8-18 Vol. 1

PROGRAMMING WITH THE X87 FPU

8.3.4 

Load Constant Instructions

The following instructions push commonly used constants onto the top [ST(0)] of the x87 FPU register stack:

FLDZ

Load +0.0

FLD1

Load +1.0

FLDPI

Load π

FLDL2T

Load log

2

 10

FLDL2E

Load log

2

e

FLDLG2

Load log

10

2

FLDLN2

Load log

e

2

The constant values have full double extended-precision floating-point precision (64 bits) and are accurate to 
approximately 19 decimal digits. They are stored internally in a format more precise than double extended-preci-
sion floating point. When loading the constant, the x87 FPU rounds the more precise internal constant according 
to the RC (rounding control) field of the x87 FPU control word. The inexact-result exception (#P) is not generated 
as a result of this rounding, nor is the C1 flag set in the x87 FPU status word if the value is rounded up. See 
Section 8.3.8, “Approximation of Pi,” for information on the π constant.

8.3.5 

Basic Arithmetic Instructions

The following floating-point instructions perform basic arithmetic operations on floating-point numbers. Where 
applicable, these instructions match IEEE Standard 754:
FADD/FADDP

Add floating point

FIADD

Add integer to floating point

FSUB/FSUBP

Subtract floating point

FISUB

Subtract integer from floating point

FSUBR/FSUBRP

Reverse subtract floating point

FISUBR

Reverse subtract floating point from integer

FMUL/FMULP

Multiply floating point

FIMUL

Multiply integer by floating point

FDIV/FDIVP

Divide floating point

FIDIV

Divide floating point by integer

FDIVR/FDIVRP

Reverse divide

FIDIVR

Reverse divide integer by floating point

FABS

Absolute value

FCHS

Change sign

FSQRT

Square root

FPREM

Partial remainder

FPREM1

IEEE partial remainder

FRNDINT

Round to integral value

FXTRACT

Extract exponent and significand

The add, subtract, multiply and divide instructions operate on the following types of operands:

•

Two x87 FPU data registers

•

An x87 FPU data register and a floating-point or integer value in memory

See Section 8.1.2, “x87 FPU Data Registers,” for a description of how operands are referenced on the data register 
stack.
Operands in memory can be in single-precision floating-point, double-precision floating-point, word-integer, or 
doubleword-integer format. They are converted to double extended-precision floating-point format automatically.