Scribd
Upload
58 views1 page

323 Assignment 2

1) The document describes an assignment to write a C program that demonstrates floating point addition is not always associative by showing that (A+B)-B may not equal A for all values of A and B. 2) Students are instructed to select values of A and B that produce accurate and inaccurate results, explaining why certain values like 1.36 cannot be represented accurately in 32-bit floating point. 3) The document provides sample code to examine the exponent, sign, and mantissa of floating point numbers to help students select values of A and B.

Uploaded by

tarafe5509
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
58 views1 page

323 Assignment 2

1) The document describes an assignment to write a C program that demonstrates floating point addition is not always associative by showing that (A+B)-B may not equal A for all values of A and B. 2) Students are instructed to select values of A and B that produce accurate and inaccurate results, explaining why certain values like 1.36 cannot be represented accurately in 32-bit floating point. 3) The document provides sample code to examine the exponent, sign, and mantissa of floating point numbers to help students select values of A and B.

Uploaded by

tarafe5509
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Yeditepe University

Department of Computer Engineering

CSE 323 - Computer Organization

Assignment 2

Due to: 26.4.2023 23:59

In floating point addition/subtraction, due to the shift operations to align the mantissas, some errors may occur. Write a
C program to show that, in floating-point arithmetic, the result of (A+B)-B may not always be equal to A. That is, after
the following operations, D may not be equal to A.
C  A+B
D  C-B

1. First, select two numbers for A and B such that when you run your code, the result can be calculated without an error
(that is D is equal to A).

2. Then, select different numbers for A and B and show that D is not equal to A.

In both cases, print A, B, C and D.

In selecting the numbers A and B, make sure that the numbers can be represented in the binary accurately.
For example:
(0.6875)10 can be accurately represented in 32-bit floating format since
1 * 2-1+ 1 * 2-3+ 1 * 2-4= 0.5 + 0.125 + 0.0625 = 0.6875 = 0.1011
Hence, its 32-bit floating point representation would be 0 01111110 01100000000000000000000

However, (1.36)10 cannot be accurately represented in 32-bit floating format, since the fractional part 0.36
=0.01011100001010001111… repeats itself. Hence, its 32-bit floating point representation would be 0 01111101
01110 00010100 011110101

Therefore, you must select numbers A and B such that their fractional parts can be accurately represented in 32- bit
floating format, as in number (0.6875)10.

Some helpful links:


● Online Binary-Decimal Converter http://www.binaryconvert.com/
● Floating Point Number Representation in C
https://www.cprogramming.com/tutorial/floating_point/understanding_floating_point_representation.html

The following code may help you to find the desired numbers:
#include <stdio.h>
#include <stdlib.h>
#include <ieee754.h>

int main(){
float a;
for(;;){
printf("\n\nENTER the value of a:\n");
scanf("%f", &a);
union ieee754_float *p_a;
unsigned int a_exp;
unsigned int a_negative;
unsigned int a_mantissa;
p_a = (union ieee754_float*)&a;
a_exp = p_a->ieee.exponent;
a_negative = p_a->ieee.negative;
a_mantissa = p_a->ieee.mantissa;
printf("exponent of a: %x\n", a_exp);
printf("negative of a: %x\n", a_negative);
printf("mantissa of a: %x\n", a_mantissa);
}
return 0;
}

Submission:
Submit your C program, outputs of the two runs, and your comments on the results. For what kind of numbers
inaccurate results are obtained?

You might also like