Program for conversion of 32 Bits Single Precision IEEE 754 Floating Point Representation

Pre-Requisite: IEEE Standard 754 Floating Point Numbers

Write a program to find out the 32 Bits Single Precision IEEE 754 Floating-Point representation of a given real value and vice versa.

Examples:

Input: real number = 16.75
Output: 0 | 10000011 | 00001100000000000000000

Input: floating point number = 0 | 10000011 | 00001100000000000000000
Output: 16.75

Approach:
This implementation is based on Union Datatype in C and using the concept of Bit Fields.



Bit Fields are assigned when we don’t require the full memory that is usually allocated to some variables but we want to limit the amount of memory taken up by those variables. In C, members of a Union share the common memory space and taken we can access the members only one at a time.

Below is the implementation of the above approach:

Program 1: Convert a real value to its floating point representation

C

filter_none

edit
close

play_arrow

link
brightness_4
code

// C program to convert a real value
// to IEEE 754 floating point representaion
  
#include <stdio.h>
  
void printBinary(int n, int i)
{
  
    // Prints the binary representation
    // of a number n up to i-bits.
    int k;
    for (k = i - 1; k >= 0; k--) {
  
        if ((n >> k) & 1)
            printf("1");
        else
            printf("0");
    }
}
  
typedef union {
  
    float f;
    struct
    {
  
        // Order is important.
        // Here the members of the union data structure
        // use the same memory (32 bits).
        // The ordering is taken
        // from the LSB to the MSB.
        unsigned int mantissa : 23;
        unsigned int exponent : 8;
        unsigned int sign : 1;
  
    } raw;
} myfloat;
  
// Function to convert real value
// to IEEE foating point representation
void printIEEE(myfloat var)
{
  
    // Prints the IEEE 754 representation
    // of a float value (32 bits)
  
    printf("%d | ", var.raw.sign);
    printBinary(var.raw.exponent, 8);
    printf(" | ");
    printBinary(var.raw.mantissa, 23);
    printf("\n");
}
  
// Driver Code
int main()
{
  
    // Instantiate the union
    myfloat var;
  
    // Get the real value
    var.f = -2.25;
  
    // Get the IEEE floating point representation
    printf("IEEE 754 representation of %f is : \n",
           var.f);
    printIEEE(var);
  
    return 0;
}

chevron_right


Python3

filter_none

edit
close

play_arrow

link
brightness_4
code

# Python program to convert a real value
# to IEEE 754 Floating Point Representation.
  
# Function to convert a 
# fraction to binary form.
def binaryOfFraction(fraction):
  
    # Declaring an empty string
    # to store binary bits.
    binary = str()
  
    # Iterating through 
    # fraction until it 
    # becomes Zero.
    while (fraction):
          
        # Multiplying fraction by 2.
        fraction *= 2
  
        # Storing Integer Part of
        # Fraction in int_part.
        if (fraction >= 1):
            int_part = 1
            fraction -= 1
        else:
            int_part = 0
      
        # Adding int_part to binary
        # after every iteration.
        binary += str(int_part)
  
    # Returning the binary string.
    return binary
  
# Function to get sign  bit,
# exp bits and mantissa bits,
# from given real no.
def floatingPoint(real_no):
  
    # Setting Sign bit
    # default to zero.
    sign_bit = 0
  
    # Sign bit will set to
    # 1 for negative no.
    if(real_no < 0):
        sign_bit = 1
  
    # converting given no. to
    # absolute value as we have
    # already set the sign bit.
    real_no = abs(real_no)
  
    # Converting Integer Part 
    # of Real no to Binary
    int_str = bin(int(real_no))[2 : ]
  
    # Function call to convert
    # Fraction part of real no
    # to Binary.
    fraction_str = binaryOfFraction(real_no - int(real_no))
  
    # Getting the index where
    # Bit was high for the first
    # Time in binary repres
    # of Integer part of real no.
    ind = int_str.index('1')
  
    # The Exponent is the no.
    # By which we have right 
    # Shifted the decimal and 
    # it is given below.
    # Also converting it to bias 
    # exp by adding 127.
    exp_str = bin((len(int_str) - ind - 1) + 127)[2 : ]
  
    # getting mantissa string
    # By adding int_str and fraction_str.
    # the zeroes in MSB of int_str
    # have no significance so they
    # are ignored by slicing.
    mant_str = int_str[ind + 1 : ] + fraction_str
  
    # Adding Zeroes in LSB of 
    # mantissa string so as to make
    # it's length of 23 bits.
    mant_str = mant_str + ('0' * (23 - len(mant_str)))
  
    # Returning the sign, Exp
    # and Mantissa Bit strings.
    return sign_bit, exp_str, mant_str
  
# Driver Code
if __name__ == "__main__":
  
    # Function call to get
    # Sign, Exponent and 
    # Mantissa Bit Strings.
    sign_bit, exp_str, mant_str = floatingPoint(-2.250000)
  
    # Final Floating point Representation.
    ieee_32 = str(sign_bit) + '|' + exp_str + '|' + mant_str
  
    # Printing the ieee 32 represenation.
    print("IEEE 754 representation of -2.250000 is :")
    print(ieee_32)

chevron_right


Output:

IEEE 754 representation of -2.250000 is : 
1 | 10000000 | 00100000000000000000000

Program 2: Convert a floating point representation to its real value

C

filter_none

edit
close

play_arrow

link
brightness_4
code

// C program to convert
// IEEE 754 floating point representaion
// into real value
  
#include <math.h>
#include <stdio.h>
  
typedef union {
  
    float f;
    struct
    {
  
        // Order is important.
        // Here the members of the union data structure
        // use the same memory (32 bits).
        // The ordering is taken
        // from the LSB to the MSB.
  
        unsigned int mantissa : 23;
        unsigned int exponent : 8;
        unsigned int sign : 1;
  
    } raw;
} myfloat;
  
// Function to convert a binary array
// to the corresponding integer
unsigned int convertToInt(int* arr, int low, int high)
{
    unsigned f = 0, i;
    for (i = high; i >= low; i--) {
        f = f + arr[i] * pow(2, high - i);
    }
    return f;
}
  
// Driver Code
int main()
{
  
    // Get the 32-bit floating point number
    unsigned int ieee[32]
        = { 1,
            1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1,
            0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
            0, 0, 0, 0, 0, 0, 0, 0, 0 };
  
    myfloat var;
  
    // Convert the least significant
    // mantissa part (23 bits)
    // to corresponding decimal integer
    unsigned f = convertToInt(ieee, 9, 31);
  
    // Assign integer representation of mantissa
    var.raw.mantissa = f;
  
    // Convert the exponent part (8 bits)
    // to a corresponding decimal integer
    f = convertToInt(ieee, 1, 8);
  
    // Assign integer representation
    // of the exponent
    var.raw.exponent = f;
  
    // Assign sign bit
    var.raw.sign = ieee[0];
  
    printf("The float value of the given"
           " IEEE-754 representation is : \n");
    printf("%f", var.f);
}

chevron_right


Python3

filter_none

edit
close

play_arrow

link
brightness_4
code

# Python program to convert
# IEEE 754 floating point representation
# into real value
  
# Function to convert Binary
# of Mantissa to float value.
def convertToInt(mantissa_str):
  
    # variable to make a count
    # of negative power of 2.
    power_count = -1
  
    # variable to store
    # float value of mantissa.
    mantissa_int = 0
  
    # Iterations through binary
    # Number. Standard form of 
    # Mantissa is 1.M so we have 
    # 0.M therefore we are taking
    # negative powers on 2 for 
    # conversion.
    for i in mantissa_str:
  
        # Adding converted value of
        # Binary bits in every 
        # iteration to float mantissa.
        mantissa_int += (int(i) * pow(2, power_count))
  
        # count will decrease by 1
        # as we move toward right.
        power_count -= 1
          
    # returning mantissa in 1.M form.
    return (mantissa_int + 1)
  
if __name__ == "__main__":
    # Floating Point Representation
    # to be converted into real 
    # value.
    ieee_32 = '1|10000000|00100000000000000000000'
  
    # First bit will be sign bit.
    sign_bit = int(ieee_32[0])
  
    # Next 8 bits will be 
    # Exponent Bits in Biased
    # form.
    exponent_bias = int(ieee_32[2 : 10], 2)
  
    # In 32 Bit format bias
    # value is 127 so to have
    # unbiased exponent
    # subtract 127.
    exponent_unbias = exponent_bias - 127
  
    # Next 23 Bits will be
    # Mantissa (1.M format)
    mantissa_str = ieee_32[11 : ]
  
    # Function call to convert
    # 23 binary bits into 
    # 1.M real no. form
    mantissa_int = convertToInt(mantissa_str)
  
    # The final real no. obtained
    # by sign bit, mantissa and
    # Exponent.
    real_no = pow(-1, sign_bit) * mantissa_int * pow(2, exponent_unbias)
  
    # Printing the obtained
    # Real value of floating
    # Point Representaion.
    print("The float value of the given IEEE-754 representation is :",real_no)

chevron_right


Output:

The float value of the given IEEE-754 representation is : 
-2.250000

Don’t stop now and take your learning to the next level. Learn all the important concepts of Data Structures and Algorithms with the help of the most trusted course: DSA Self Paced. Become industry ready at a student-friendly price.




My Personal Notes arrow_drop_up


If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to contribute@geeksforgeeks.org. See your article appearing on the GeeksforGeeks main page and help other Geeks.

Please Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.



Improved By : amit_mangal_