JPH0312735A - Maximum/minimum value selecting circuit - Google Patents

Abstract

PURPOSE:To perform the comparison between two numerical data and to output the maximum or minimum of both data at high speed by propagating the difference of values of each digit of both numerical data to the lowest-order digit from the highest- order digit. CONSTITUTION:The logical gates 1 - 23 produce the value deciding function gij and the value holding function Pij of each digit. Simultaneously, the logical gates 23 - 24 produce the value deciding functions and the value holding functions of the digits covering a high-order one through its own one. The logical gates 47 - 54 select the input data X1 and Y1. The gij (0<=i<=7) and the Pij (0<=i<=7) of each digit itself are obtained from the input data by the prescribed computing operations. Thus g77 is equal to 1 with X7<Y7 and Y7 is outputted to Z7 from a gate 36. In this case, g67, g47 and digit borrow g07 are all equal to 1 and Yi (0<=i<=7) is outputted to Zi (0<=i<=7). When Xj YYj is satisfied with Xj (4<=j<=7) and Yj (4<=j<=7), g47 and P47 are equal to 0 and 1 respectively and Xj is outputted to Zj. Furthermore it is easily known at a gate 43 that Yk (0<=k<=3) is outputted with X3<Y3. An output selection signal is divided into g77, g67, g47 and g07 and therefore the load of each selection signal is never increased at one time despite the increase of the number of digits.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an arithmetic device, and more particularly to an arithmetic device that calculates the maximum or minimum value of two pieces of data.

BACKGROUND OF THE INVENTION Conventionally, the maximum value or minimum value of two numerical data has been determined by the following method. First, two numerical data are subtracted, and at this time, the magnitude relationship between the two numerical data is determined depending on whether a digit is borrowed from the most significant digit.
One of the two numerical data was selected and output. FIG. 2 shows an example of a configuration for realizing the method described above.

In Figure 2, 8-digit numerical data X = X7XsX6X4
X3X2XI X@ and Y = yT ys V s Y
Whether a digit is borrowed from the most significant digit by subtraction X-Y with a V s V 2V + ya is indicated by a digit borrow b7. In other words, when b7 is "0", no borrowing from the most significant digit occurs, and when b7 is "1", borrowing from the most significant digit occurs. Furthermore, if borrowing occurs, the magnitude relationship between X and Y becomes X<Y, and if borrowing does not occur, X≧Y. logic game) 100~
107 outputs X when b7 is "0" and Y when b7 is "1" (7) TeZ = Z? Zs, Zs ZJ
Zs Zg Z+ Z@ represents the maximum value of X and Y.

Here, the logic for obtaining b7 from X and Y will be explained in detail. Generally, 0-digit data W=v, w, -,...
WIN-to v=vllV, -1...v, va mino calculation W
- In ■, borrow a digit from the most significant digit. is a borrow generation function■. +n-1+digit borrow propagation signal 1pn+s-1+n
It is expressed by the following formula by borrowing bn-1 from -1 digit.

t)+"lH+ +a-+"lpn+n-+ ・1)n
-1-1i) Here, for i and j where i≧j, the borrow generation function '@ l * J means that when subtraction is performed from the j-th digit to the i-th digit, a borrow is generated in the upper order. Represents +,,.

1 indicates that when subtraction is performed from the j-th digit to the i-th digit, if a digit is borrowed from a lower digit, the carry is propagated to an upper digit. From this, the borrow generation function 1g1.1 and the borrow propagation function 11.1 for each digit are given as follows from the value wl +Vl of each digit of the input data. That is, +1. , Knee W, ・V...(2) 1 m + , + ”Komachi 1 ``V + or 11
+Vl...(3) holds true. Also, t+J and k have the relationship i≧j≧
The following relational expression holds true for . In other words,! . +, b: Ig++1"Iel, +・I@J-1,k
...(4) Iel, b: Iel, t・Iel
-1,k...(5). this(
2) to (4), a borrow generation function L11 from digit k to digit i serving as a certain reference. 1 and the number of negative reactions
11. k can be found. That is, from each digit to be subtracted, find the borrow generation function IQ l , I and the borrow propagation function 1@1.1 for each digit represented by (2) and (3), and then calculate (4) and Equation (5) is repeatedly used to generate a borrow generation function 1++I+ from a certain reference digit k to each digit, and a borrow propagation function 1p1. k can be found. From the above, when considering subtraction of all digits, the i-th digit borrow bl is given by the following formula, which is an extension of (1): bl:I. l+1"Iel-+, 3・bj-1...(
6) Also, if we set j=O in equation (6), then 1) l”ll meeting + summer 9th - goose, ・ ・ b-1 ・ ・
・(7) becomes. Here, means borrowing to the least significant digit, and by setting b-1 to "0", you can find the borrowing from each digit for the operation of W-■, and 1
By setting = n, it is possible to find the borrowing of digits from the most significant digit. Figure 2 shows bl, +1.1.
111 is shown by entering the actual values in i and j. As mentioned above, in the conventional configuration, the borrow from each digit for the operation of The maximum or minimum value of X and Y was determined by selecting either one of X and Y.

Problems to be Solved by the Invention In the configuration of the data maximum/minimum value selection circuit as explained above, if it is known whether the borrowed digit 1°7.7 of the own digit, such as the most significant digit z7, is O or I. Even if it is possible to immediately decide whether to select digits x or 7, if you do not wait for the digit borrowing from the lower digit, then 177 selections could not be determined. Also, since all digits of data are selected only by borrowing from the most significant digit, the larger the number of digits in the data, the greater the load on borrowing from the most significant digit, resulting in slower output of maximum and minimum values. We had the challenge of becoming.

The present invention has been made in view of the above points, and by changing the configuration of the circuit that generates the selection signal of the maximum and minimum data values, it is possible to output the maximum and minimum data values with the same circuit scale as the conventional one and at a higher speed than the conventional one. The purpose of the present invention is to provide a maximum/minimum value selection circuit that enables the maximum/minimum value selection circuit.

Means for Solving the Problems In order to solve the above problems, the present invention compares the values of each digit of two numerical data, and calculates the magnitude relationship of each digit from the most significant digit to the least significant digit. By propagating toward the digit, the two numerical data are compared and the maximum or minimum value thereof is output.

Effect of the Invention With the above-described configuration, the present invention propagates the magnitude relationship of each digit of two numerical data from the most significant digit to the least significant digit. In other words, the magnitude relationship is determined in order from the most significant digit to the least significant digit, and tt j that satisfies i>j
Lie on X = X, X, -+ ...X+ Xe-Y
” 3'*711-+...3'+ 3'@2 horned hawk, when xI>y, there is no need to compare the magnitude relationship of the lower digits than the first digit, and X>Y clearly holds true. That is, by determining the magnitude relationship of each digit in order from the most significant digit to the least significant digit, the magnitude relationship of the numerical data of 2 can be determined at high speed.

Embodiment FIG. 1 is a circuit diagram of a maximum/minimum value selection circuit in an embodiment of the present invention. In FIG. 1, I to 23 are logic gates for generating a magnitude relationship determination function and a magnitude relationship reservation function for each digit, and 24 to 34 are magnitude relationship determination functions and magnitude relationship reservation functions from the upper digit to its own digit. Logic gates 47 to 54 for generating input XnCn=0. I, 2.3.4.5
．． e, 7) or Y++ (n=o, I.

2, 3.4.5. llf, 7) and Zn (n=0. 1.2. 3.4. 5.6.
7), and 35 to 4B are inverters for logic inversion. Here, the above-mentioned magnitude relationship determining function and magnitude relationship reservation function will be explained. i≦j
is i.

In j, the magnitude relationship determination function (hereinafter referred to as determination function) g
l, + indicate that a specific magnitude relationship has been determined from the i-th digit to the j-th digit. P+ and + (hereinafter referred to as pending functions) indicate that the specific magnitude function from the i-th digit to the j-th digit has not been determined yet. From this, the decision functions g+, +* and reservation functions P+, j for each digit are input data XI+7. It is given as follows. That is, g1. + """XI" Y I...(7) h, +=L--71- or: □, +7...(8) holds true. Also, I, which fosters the relationship i≦J:ak,
The following relational expression holds true for J and k. That is, g +, *= g +, +P+, + ” g 1.J
−1°゛(9)h, h=P1. k” P+, No 1
...(10). These (7) to (10)
) determination function g+, *t reservation function P1. k can be found. Based on the above, the actual operation of the circuit shown in FIG. 1 will be specifically explained. First, the decision function g+, +(0≦1≦7) of each digit and the reservation function P+,+(
0≦1≦7) are obtained. If XT<YT, g1□ becomes "!", so y7 is output from logic gate 3G to 27. At this time, g77 becomes "l", so g 671 g 47 + g a7 are all.
■” becomes Z+ (0:a 1≦7) Niha Y + (0≦
1≦7) Force is output. Also, X, (4≦j≦7), Y+
(4≦j≦7), when Xj≧y1, ga7 is I
IQ +1. Since p, 7 becomes "Ill", z, (
If 4≦j≦7), χ, (4≦j≦7) is output. In addition, in the logic gate 43, z until glI3 is determined.
, (φ≦≦3), Xk (O≦≦3), yk (0≦
I don't know which one (≦3) will be output, but if X3<
It is immediately clear that when y3, yb (0≦≦3) is output. In this embodiment, the maximum value output has been described, but the minimum value output is configured in the same manner. As you can see from the above explanation, if you compare the magnitude relationship of two numerical data for each digit from the most significant digit to the least significant digit and propagate the results sequentially, you can quickly reach the maximum value of the two numerical data. You can choose a value or a minimum value. moreover(
Equations 2) to (5) and Equations (7) to (10) have different variables but are logically the same, so the circuit scale is almost the same as the conventional one. Note that, as shown in Figure 1, the signal for selecting the output is not a signal common to all digits, but g7? I
g 6? Since it is divided into l g 47 * g a7, even if the number of digits increases, the load on each one does not increase at the same time.

Effects of the Invention As described above, according to the present invention, by propagating the magnitude relationship for each digit of data from the most significant digit to the least significant digit, the circuit scale is the same as that of the conventional one, and the circuit size is the same as that of the conventional one. It is possible to obtain maximum and minimum value output of data at high speed, and is extremely useful in practice.

[Brief explanation of the drawing]

FIG. 1 is a circuit diagram of a maximum/minimum value selection circuit according to an embodiment of the present invention, and FIG. 2 is a circuit diagram of a maximum/minimum value selection circuit showing the configuration of a conventional method. 1 to 23... Logic gates that generate magnitude relation determination functions and magnitude relation reservation functions for each digit, 24 to 34 ... magnitude relation determination functions and magnitude relation reservation functions from the upper digit to its own digit. Logic gates to generate, 47 to 54...Logic gates to select input data.

Claims (1)

[Claims] The two numerical data are compared by comparing the values of each digit of the two numerical data and propagating the magnitude relationship of each digit found from the result of this comparison from the most significant digit to the least significant digit. , a maximum/minimum value selection circuit which selects and outputs the maximum value or the minimum value from the two numerical data.