this:
my socket7 cpu benchmark results

‎‎‎

A while back I did some benchmarking on a VIA MVP3 based motherboard, and a couple of the early setups matched what you're asking here. One CPU was a Pentium MMX at 233/66, the other was a K6-2 450MHz clocked down to 233/66. The K6-2 I used was a CXT core, while I think a "real" 233MHz chip would be a bit different. I don't know how much difference that makes.
Everything I tested with these CPUs was with a Riva 128 ZX AGP card.

3DMark99 640x480x16 triple buffered, "Intel" optimized mode, not 3DNow: MMX 233MHz = 410, K6-2 233MHz = 813
3DMark2000 640x480x16 triple buffered, "Software" T&L mode (not 3DNow): MMX 233MHz = 443, K6-2 233MHz = 682
The difference diminishes greatly at higher resolutions, probably because of the video card.
The only real games I tested with these CPUs were Unreal, DOOM, and Quake. None of them showed such a big difference, actually they all favored the MMX slightly.

Unreal, software rendered 320x240 flyby demo: 15.4fps on K6-2, 16.1fps on Pentium MMX - (MMX is 4.5% faster)

DOS:
3DBench 1.0c: Pentium MMX 142.4, K6-2 219.8 - (K6-2 is 54% faster)
Chris 3D Benchmark: MMX 229.8, K6-2 265.5 (K6-2 faster by 15.5%)
Chris 3D 640x480: MMX 70.7, K6-2 73.4 (K6-2 faster by 3.8%)
PC Player Benchmark 320x200: MMX 56.6, K6-2 75.1 (K6-2 faster by 32.7%)
PC Player Benchmark 640x480: MMX 23.2, K6-2 25.4 (K6-2 faster by 9.5%)
DOOM max details, time in "realtics": MMX 775 , K6-2 823 - (MMX faster by 5.8%)
Quake 320x200: MMX 48.2fps, K6-2 47.2fps - (MMX faster by 2.1%)
Quake 360x480: MMX 20.9fps, K6-2 20.8fps - (MMX faster by 0.5%)
Quake 640x480: MMX 16.4fps, K6-2 14.4fps - (MMX faster by 13.9%)

‎‎‎

ElectroMan wrote:

Thanks. […]
Show full quote

Thanks.

Although, correct me if I'm pulling a genuine idiot - I am fairly new in the retro computing after all, mind you, but it appears to me that:

1. It only covers via c3 (233)
2. Shouldn't via c3 frequency range from 500 MHz to 1.4 GHz? (I've seen the comment under the chart about it performing worse than S7 CPUs)

But overall I believe I've got the picture from other CPU comparison.

didn't you see that i compared them at 300?

Feel free to my very recent and very similar thread here: Pentium 233 MMX to K6-2/300, worth it?

‎‎‎

Ooops, apologies..

The best argument for a P-MMX 233 vs a K6-233 is K6-233s are power hungry and hot. Some are 3.2V core though there was also a 2.9V version IIRC. Some boards may not cope with them without extra cooling on the voltage regulators, and some, while they could boot one in their prime, may now not, due to aged capacitors. K6-266s less hot, less problem, they did a respin. (In fact they'll probably o/c close to 400.)

However, the 3.2V K6 does have the advantage that you can say "screw needing split plane" and run it on some early 430VX or TX boards that didn't have voltage regs, or had an addable plugin reg that nobody has ever seen in the wild. Yes, this is even slightly hotter still and I recommend bolting huge sinks to mosfets and making sure they sit in airflow. Some FX boards were meant to be theoretically capable of running MMX class, but I never met one of those that i) actually had the VR needed to run MMX class, ii) claimed to run higher than a P166, iii) had a flashable and available BIOS that had any better than P166 or so MMX support in it, never mind K6, so have never personally confirmed that that was possible.