The 486SX processors were the same as their DX counterparts in design, but the built in math co processor was disabled. There was a 487 which was actually a 486 with the math co processor enabled. If the motherboard had this slot for the 487, the SX cpu would be disabled and the 487 cpu would take over all processing.

Was the i486SX "Cripple-chip" really a planned release, or was it actually as rumoured to be, a genuine i486DX -33, that failed the FPU production-line tests???? I was donated an i486SX-20, and I was also donated a "full-size-AT form-factor" 486-33 m/board. This crazy-large m/board had an 80486 socket, and 4167 socket too. It also had sixteen (count them, 16) SIMM sockets, cause, what was a 16MByte SIMM??? It only understood 256kByte and 1MByte SIMMs, and it only understood the 9-chip SIMMs, not the later 3-chip SIMMs. This proved to be a bane to me, as the 9-chip SIMMs cost more than the newer/faster 3-chip units.
I put two-and-two together, and that little PGA SX-20 chip ran at 33MHz just fine. As a test, I tried to smoke-it-out by covering it with cardboard etc. to fry the thing. It would not stop running normally! (Without the cardboard), I ran it at 33MHz for years, before I sold that sytem. Last I heard, it still was fine.

This is an SMD package CPU mounted on a PGA board to fit into a standard desktop socket.

Great little sub-unit this! Overclocked it to 40MHz no probs. Used it for years and years. No heatsink needed at 40 either! Not bad for a so-called -25.

I lusted for one of these! Time marches on, and I upgraded to Intel DX4 -100 after all. Never got to see one of these SX2's.