MathSharp, a vector and matrix library written in C# using hardware intrinsics

Only 11 days to go until the official 3.0 release of .Net Core. We're pretty excited about all the new features that were added (such as the hardware intrinsics here). The biggest one we want to play around with is actually the new JsonSerializer. UTF8 vs UTF16 as the underlying storage type could have a huge impact on performance for some of our larger JSON contracts.

"Parsing a typical JSON payload and accessing all its members using the JsonDocument is 2-3x faster than Json.NET with little allocations for data that is reasonably sized (that is, < 1 MB)."

From: https://docs.microsoft.com/en-us/dotnet/core/whats-new/dotne...

I'll definitely be checking this math(s) library out after I get our solutions updated from 2.2. I was playing around with some ray tracing in C#-land and something like this could come in really handy.

I took a look at the code and I was wondering about the implementation performance where there are a lot of if() statements in the low level functions. Like Vector4F DotProduct2D() has if (Sse41.IsSupported) else if (Sse3.IsSupported) else if (Sse.IsSupported) and finally a default implementation. This seems inefficient but I don't know if somehow the booleans are more like compile time flags and the extraneous code disappears in a puff of logic.

The following call which is in MatrixOperations for instance is about cameras - I assume this is some scene-object type of logic regarding 3d graphics?

That does not look like a general-purpose linear algebra library to me.

public static MatrixSingle CreateBillboard(in Vector4FParam1_3 objectPosition, in Vector4FParam1_3 cameraPosition, in Vector4FParam1_3 cameraUpVector, in Vector4FParam1_3 cameraForwardVector)

Looks like this claims to be faster than other libraries but has no benchmarks.

There is already a Microsoft library for this - System.Numerics.Vectors, see documentation at https://docs.microsoft.com/en-us/dotnet/api/system.numerics?...

Not sure how MathSharp compares in performance or usability, however.

> How to use

> TODO

:(

I suggest that you use run-time code generation to create optimized classes that do not check for intrinsic features on each function call.