I would title this blog post by Doug Bates: "Julia: Fast for Free" http://dmbates.blogspot.com/2012/05/simple-gibbs-example-in-julia.html … The Julia team is doing amazing work!
-
-
@hadleywickham@johnmyleswhite ... but average dynamic language implementors would not have the skill to make that design performant :) -
@avibryant@hadleywickham If you want to understand how Julia gets its speed, consider this toy example: https://gist.github.com/johnmyleswhite/6195659 … -
@johnmyleswhite I don't think that's a great example b/c depends on knowing julia's name binding rules -
@hadleywickham In what sense? -
@johnmyleswhite because type of x can't change within function but can at global level? -
@hadleywickham In principle, R could use the same tricks, but my memory is that R has many functions whose type output is hard to predict. -
@johnmyleswhite yes, way too many. bad for both performance and understanding -
@hadleywickham It would be interesting to see how many functions need to be redefined to make a variant of R that's type-stable. - 3 more replies
New conversation -
-
-
@avibryant so it's fast, but not surprisingly fast. makes sense! -
@hadleywickham@avibryant A big part is that they don't have Rinternals.h specifying the memory layout of all objs. Python has same problem -
@perisaccadic
@hadleywickham@avibryant Fast Julia code is (per their explanation) devectorized, aka optimized in abstraction busting ways.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.