JPH0667894A - Power function processing system - Google Patents

Abstract

PURPOSE:To execute fast power function processing by providing power function processing system with a 15th power judging means for judging whether the calculation of the 15th power is necessary or not during the calculation and a 15th power calculating means for fast executing 15th power calculation. CONSTITUTION:In the case of processing a power function by a FORTRAN compiler in the power function processing system of a computer system, power is rapidly calculated by power function processing 1 including the 15th power judging means 2 for judging whether 15th power calculation is necessary or not in calculation and the 15th power calculating means for rapidly calculating 15th power. This calculation makes it possible to find out a 15th power value by the multiplication frequency of floating point variables which is less than a convensional method only by once. Namely the multiplication of 4 floating point variable can be reduced as compared with convensional algorithm.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an F system on a computer system.
The present invention relates to a power function processing method of an ORTRAN compiler.

[0002]

2. Description of the Related Art FORT in a conventional computer system
The exponentiation function processing method of the RAN compiler uses the binary calculation method to perform exponentiation.

FIG. 3 shows an example of a flow chart for calculating A = X ** N by an algorithm of a conventional power function function processing method. As shown in FIG. 3, first, the output decimal point variable A is initialized to 1, and the temporary floating point variable T
Is initialized with X (step 21). Next, a remainder obtained by dividing N by 2 is obtained (step 22), and if 1, A = A *
5 is calculated (step 23).

Next, a quotient of N = N / 2 is calculated (step 24). If N is 0 or less, the process ends (step 2
5) Otherwise, calculate T = T * T (step 26) and return to step 22. The power is calculated by this repetition.

For example, using this algorithm, X **
Calculating 15, requires 8 multiplications of the floating point variable.

[0006]

In the conventional power function processing method of the FORTRAN compiler in the computer system, since the power calculation is performed using the binary calculation method, the result is always obtained with the minimum number of multiplications. There is a problem in that

[0007]

The exponentiation function processing method of the present invention is the exponentiation function processing method of a computer system.
When processing the exponentiation function with the ORTRAN compiler,
It is configured to calculate the exponentiation at high speed by the exponentiation function process having means for determining whether or not the exponentiation of the 15th power is necessary during the calculation and means for calculating the 15th power at high speed.

[0008]

Embodiments of the present invention will now be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the power function processing method of the present invention. As shown in FIG. 1, the exponentiation function process 1 includes a fifteenth power determination unit 2 that determines whether or not a fifteenth power is required during calculation, and a fifteenth power calculation unit 3 that performs a fifteenth power calculation at high speed. Have

FIG. 2 shows an example of a flow chart for calculating A = X ** N by a new algorithm of the power function function processing method of the present invention. As shown in FIG. 2, first, the output floating point variable A is initialized to 1, and the temporary floating point variable T is initialized to X (step 10).

Next, the remainder obtained by dividing N by 16 is obtained (step 11). If not 15, the remainder obtained by dividing N by 2 is obtained (step 12), as in the conventional algorithm.
If 1, then A = A * T is calculated (step 13). Next, the quotient of N = N / 2 is calculated (step 14), and N is 0.
If it is below, the process is terminated (step 15), and if not, T = T * T is calculated (step 16) and the process returns to step 11.

When the remainder of N / 16 becomes 15 in step 11, T1 = T * T, T1 = T1 * T, T2 =
T1 * T1, T2 = T2 * T2, T2 = T2 * T1, A
= A * T2 is calculated (step 17).

Next, a quotient of N = N / 16 is calculated (step 18). If N is 0 or less, the process is terminated (step 19). Otherwise, T = T * T2 is calculated. (Step 20), the process returns to step 11. The power is calculated by this repetition.

By this calculation, the 15th power can be obtained with the number of multiplications of the floating-point variable being seven, which is one less than in the conventional method. That is, in this new algorithm, each time the determination condition of step 11 is satisfied, one multiplication of the floating point variable is less than that of the conventional algorithm. For example, if X ** 15 is calculated, the floating point can be multiplied by a number in 7 times.

[0014]

As described above, the power function processing method of the FORTRAN compiler in the computer system of the present invention has a fifteenth power judgment means for judging whether or not fifteenth power is necessary during the calculation, and a fifteenth power judgment means. High speed calculation 15
The power calculation means has the effect that the power function processing can be executed at high speed.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a power function processing method according to the present invention.

FIG. 2 shows an example of a flowchart for calculating A = X ** N by a new algorithm of the power function function processing method of the present invention.

FIG. 3 shows an example of a flowchart for calculating A = X ** N by an algorithm of a conventional power function function processing method.

[Explanation of symbols]

 1 Power Function Processing 2 15 Power Determining Means 3 15 Power Computing Means

Claims (1)

[Claims] 1. A means for processing a power function in a computer system, for processing a power function by a FORTRAN compiler, a means for determining whether or not a power of 15 is necessary during the calculation, and a high speed calculation of the power of 15 A power function processing method of a FORTRAN compiler in a computer system, wherein the power function is calculated at high speed by a power function processing having a means for performing.