZB in that image. Perhaps there were revision of the 378ZB had APIC multi-processor support, or perhaps it is not entirely necessary to have APIC multi-processor support? It would be interesting to solder on the extra socket 5 with accompanying SMD caps, and whatever else is missing, to see what limitation there is, if any, in this dual processor implementation. Although I have never owned one, these early socket 5 dual processor systems seem quite interesting.

feipoa wrote:

I also see 378ZB in that image. Perhaps there were revision of the 378ZB had APIC multi-processor support, or perhaps it is not entirely necessary to have APIC multi-processor support?

I have done my homework and read up a little on it. It seems that APIC is a necessary part of the specification for multiprocessor support. From P54C on, the functionality was split into a local APIC built into the processors and a discrete I/O level part. It needs the I/O level to function as a multi processor system. It would probably run without it, but with only one cpu taking part in the running.

feipoa wrote:

It would be interesting to solder on the extra socket 5 with accompanying SMD caps, and whatever else is missing, to see what limitation there is, if any, in this dual processor implementation.

Yes, it is tempting to trust that image and go for it.. 🤣 It is always sweet to tinker to success.
My suspicion though is that I would be biting the dust and better off hunting for an already dual socket board.

I learned also that windows 9x doesn't support APIC and hence multiple cpus, since it would break support for MSDOS drivers. (Or am I wrong?)
But I guess OS/2 supports it, and it would have been a nice reason to have a first acquaintance with that one.

flupke11 wrote:

Hello,

I don't know whether it helps, but I made a picture of the clock on the Asus PCI/I-P54NP4 I have:

Thanks! I appreciate the image. The part I needed is on the opposite side of the socket, but since I'm not pursuing this for now, I wont ask you for another. If I would get back to this in the future, maybe I can contact you for another image.

Are you restricting your search to dual socket 5 boards, or would you also consider dual socket 7 boards? I like dual socket 5 boards because they still maintain that pre-1995 ancient feel to them.

...How would you even do this? Surely the task of manually soldering 320 pins easily outweighs the difficulty of just procuring a new board?

Jorpho wrote:

...How would you even do this? Surely the task of manually soldering 320 pins easily outweighs the difficulty of just procuring a new board?

I see your point. I guess I would be more skilled with the soldering iron afterwards, which would be a gain.

feipoa wrote:

Are you restricting your search to dual socket 5 boards, or would you also consider dual socket 7 boards? I like dual socket 5 boards because they still maintain that pre-1995 ancient feel to them.

I'm not exactly searching for one. It was more of seeing an opportunity since I happened to have this board. But I understand what you mean. This Asus and the Intel Plato is my only socket 5 boards so far. I ended up building a machine around the Plato the other day, because it felt even more "ancient" than the Asus. Although I suspect Asus is a somewhat better board. I will probably try a dual board if I stumble over one, even if I have already too many old computers that I don't "need". But maybe I have passed the "point of no return" in this regard? 😵