I think it varies by motherboard manufacturer, depending on how they implemented it. But your best bet is to run some benchmarks, then compare to known results from slower systems. This will put you in the ballpark. See the cache disabled benchmark chart in the link in my signature. There's a bunch of historic results of systems from slow 386 up to Pentium and higher.

You can see if CHKCPU detects your non turbo speed.

without a CPUID instruction, you have to do things like execute an instruction that you know the timing of in a loop.

As far as I know, Deturbo usually interjects wait states to the l1 and l2 cache, not sure what it does to the FSB speed. I usually just set the deturbo LED display to the FSB freq. for simplicity.

And for what it's worth, my DX2-66 runs about like a 20Mhz 486 according to most benchmarks when de-turboed.

Thanks for all your replies. I will try to benchmark my way to find an approximate CPU speed. I had to overcome some L1 caching issues in order to get the computer to even boot.
It appears that my motherboard was incorrectly configured for my DX4-100 EW CPU (the one with writeback L1 cache). I had to set jumper configuration for P24D instead of 80486DX4.

Any suggestions as for benchmarks besides Navrátil Software System Information?

@dfir -- check the cache disabled thread in my signature. There is a spreadsheet of reference results using 3dbench, pcpbench, speedsys, and Doom. My DX2/66 is in there with Turbo on and off.

@clueless1 Dude, that is one hell of a spreadsheet. Nice job! I will make sure to crunch it soon and perhaps find similarities from my own benchmarkings. Sweet...