1
$\begingroup$

MATLAB/Octave is a pretty useful language for prototyping scientific computation use cases, however I would like to know if the sheer nature of being a high level language bottlenecks how fast the performance can get. (I'm not talking about writing optimal code like vectorization, but a general impact on CPU Time, like how C++ can be made faster than equivalent Python programs by really large magnitudes).

My use case involves running a lot of statistical computations (like Sample Entropy for a 1D temporal time series) and I don't have equivalent code ready in C right now to test my hypothesis.

As we may agree upon as a rule of thumb, lower level languages outperform higher level languages at the cost of code readability and learning curves, but I had also heard that MATLAB interfaces super fast Fortran libraries for many of its numeric computations so I'm unsure how to proceed without unnecessarily wasting my time.

|cite|improve this question
$\endgroup$
  • $\begingroup$ Well yes, although that is a much more general question (and thus I missed it in my searches) it does answer my questions, what do I do with this thread in that case? (Unless there is scope for more productive discussion on this topic) $\endgroup$ – fibonachoceres Mar 16 '17 at 16:38
  • $\begingroup$ "As we may agree upon as a rule of thumb, lower level languages outperform higher level languages at the cost of code readability and learning curves": why not just use Julia? $\endgroup$ – Chris Rackauckas Mar 16 '17 at 19:44

Browse other questions tagged or ask your own question.