Statistically accurate benchmarking
The tools described in this chapter, particularly the @time macro, are useful to identify and investigate bottlenecks in our program. However, they are not very accurate for a fine-grained analysis of fast programs. If you want to, for example, compare two functions that take a few milliseconds to run, the amount of error and variability in the measurement will easily swamp the running time of this function.
Using Benchmarks.jl
The solution then is to use the Benchmarks.jl package for statistically accurate benchmarking. This package is not yet published in the official repository, but is stable and high-quality nevertheless. It can be installed with Pkg.clone("https://github.com/johnmyleswhite/Benchmarks.jl.git") ...
Get Julia: High Performance Programming now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
