I think I discovered something new about the Cyrix/IBM 6x86

[stamasd]

I may have accidentally stumbled upon something previously unknown about the venerable and otherwise not very interesting 6x86 CPUs.

According to the datasheet at http://datasheets.chipdb.org/IBM/x86/6x86/6x86_ALL.pdf the only multiplier these CPUs support are 2x and 3x. They have only one multiplier control pin, CLKMUL, and depending on the state of that pin low/high the multiplier is either 2x or 3x (with not connected = same as low because it has an internal pull-down resistor).

So 2x and 3x is all there is huh?

I was playing around with one of them in a socket7 motherboard, and at one point tried to find what happens if the multiplier jumpers were configured in the positions that were unmarked for Cyrix CPUs. One of the two undocumented settings did not POST, but in the other one the CPU posted with a multiplier of 1x.

I don't know how this motherboard is able to do that given the way multiplier control is implemented, but it's reproducible and reliable. I have verified 2 6x86 CPUs to behave this way: a 6x86-P166+GP and a 6x86L-P200+GP (meaning, both of the ones I had on hand for testing).

So I was able to get the 6x86 running at frequencies as low as 50MHz.
I've documented this, including pictures, on Vogons: https://www.vogons.org/viewtopic.php?f=46&t=85855&start=20 (second half of the 2nd page in that thread)

[frag_]

It's known fact.
https://www.vogons.org/viewtopic.php?t=41832
https://www.vogons.org/download/file.php?id=16529

[stamasd]

Hmm interesting, I wasn't aware of that. Not common knowledge.

I've updated my Vogons thread to clarify. BF1 is still undocumented and "reserved" on the 6x86.