Multithreading in Julia

In this recipe, we use Julia's multithreading mechanism to speed up computations of statistics over a largeDataFrame.

Getting ready

For this recipe, you need to use theDataFrames.jlandBenchmarkTools.jlpackages, which can be installed with the Julia package manager. If you need to add them then in the Julia command line, press the] key and run these commands:

(v1.0) pkg> add DataFrames
(v1.0) pkg> add BenchmarkTools

This will install theDataFrames.jlandBenchmarkTools.jlpackages and all their dependencies.

 

In this recipe, we use Julia's threading mechanism. The number of threads is controlled via theJULIA_NUM_THREADSsystem variable, which needs to be assigned before starting thejuliaprocess. In order to set the variable, start a new command-line console and execute this command (Windows version):

C:\ set JULIA_NUM_THREADS=4

In a Linux console, run this:

$ export JULIA_NUM_THREADS=4

Now, start Julia:

$julia

You can check whether the number of threads has been properly allocated...