Hacker News Reader

Hacker News

Comment by neonsunset | Hacker News Reader

Full thread

neonsunset

C# let's you write intrinsics-based and cross-platform vectorized kernels without C++ code ;)

(GC does have AVX512-based vxsort written in C++ though)

View on HN

Java does as well, check out the Vector API.

AFAIK the Java Vector API is like Vector<T> not like the intrinsics one

Not quite:

* https://openjdk.org/jeps/417

* https://youtu.be/1JeoNr6-pZw

I expect they are referring to the "Vector API" described in JEP 426: Vector API [1]. From the summary:

    Introduce an API to express vector computations that reliably compile at runtime to optimal vector instructions on supported CPU architectures, thus achieving performance superior to equivalent scalar computations.

[1] https://openjdk.org/jeps/426

Yes. Where does it expose the AVX or ARM intrinsics? I only see generic add/neg/mul methods (thus the comparison to .NET Vector<T>) and not the intrinsics exposed thhemselves; here is a random pick from the .NET ones: https://learn.microsoft.com/en-us/dotnet/api/system.runtime....*)

Reply on news.ycombinator.com