MUL

MUL takes a single operand, the multiplier. The other factor and the place for the result are fixed by the operand size:

The result is twice as wide as the factors, so it always fits: 255 × 255 = 65025 fits in AX. The numbers are treated as unsigned; for signed values use IMUL.

MOV     AL, 6
MOV     BL, 7
MUL     BL         ; AX = 42
AAM                ; AH = tens, AL = ones
OR      AX, 3030h  ; both digits -> characters
MOV     BX, AX
MOV     DL, BH
MOV     AH, 02h    ; DOS: print DL
INT     21h
MOV     DL, BL
MOV     AH, 02h
INT     21h

What is not allowed

There is no immediate form on the 8086: MUL 10 does not exist, so load the constant into a register or keep it in memory. A memory operand needs a size when nothing implies it, MUL BYTE PTR [SI] or MUL WORD PTR [SI]. The size chosen decides which registers are used, so MUL BX overwrites DX even if you only expected a small result.

MOV     AX, 1000
MOV     BX, 300
MUL     BX  ; DX:AX = 300000 = 0004h:93E0h, CF = OF = 1

Flags

CF and OF are both set to 1 when the upper half of the result (AH for a byte multiply, DX for a word multiply) is not zero, and both are cleared when it is zero. So after MUL a JNC can tell you that the product still fits in AL or AX alone. SF, ZF, AF and PF are undefined: do not test ZF to see whether the product was zero.

Idioms and pitfalls

Multiplying by a power of two is often written as shifts (SHL AX, 1, or SHL AX, CL for larger counts), which also leave DX alone. Remember that DX is clobbered by a 16-bit MUL; save it first if it held something else.

Later CPUs

The 386 adds a 32-bit form, MUL ECX, which multiplies EAX and leaves the 64-bit result in EDX:EAX. MUL never gained an immediate or a two-operand form; those exist only for IMUL.