RapidCAD is supposed to be an intel 486DX designed for a 386DX socket, though it is missing the L2 cache and a few 486 specific instructions. However, the math-coprocessor remains in the CPU core, making the floating point operations as fast as on a real 486. As far as I know it is the fastest floating point solution available for 386 systems. The RapidCAD-1 should always be installed with RapidCAD-2, which is a terminator for the FPU socket.
