JPH01159727A - Data processor - Google Patents

Abstract

PURPOSE:To speed up the magnitude comparison of numerical value by binary comparison for every processing unit by expressing the internal expression of numerical value data by floating decimal point expression, and making the length of an exponential part including a sign and a mantissa part be integer times of the processing unit. CONSTITUTION:The numerical value data held in a storage device 11 is called out to a data converting part 12 by the instruction of a controller part 13. The data converting part 12 converts this numerical value data into the internal expression of the floating decimal point corresponding to the length, etc. of a base and the exponential part to be given from the controller part. This internal expression is expressed by the sign to show the positive or the negative of the mantissa part, the characteristic part and the mantissa part. This internal expression is normalized preliminarily and it is constituted so as to show a minimum value when both the exponential part and the mantissa part including the sign are 0. When the data of such the internal expression is compared, the magnitude can be determined by the binary comparison without being conscious of scaling, the sign, precision, etc.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a data processing device that performs comparison operations on numerical data.

(Prior Art) Generally, data processing devices such as database machines that process data in various formats often perform comparison operations between numerical data in different formats. In such cases, conventionally, it has been necessary to compare the formats of each numerical data by taking into consideration the sign, scale, precision, etc.

For this reason, the comparison calculation process becomes complicated, and the data processing device itself, which uses a lot of comparison calculations, also becomes complicated, which has the drawback of increasing the processing time.

(Problems to be Solved by the Invention) As mentioned above, conventional data processing devices have the disadvantage that processing time increases when numerical data comparison processing is frequently used, and the device itself becomes complicated. .

The present invention has been made in view of the above points, and it is an object of the present invention to provide a data processing device that can simplify comparison operations of numerical data of different formats and improve processing speed.

[Structure of the Invention (Means for Solving the Problems)] This invention represents the internal representation of numerical data in a normalized floating point representation in a data processing device that performs comparative calculation processing of numerical data, and The sign bit in floating point representation is 1 when the mantissa is positive or zero, and 0 when it is negative. Convert numeric data to consist of bits, and convert this converted 2
The system is configured to compare two numerical data for each processing unit starting from the code bit and determine the magnitude of the numerical data.

(Function) In the data processing device configured in this way, when comparing numerical data, there is no need to consider the sign, scale, precision, etc. of the data format; Since it is possible to determine the magnitude by comparing with , it is possible to simplify the comparison calculation process.

(Example) Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a data processing device according to an embodiment of the present invention.

In the figure, numeral 11 is a storage device that stores processed data, and holds numerical data in various formats. Reference numeral 12 denotes a data conversion section, which reads numerical data stored in the storage device 11 according to instructions from the controller section 13 and converts it into an internal representation to be described later. Reference numeral 14 denotes a data comparison section, which is controlled by the controller section 13 and performs a binary comparison operation on the numerical data converted into internal representation by the data conversion section 12 for each appropriate processing unit.

The operation of the data processing apparatus configured in this way will be explained.

First, numerical data held in the storage device 11 is called to the data conversion section 12 in accordance with an instruction from the controller section 13 that controls the entire apparatus. The data conversion unit 12 is
The called numerical data is converted into an internal representation corresponding to the base, the length of the exponent, etc. given from the controller section 13. This internal representation is configured as a type of floating point representation as shown in FIG. In the figure, the most significant bit (MSB) is a sign bit that indicates the significand, the n bits after this sign bit represent the exponent, and the n+1 to m bits thereafter represent the mantissa. It represents. The above sign bit indicates 1 when the mantissa is positive or zero, and o when it is negative, and the bit length n of the exponent part including the sign bit, and the bit length m-n of the mantissa part.
is an integral multiple of the processing unit. Further, this internal representation is normalized and is configured to indicate the minimum value when both the exponent part and the mantissa part including the sign are O. For example, if the exponent part consists of 7 bits, -64
゜. , all bits are 0, 63.1o,
127, , in the conventional display format. The bit configuration is the same as that shown in .

Further, as for the base of internal representation (floating point representation) and the length of the exponent part including the sign of the mantissa part, one appropriate value is determined between the two data to be compared. Furthermore, the length of the mantissa part is variable and can be an arbitrary length that is an integer multiple of the processing unit for each data.

In this way, the base and exponent lengths are arbitrary under the condition that the two numerical data being compared are the same, so this problem occurs, for example, when converting decimal data to hexadecimal floating point representation. It is possible to configure internal representations so that errors and confusion do not occur. In addition, since the mantissa is variable for each data, when comparing numerical data, if the values are the same, the longer data length can be determined to be larger and the comparison can be made correctly, which improves the accuracy of the individual data. No need to consider. Furthermore, the internal representation of numerical data is such that the minimum value is when all the bits constituting the data are 0, so during comparison operations, the sign and
The magnitude of numerical data can be easily and efficiently determined by simple binary comparison in the order of exponent and mantissa.

In this way, when comparing numerical data in different formats, by comparing the data converted to the internal representation configured as above, it is possible to compare numerical data from the beginning without worrying about scale, sign, precision, etc. Comparison operations can be easily performed because the magnitude can be determined simply by performing a binary comparison for each processing unit.
Improving the processing speed of data processing devices that use a lot of comparison operations,
And the device itself can be simplified. Furthermore, by appropriately determining the lengths of the base and exponent part, it is possible to eliminate errors and blurs that occur during conversion to internal representation.

[Effects of the Invention] As described above, according to the present invention, the internal representation of numerical data in a data processing device is expressed in normalized floating point representation, and the exponent part including the sign of the mantissa part and the length of the mantissa part are By using a numerical data format in which the value is an integer multiple of the processing unit, and the exponent and mantissa including the sign are all 0, the minimum value is expressed, it is possible to simply calculate each processing unit from the beginning of the internal representation. Since the magnitude of numerical values can be determined by binary comparison, the speed of comparison calculation processing can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a data processing apparatus according to an embodiment of the present invention, and FIG. 2 is a diagram showing the format of internal representation of numerical data in the embodiment. 11...Storage device, 12...Data converter, 13.
...Controller section, 14...Data comparison section. Applicant's agent Patent attorney Takehiko Suzue

Claims (1)

[Claims] In a data processing device that performs comparison operations on numerical data, the internal representation of numerical data is expressed in a normalized floating point representation, and the sign bit in this floating point representation is set to 1 when the mantissa is positive or zero, and negative. , the numerical data conversion means converts the numerical data so that the exponent part including the sign bit and the mantissa part continuous to the exponent part are composed of bits that are an integral multiple of the processing unit; 1. A data processing device comprising: numerical data comparing means for comparing two numerical data converted by the numerical data converting means for each processing unit starting from the code bit, and determining the magnitude of the numerical data.