The fastest Statistical Programming Language is …Javascript?

#1

Post by TalGalili » Sat, Apr 28, 2012, 7:44 AM UTC

The fastest Statistical Programming Language is …Javascript?
r-bloggers.com

Re: The fastest Statistical Programming Language is …Javascript?

#2

Post by Tichy » Sat, Apr 28, 2012, 10:17 AM UTC

The fascinating question is: why is Javascript fast? I suppose it is because of the competition between the browsers. Yay for competition!

Re: The fastest Statistical Programming Language is …Javascript?

#3

Post by disgruntledphd2 » Sat, Apr 28, 2012, 10:25 AM UTC

Urrgh. This came through on my feed earlier today, and I left a comment on it. While javascript is fast (and can be used for many things), the real issue with using it for stats is the lack of libraries. More specifically, as far as I know it cannot interface with Fortran. That's a death knell for any statistical programming language, as it means no LAPACK, and no-one (sane) is going to rewrite all of those linear algebra libraries. So regardless of how fast it is, its not going to make it as a stats language.

That being said, it does make it easier to develop statistically aware web-apps (a particular interest of mine), so that's definitely good.

Re: The fastest Statistical Programming Language is …Javascript?

#4

Post by NonEUCitizen » Sat, Apr 28, 2012, 10:33 AM UTC

His table shows js is 40x slower on matrix multiplication.

Re: The fastest Statistical Programming Language is …Javascript?

#5

Post by platzhirsch » Sat, Apr 28, 2012, 10:38 AM UTC

NonEUCitizen wrote: post #4

His table shows js is 40x slower on matrix multiplication.

Ergo, JavaScript isn't the fasted language for that matters, because matrix multiplication is too important.

Re: The fastest Statistical Programming Language is …Javascript?

#6

Post by lucian1900 » Sat, Apr 28, 2012, 10:43 AM UTC

PyPy is often faster than v8, so a more complete benchmark should include it.

Re: The fastest Statistical Programming Language is …Javascript?

#7

Post by sycren » Sat, Apr 28, 2012, 10:44 AM UTC

platzhirsch wrote: post #5

NonEUCitizen wrote: post #4

His table shows js is 40x slower on matrix multiplication.

Ergo, JavaScript isn't the fasted language for that matters, because matrix multiplication is too important.

or there is no dedicated library for matrix multiplication compared to the other languages..

Re: The fastest Statistical Programming Language is …Javascript?

#8

Post by regularfry » Sat, Apr 28, 2012, 10:50 AM UTC

disgruntledphd2 wrote: post #3

Urrgh. This came through on my feed earlier today, and I left a comment on it. While javascript is fast (and can be used for many things), the real issue with using it for stats is the lack of libraries. More specifically, as far as I know it cannot interface with Fortran. That's a death knell for any statistical programming language, as it means no LAPACK, and no-one (sane) is going to rewrite all of those linear al…

It does mean that there might be mileage in a native tool wrapping LAPACK et al in v8.

Re: The fastest Statistical Programming Language is …Javascript?

#9

Post by friggeri » Sat, Apr 28, 2012, 10:51 AM UTC

platzhirsch wrote: post #5

NonEUCitizen wrote: post #4

His table shows js is 40x slower on matrix multiplication.

Ergo, JavaScript isn't the fasted language for that matters, because matrix multiplication is too important.

Except that the matrix multiplication benchmark uses pure javascript (see the source here: https://github.com/JuliaLang/julia/blob/master/test/perf/per...). I wonder how JS would do with bindings to LAPACK.

And then there is a bias in those benchmarks, see for example the ones for quicksort: in Python they only time the duration of the sort itself, whereas (at least in Julia and JS) they time both the creation of the random array and the time needed to sort it.

Re: The fastest Statistical Programming Language is …Javascript?

#10

Post by icebraining » Sat, Apr 28, 2012, 10:52 AM UTC

disgruntledphd2 wrote: post #3

Urrgh. This came through on my feed earlier today, and I left a comment on it. While javascript is fast (and can be used for many things), the real issue with using it for stats is the lack of libraries. More specifically, as far as I know it cannot interface with Fortran. That's a death knell for any statistical programming language, as it means no LAPACK, and no-one (sane) is going to rewrite all of those linear al…

What about using https://github.com/rbranson/node-ffi + clapack?

Not that I know what I'm talking about, I've never used Fortran.