The x87 instruction set was Intel’s first floating-point architecture, introduced in 1981 with the 8087 hardware coprocessor. The architecture provides an array of eight 80-bit data registers, which is managed as a circular stack. Each numerical operand is located either in an x87 data register or in memory and is interpreted according to one of the data formats defined by Definition 5.3. Data register contents are always interpreted according to the double extended precision format (EP). Memory operands may be encoded in the single (SP), double (DP), or double extended precision format. Numerical results are written only to the data registers in the EP format. The architecture also provides distinct 16-bit control and status registers, the FCW and the FSW, corresponding to the single SSE register MXCSR.