Making cross platform, compute intense vector problems less... vexing.
Vecxt aims to free you from the Architectural tyranny of having to choose which platform you run your
Aims to provide convienent and intuitive syntax for vector computations, whilst guaranteeing best-in-class performance on all platforms.
| JVM | JS | Native | |
|---|---|---|---|
| Data structure | Array[Double] |
Float64Array |
Array[Double] |
| Shims to | https://github.com/luhenry/netlib | https://github.com/stdlib-js/blas | CBLAS |
https://github.com/dragonfly-ai/narr
Surprise! Add that dependancy, express your computation in NArray and get best-in-class, native performance (no runtime overhead) on all platforms.
Note: Narr is not distributed "out the box", but it is recommended for anything cross platform.
Supported operationsa re currently very basic, but we share a single test suite across all platforms, e.g.
val v1 = NArray[Double](1.0, 2.0, 3.0)
val v2 = NArray[Double](3.0, 2.0, 1.0)
val v3 = v1 + v2
assertEqualsDouble(v3(0), 4, 0.00001)
assertEqualsDouble(v3(1), 4, 0.00001)
assertEqualsDouble(v3(2), 4, 0.00001)
Not?