ScalaCL is still very young but you can already play with it in a few seconds… Here’s how :


  • You must have an OpenCL implementation installed on your computer : it’s automatically the case under MacOS X 10.6 (Snow Leopard) or later, you’ll need to install one from NVIDIA, ATI or Intel otherwise (see ScalaCL’s wiki for more information)
  • You need to install Scala 2.8.1. If you haven’t yet, I suggest running the IzPack Installer (on Windows 64 bits, make sure to install it in `c:\Program Files`, not `c:\Program Files (x86)`)

Start your chronometer

  • Install ScalaCL’s compiler plugin & collections using sbaz (sbaz is part of the Scala distribution) :

    sbaz update
    sbaz install scalacl
  • Open a Scala console in command line :

  • Add some imports :

    import scalacl._ ; import scala.math._ 
  • Create an OpenCL context (has to be ‘implicit’) :

    implicit val context =
    // prefer CPUs ?
  • Create an OpenCL collection :

    val a = (0 until 100000).cl // this gives a CLRange 
  • Run some maps and filters on your GPU :

    val result = => cos(x / 100.0f).toFloat).zipWithIndex map { case (c, i) => c * 10 + i } filter { v => (v.toInt % 2) == 1 } 
  • Need to get back an Array[Float] ? CLArray[T], CLRange and other CLIndexedSeq[T] collections behave as expected :


That’s it !

You’ve just run some Scala code on your graphic card !!!

Wanna get juicy details on what’s happening under the hood ? Set the `SCALACL_VERBOSE` environment variable to `1` before launching the Scala console (or the Scala compiler) :

  • On Windows : set SCALACL_VERBOSE=1
  • With a bash shell (MacOS X and Linux default) : SCALACL_VERBOSE=1 scala
  • With a tcsh shell : env SCALACL_VERBOSE=1 scala

Reactions ? Join NativeLibs4Java’s Google Group !

Wanna learn more ? Read ScalaCL’s FAQ.