I remember having a 486DX4/100 (a clock-tripled 33 MHz bus and 99 MHz core). I downloaded the reference manuals from Intel and found the pins I could jumper to

I remember having a 486DX4/100 (a clock-tripled 33 MHz bus and 99 MHz core). I downloaded the reference manuals from Intel and found the pins I could jumper to change the clock multipler from x3 to x2, and then I increased the bus speed to 50 MHz to effectively have a 486 DX2/100. I don't remember all the hardware details anymore, but I think I tested system stability by recompiling XFree86 on Linux. This was something I already knew how to do as a beta tester for 2D accelerator drivers, and it took many hours and so seemed like a good test...