TheDeveloperBlog.com


C# Float Notes

Float is a floating-point number. The C# language provides this type as a single-precision floating point number representation. Float is less precise than a double. It occupies four bytes and is no larger than an int.


Suffixes. You will need to use the character suffix "f" or "F" on constants you want to be treated as float values. If you don't specify "f" or "F", you will get a compile-time error. The C# compiler treats these values as doubles by default.

Next: This program shows how to specify floats with "f" or "F". It is often preferred to use the uppercase F to reduce confusion.

Suffix
C# program that uses float suffixes

using System;

class Program
{
    static void Main()
    {
	float val1 = 1.001f; // Lowercase f suffix.
	float val2 = 1.002F; // Uppercase F suffix.

	Console.WriteLine(val1);
	Console.WriteLine(val2);
    }
}

Output

1.001
1.002


Test float. The float type can be used in the same ways that other numeric types in the C# language can be used. You can convert floats to strings, either with ToString or through Console.WriteLine. You can use negative floating point numbers.

ToStringConsole.WriteLine

Also: You can use the == equality operator on floats, and other arithmetic operators.

C# program that tests float type

using System;

class Program
{
    static void Main()
    {
	// Use float type.
	float number = 1.5F;
	Console.WriteLine(number);
	// Set to negative value.
	number = -1.001F;
	Console.WriteLine(number);
	Console.WriteLine(number == -1.001F); // Use == operator
	Console.WriteLine(number + 200); // Use + operator
	Console.WriteLine(number.GetType());
	Console.WriteLine(typeof(float));
	Console.WriteLine(float.MinValue.ToString("0"));
	Console.WriteLine(float.MaxValue.ToString("0"));

	// Find the memory usage for a float value.
	long bytes1 = GC.GetTotalMemory(false);
	float[] array = new float[1000 * 1000];
	array[0] = 1;
	long bytes2 = GC.GetTotalMemory(false);
	Console.WriteLine("{0} bytes per float", ((bytes2 - bytes1) / (1000000)));
    }
}

Output

1.5
-1.001
True
198.999
System.Single
System.Single
-340282300000000000000000000000000000000
340282300000000000000000000000000000000
4 bytes per float

The expression typeof(float) in the program, as well as the GetType method, both return the System.Single type. The C# language aliases the float keyword to System.Single so they are precisely equivalent.

TypeofSingle and Double Types

Also: The memory usage of floats in the C# language is four bytes per float, which is the same as the int type.

Int

Tip: If you need to save memory, using floats instead of doubles can be beneficial.


Constant fields. There are several constant fields on the float type. This program shows the float.Epsilon constant, which is the smallest float value greater than zero. The NaN constant represents not a number.

And: The NegativeInfinity and PositiveInfinity constants are tested with float.IsNegativeInfinity and float.IsPositiveInfinity.

C# program that reveals float constants

using System;

class Program
{
    static void Main()
    {
	Console.WriteLine(float.Epsilon);
	Console.WriteLine(float.NaN);
	Console.WriteLine(float.NegativeInfinity);
	Console.WriteLine(float.PositiveInfinity);
    }
}

Output

1.401298E-45
NaN
-Infinity
Infinity


Epsilon. One major problem with comparing two float values is that they may be almost, but not exactly, equal. Your program probably wants to consider the two values equal. This can be solved with an epsilon test.

The float.Epsilon constant is the smallest possible float value. We take the absolute value of the difference between the two values, and see if the difference is less than float.Epsilon. If so, the two values can be considered equal.

Math.Abs

Discussion. So when should you use the float type in your C# programs? Typically, if you need to use floating point numbers, using the double type instead is a better choice because it has more precision and is probably more common.

However: The double type uses more memory than the float type. In a large array this difference becomes relevant.

Arrays

Therefore: Using a float instead of a double can be beneficial if you are sure the loss of precision is immaterial.

Another reason to use float is simply for compatibility (interoperability) with other software that uses floats. It is best to match the types closely to existing software to alleviate inconsistencies between different code bases.


Summary. Float is commonly needed low-level types. Instead, its best uses are for compatibility issues when you need to represent floating-point numbers. Double is usually better for storing large numbers with decimals.

Double