Released in the early 1990s, the Intel 487SX served as a floating-point coprocessor (FPU) upgrade for the 486SX microprocessor. Despite being marketed as a separate math coprocessor, the 487SX was essentially a full 486DX processor with an extra pin to fit the coprocessor socket.

When installed, the original 486SX remained on the motherboard but was completely disabled, a design likely intended to prevent its reuse and drive demand for upgrades. This configuration meant that many users unknowingly installed a far more powerful 486DX chip in their PCs. While technically possible to remove the 486SX and use the 487SX in its place, Intel did not guarantee reliable operation in that scenario.

The 487SX mirrored the 486DX’s specifications, featuring 1.2 million transistors, an integrated FPU, and an 8 KB cache. Intel also offered the 486-overdrive as an alternative upgrade, which installed in the 487SX socket, sped up the main processor, and disabled the original 486SX. Clock speeds for the 487SX were available at 25 MHz and 33 MHz.

The coprocessor used a 169-pin PGA socket, compatible with several upgrade processors, including the 486DX2, 486DX4, and later CPUs from AMD and Cyrix like the 5x86. While Intel licensed other manufacturers for general CPU production, the 487SX and related FPU designs were retained exclusively by Intel