OjAlgo's use of SIMD instructions

[sonideft]

I understand that OjAlgo did something similar internally to jdk.incubator.vector to make use of SIMD instructions; for faster processing. I have two questions:

1. Is there a way to tell if certain matrix or vector array operations use SIMD instructions?
2. Is OjAlgo planning to make use of the jdk.incubator.vector SIMD ability in a near-term release?

While I solve a lot of A x = B matrix's, which have been well documented as being very fast already, I also have a number of vectors ( 1 row, many columns) that I need to multiply, sum, etc and would like to know if those available functions already benefit from SIMD instructions. If they do, I won't bother trying to invent my own methods.

[apete]

With Java getting SIMD instructions in indirect. The JVM has support for auto-vectorisation, but all you can do is to write code that is likely to be vectorised. ojAlgo does this, and vectorisation kicks in in many different situations. This is part of why ojAlgo performs well.

As soon as the vector api or any part of Valhalla is released as finalised (not previews or incubating) ojAlgo will make use of it.

The fact that the vector api has been incubating for so long is unfortunate. It's waiting for Valhalla, and that's still several versions/releases away.

[sonideft]

Thanks for the clear and direct answer.