Only one way of finding out...

It's pretty safe though, so worst case it simply won't post.

I'd say there's a good chance the 25 might work at 33MHz, it may even have the same silicon as a 33. If it works, it works, if not, oh well. I wouldn't try the 16 on it though.

Please correct me if I'm wrong, because maths isn't my strong point, but doesn't an overclock from 25Mhz to 33Mhz represent a 24% increase?

Just for your information: I tried using an Intel 386DX-33 CPU, with a 386 mainboard that had an 80MHz oscillator in it, and it worked fine. I think that this represents an 18% overclock. I then replaced this 80Mhz osci with a 100Mhz osci, and I saw no BIOS POST on the screen. I then replaced the 100 osci with an 80 osci, and it worked fine again.

Well, 25*1.24=31, while 33/25=1.32. It's 24 if you compare the delta with the "destination" value, but one usually refers to the original value when expressing increases or decreases 😀.

retro games 100 wrote:

Please correct me if I'm wrong, because maths isn't my strong point, but doesn't an overclock from 25Mhz to 33Mhz represent a 24% increase?

Just for your information: I tried using an Intel 386DX-33 CPU, with a 386 mainboard that had an 80MHz oscillator in it, and it worked fine. I think that this represents an 18% overclock. I then replaced this 80Mhz osci with a 100Mhz osci, and I saw no BIOS POST on the screen. I then replaced the 100 osci with an 80 osci, and it worked fine again.

Nope, 32%. To determine the overclock you take the amount of the overclock (in this case 8mhz) and divide it by the original clock speed (in this case 25mhz). 8/25 =.32 or 32%

Your example of overclocking a 33mhz chip to 40mhz is a little over a 21% bump. 7/33=.21212121.....or just over 21%.

I've found that overclocks of 20% of the chips rated speed is about all a 386, 486, or slow Pentium can take. Pushing beyond that raises the risk of burning up the chip past the point of my risk tolerance. I would personally not risk pushing a 25mhz 386 to 33mhz because 32% is too far for me. If you can find an oscillator that lies in between 50 and 66mhz, then that might be doable assuming the rest of the system can handle the new. non-standard clock rate. A 60mhz oscillator would give you a 30mhz processor speed and that would be 20%.

I just accidentally discovered that starting the board (GA-386UM REV.2) with its Turbo switch off makes it report "CPU frequency 16 MHz", so I was able to successfully test both CPUs at 16 MHz (I didn't make any tries at full speed). Let the 386 collecting begin 😀...