FDiv

FDIV is an instruction in the x87 floating-point unit (FPU) used to perform floating-point division. It operates on the FPU’s register stack, where computations are carried out using ST(0) as the primary accumulator and another source operand either from another stack register (ST(i)) or from a memory operand. The result of the division is stored back into ST(0).

There are two main forms of the operation. The standard form, FDIV ST(i), computes ST(0) = ST(0) / ST(i). The reciprocal form, FDIVR ST(i), computes ST(0) = ST(i) / ST(0). In addition, FDIV can take a memory operand, such as FDIV m32fp or FDIV m64fp, which divides ST(0) by the floating-point value located in memory, with the result stored in ST(0). The destination of the operation is always ST(0); other stack registers remain unchanged.

FDIV is part of the broader x87 instruction set introduced with the 8087 coprocessor and carried through later x86 processors. Its behavior, including exception handling for conditions like division by zero, overflow, underflow, invalid operation, and inexact results, depends on the current floating-point control word and status word settings.

In modern software, FDIV and related x87 instructions have been superseded for many tasks by vectorized and scalar floating-point instructions in SSE and AVX, but they remain available for compatibility with legacy code and certain low-level numerical routines that rely on the x87 stack architecture.