👁 Image
NeinMath 1.5.1

NeinMath

👁 Latest package
👁 Download tracker
👁 GitHub status
👁 Code coverage

NeinMath is playing around with arbitrary precision integers, written in pure managed code, not using any unsafe stuff, and a bit faster than the build-in .NET type for integers with a few thousand bits.

To install NeinMath, run the following command in the NuGet Package Manager Console.

PM> Install-Package NeinMath

It's generally based on the integer implementation of this work, but rewritten to not use pointer arithmetic and other fancy things. Thus, it's a bit slower albeit portable.

Note: starting with the new .NET Core this project becomes a bit obsolete, because the performance gains disappear since some improvements have been contributed. 🎉

Performance

Let's start with a simple comparison (time per 100 operations).

Note: ensure you're running a 64-bit process. Handling this with just 32-bits is a huge impediment for both, BigInteger and Integer.

Note: these results are from "my machine". A basic (very basic) benchmark utility is included to verify / disprove them.

Integers

Like BigInteger a structure Integer provides all the operators you would expect from an integer, so it should be quite compatible to existing .NET code. In fact, there are tests based on BigInteger to ensure it computes correctly most of the time.

To get an idea, this is an example for calculating the Greatest Common Divisor:

Integer Gcd(Integer left, Integer right)
{
 var a = left.Abs();
 var b = right.Abs();

 while (b != 0)
 {
 var c = a % b;
 a = b;
 b = c;
 }

 return a;
}

Note: calling left.Gcd(right) is much faster, since the internal implementation is based on a more sophisticated algorithm.

Rationals

Coming sometime... maybe... who knows?