javascript - numpy-like package for node -

during years on python development, i've been amazed @ how much faster things become if manage rewrite code loops though ndarray , something, numpy functions work on whole array @ once. more i'm switching more , more node, , i'm looking similar. far have turned things, none of promising:

• scikit-node, runs scikit-learn in python, , interfaces node. haven't tried it, don't expect gives me cutting edge speed like.
• there rather old, , newer, javascript matrix libraries (sylvester, gl-matrix, ...). in addition not being sure work matrices larger 4x4 (which useful in 3d rendering), seem native javascript (and some, not sure these, use webgl acceleration). great on browser, not on node.

as far know, npms can written in c++, i'm wondering why there no numpy-like libraries node. there not enough interest in node yet community needs kind of power? there hope es6 features (list comprehensions) allow javascript compilers automatically vectorise native js code c++ speeds? possibly missing else?

edit, in response close-votes: note, i'm not asking "what best package xyz". i'm wondering if there technical reason there no package on node, social reason, or no reason @ , there package missed. maybe avoid many opinionated criticism, want know: have 10000 matrices 100 x 100 each. what's best (* correction, reasonable fast) way add them together?

edit2 after more digging, turned out googling wrong thing. google "node.js scientific computing" , there links interesting notes:

• https://cs.stackexchange.com/questions/1693/a-faster-leaner-javascript-for-scientific-computing-what-features-should-i-kee
• http://www.quora.com/can-node-js-handle-numerical-computation-the-same-way-that-languages-like-r-or-julia-can
• javascript , scientific processing?

basically far understand now, no-one has bothered far. also, since there major omissions in js typedarrays (such 64bit ints), might hard add support using npms, , not hacking engine --- defeat purpose. again, didn't further research last statement.

the majority of node work seems in web "full stack" universe, lot less work being done in areas fast numeric processing advantage.

in areas fast numeric processing advantage, python, r, etc have dominant mindshare.

combine 2 facts, , end not lot of people putting effort node numeric processing libraries.