User Guide

General-Purpose Programming 53
24592—Rev. 3.15—November 2009 AMD64 Technology
The SHx instructions (including SHxD) perform shift operations on unsigned operands. The SAx
instructions operate with signed operands.
SHL and SAL instructions effectively perform multiplication of an operand by a power of 2, in which
case they work as more-efficient alternatives to the MUL instruction. Similarly, SHR and SAR
instructions can be used to divide an operand (signed or unsigned, depending on the instruction used)
by a power of 2.
Although the SAR instruction divides the operand by a power of 2, the behavior is different from the
IDIV instruction. For example, shifting –11 (FFFFFFF5h) by two bits to the right (i.e. divide –11 by
4), gives a result of FFFFFFFDh, or –3, whereas the IDIV instruction for dividing –11 by 4 gives a
result of –2. This is because the IDIV instruction rounds off the quotient to zero, whereas the SAR
instruction rounds off the remainder to zero for positive dividends, and to negative infinity for negative
dividends. This means that, for positive operands, SAR behaves like the corresponding IDIV
instruction, and for negative operands, it gives the same result if and only if all the shifted-out bits are
zeroes, and otherwise the result is smaller by 1.
The SAR instruction treats the most-significant bit (msb) of an operand in a special way: the msb (the
sign bit) is not changed, but is copied to the next bit, preserving the sign of the result. The least-
significant bit (lsb) is shifted out to the CF flag. In the SAL instruction, the msb is shifted out to CF
flag, and the lsb is cleared to 0.
The SHx instructions perform logical shift, i.e. without special treatment of the sign bit. SHL is the
same as SAL (in fact, their opcodes are the same). SHR copies 0 into the most-significant bit, and
shifts the least-significant bit to the CF flag.
The SHxD instructions perform a double shift. These instructions perform left and right shift of the
destination operand, taking the bits to copy into the most-significant bit (for the SHRD instruction) or
into the least-significant bit (for the SHLD instruction) from the source operand. These instructions
behave like SHx, but use bits from the source operand instead of zero bits to shift into the destination
operand. The source operand is not changed.
3.3.8 Compare and Test
The compare and test instructions perform arithmetic and logical comparison of operands and set
corresponding flags, depending on the result of comparison. These instruction are used in conjunction
with conditional instructions such as Jcc or SETcc to organize branching and conditionally executing
blocks in programs. Assembler equivalents of conditional operators in high-level languages
(do…while, if…then…else, and similar) also include compare and test instructions.
Compare
• CMP—Compare
The CMP instruction performs subtraction of the second operand (source) from the first operand
(destination), like the SUB instruction, but it does not store the resulting value in the destination
operand. It leaves both operands intact. The only effect of the CMP instruction is to set or clear the
arithmetic flags (OF, SF, ZF, AF, CF, PF) according to the result of subtraction.