COBOL Float Data Type

Float data type is aimed to store floating-point numbers. Depending on values, float data type can be 4- or 8-byte size. There are two parts in floating-point numbers: (1) mantissa and (2) exponent. To get a floating-point value, the mantissa is multiplied by 10 raised to the power of the exponent.

The summary of float data type can be found in COBOL Data Types article.

Storage Description

Sample

35791 value

3.5791 is the mantissa and 4 is the exponent.

3.5791 * (10 ** 4) = 35791

FLOAT 4 bytes

It contains 8 digits in mantissa and it's called single floating point data type. The leftmost 8 bits stores exponent and the remaining 24 bits stores mantissa.

FLOAT 8 bytes

It contains 16 digits in mantissa and it's called double floating point data type. The leftmost 12 bits stores exponent and the remaining 52 bits stores mantissa.

COBOL Representation

Samples 4 bytes

01 VAR-FLOAT PIC 9(5) USAGE COMP-1.
01 VAR-FLOAT PIC 9(5) COMP-1.
01 VAR-FLOAT PIC 9(5)V9(2) COMP-1.

Samples 8 bytes

01 VAR-FLOAT PIC 9(05) USAGE COMP-2.
01 VAR-FLOAT PIC S9(5) USAGE COMP-2.
01 VAR-FLOAT PIC 9(5)V9(2) USAGE COMP-2.
01 VAR-FLOAT PIC S9(5)V9(2) COMP-2.

Resources

ebcdic-parser tool for converting of mainframe EBCDIC data into Unicode ASCII delimited text
FLOAT(n)