Decimal-32 floating point converter

About The Project

Description

The website serves as a decimal-32 floating-point converter, capable of processing decimal inputs in base-10 exponential format and converting them into binary and hexadecimal forms. The 32-bit value is partitioned into 1 bit for the sign, 5 bits for the combination field, 6 bits for the exponent continuation, and 20 bits for the coefficient continuation. Special cases such as NaN and Infinity are also detected and converted. Access the website through this link

Steps

1. Check if input is valid
2. Normalize the input
3. Shift input (if possible)
4. Choose rounding method
5. Get Sign Bit from input
6. Get Combinational Field from input
7. Get Exponent Continuation from input
8. Convert remaining input to Densely Packed BCD
9. Compile and Display

Problems encountered

1. Javascript's floating point representaion

Explanation: Some of the floating point numbers were represented differently when operations were performed. For example, 0.5 minus 0.4 resulted in 0.09999999999999964.
Fix: The input was taken as a string instead of an integer.

2. Rounding with strings

Explanation: Since the inputs are stored as strings, rounding '9' up to '10' was a challenge 
Fix: The input was first parsed into an integer before incrementing. Afterwards, it will be parsed back into a string.

3. Exponents with floating points

Explanation: Exponents with floating points are being rounded and converted.
Fix: The convert button is disabled when there is a radix point in the input.

4. Integer representaion of javascript

Explanation: If the input is larger than a certain number, it is not being represented properly.
Fix: When the number is too large, 'Infinity' is shown instead of the actual value.

5. Whitespaces in input produces NaN

Explanation: Inputs with white spaces are being accepted as they are not marked as NaN by JavaScript's isNaN function.
Fix: Self-created checker for white spaces were made.

6. Representable inputs with >90 exponents

Explanation: Some inputs are still representable even if the exponent is greater than 90. For example, 1 x 10^91 can be represented as 10 x 10^90.
Fix: A checker was created to verify if there are '0's on the left side that can be truncated so that a '0' can be added on the right side.

Usage

Inputs

Decimal: Base-10 Number
Exponent: Base-10 Number (No Floating Points)
Rounding mode: Truncate/Floor/Ceiling/RTN-TE

Outputs

Rounded number: 7-Digit Base-10 Number
Exponent after rounding: Base-10 Number (w/o Bias)
Binary Form: 32-Bit Conversion of the Input
Hexadecimal Form: 8-Digit Hexadecimal Conversion of the Input

Demo

The video demo may also be accessed through this link
Access the test cases here

Authors

1. Cabrera, Jean
2. Cabrera, Alyanna
3. Lu, Bentley
4. Tan, Arvin