Quake is the game that killed Cyrix and hurt AMD. It's very FPU-dependent and Intel CPUs of this era excel there. If you're specifically interested in that, look to an Intel chip, preferably a P6.

That said, within the So7 platform, you've already identified the biggest non-CPU factor: motherboard chipset. The performance delta between slowest and fastest So7 chipset is more than 100%, so comparing CPUs across chipsets will completely skew results. Your i430TX is one of the fastest perforing chipsets, so no surprise it scores better than the same CPU on an i430VX (not bad, but nowhere near as good). The difference with Thandor's results are purely due to chipset. ALi Aladdin V also isn't bad and clocks higher than any other chipset (I have one stable on 133MHz FSB), but clock-for-clock it's slower than i430TX (but faster than i430VX).

The Ultimate 6x86 benchmark used a motherboard chipset (MVP3) that supports linear burst mode, which is a Cyrix-specific optimalisation. No surprise that the delta between K5 and 6x86 is bigger there.

With your i430TX, you don't have linear burst, so you won't see the same delta as in the Ultimate 6x86 benchmark. Compared to Thandor's benchmarks you should be a bit closer. Your K5 performs better, but a 6x86 on TX would also perform slightly better than on Aladdin V.

All in all, it would probably give you a measurable increase, but it's probably not enough to go from unplayable to playable.

Thank you that explains everything perfectly; I would probably get around 15% more fps in Quake, and it would probably also have a much better ALU performance otherwise as well..

It's probably not worth paying money to get a 6x86MX CPU then. My K5 drops down to around a 386DX40 w/ cache levels of performance when I disable L1 so it's proven to be fairly more versatile than I thought it would be, I'll stick with it 😀 Quake at 320x200 is playable if not great, but it's pretty much as I remember it was on my Pentium 133, or there abouts..