OK Mr. Junkie,
I think we have a misunderstanding. Of course we both can count, I don't really doubt it 😵
In my original post I said there is no "standard drop in replacement" for socket PGA132 which is faster than the SXL2-50. What I really wanted to say is that there is no native CPU above the SXL2-50 in this CPU line. Of course there are socket PGA168 CPUs on a converter for the PGA132 socket, and the nice SXL2-66 you showed in the thread, but I wouldn't call then native PGA132 CPUs. I hope this claifies what I intended to say. I think my term "standard drop in replacement" was a bit misleading.

Cheers,
Fabian

It really just depends on what you want to include in your definition. This doesn't really seem like a point worth debating. If including interposers, there is also the IBM Blue Lightning BL3, which is faster than the SXL2-66. The fastest PGA-132 CPU without an interposer is either the Cyrix DRx2-66 or the TI SXL2-50.

When I read the OP title, I interpret this as why are there no PGA-168 to PGA-132 adapters to allow for use of any PGA-168 CPU on PGA-132 motherboard. For this case, there is the Transcomputer module, which seems to work pretty well.

I think that's really depending on the each person's point of view.
Some chips left the factory with interposers, some got them aftwerwards.
Same with mainboards/chipsets. Some were pure (386 or 486), some were hybrids (386+486)..

386SX wrote:

did a socket adapter ever exist to put a 486 maybe the lower models onto a 386 socket?

It seems now that it did...

I hope to reverse engineer and create a copy from this one... maybe even make a socket version... rather than QFP.

What the heck is this? Why would anyone want to use a 386 CPU on a 486 motherboard? Very fascinating find. Did you try this in a socket 3 motherboard?

That might make sense with respect to the slowest 486 cpus, like 486sx16 and 486sx25.

386_junkie wrote:

I hope to reverse engineer and create a copy from this one... maybe even make a socket version... rather than QFP.

Have you tested this to work? It is very interesting indeed. It seems to be a relatively simple device. But what is the deal with the two extra rows of contacts within the socket 3 pins? Was another chip supposed to be placed there?
Hope you can share the wire routing diagram at some point.

gerwin wrote:

386_junkie wrote:

I hope to reverse engineer and create a copy from this one... maybe even make a socket version... rather than QFP.

Have you tested this to work? It is very interesting indeed. It seems to be a relatively simple device. But what is the deal with the two extra rows of contacts within the socket 3 pins? Was another chip supposed to be placed there?
Hope you can share the wire routing diagram at some point.

Have created a separate thread which will get updated as I go, there you will find progress so far: -

486 to 386 CPU module

Sure, any work I do on it will be freely available to all.

feipoa wrote:

Why would anyone want to use a 386 CPU on a 486 motherboard?

Let me retort... what technology is available to a 486,.. that is not to a 386?

...

I'm thinking along the lines of... PCI bus, 3D cards, chipsets etc... but I guess that would only be the beginning, it opens a door.

As I stated in the other thread, this was likely an adapter for certain hybrid 386/486 boards. I really doubt it would work on a PCI motherboard.

386SX wrote on 2015-09-16, 10:44:

I've read somewhere that a 386 clocked 50Mhz was scheduled but not released. Is it true?

There were 50, 66,75 and even 100mhz overdrives that theoretically fit a 386 board albeit labeled 486dlc

AMD considered an actual 386 early on around the time they released the AM386sx but 40mhz 386 boards were already built with specially selected and binned components and cache so it was scrapped. As the AM386dx was releasing AMD was already in the middle of developing an AM486

Add to this the 386dx40 already was competing with the 486sx 25 and a 50mhz 386 variant brought nothing additional to the marketing dept