Link Details

Very interesting benchmark. One of the more interesting conclusions: Java is faster than C++...by a wide margin. Ruby, Groovy, Python and friends all get a showing as well. Worth the read.

Would be nice to explain why some voted this down.

Two reasons I voted down:

1. A "one-algorithm test" is hardly a way to compare language/compiler performance.
2. The end results don't jibe with my (semi-objective) reality that :

C++ > Java > (Python 2.5.1 ~ Groovy 1.6 ~ Ruby 1.9)

Meaning C++ is generally faster than Java, which is generally faster than the set of popular, dynamic languages.

Check out http://shootout.alioth.debian.org/gp4/index.php for a broader set of performance metrics than one piece of specific code.

alphadog,

I am just a little surprised - both your views are actually quite explicitly stated in the blog post (they are not even tucked into the fine print).

I believe there is indeed a case for one-algorithm tests to be blogged about.. something I am starting to suspect (though have no way to offer any additional supporting evidence or to reach a definitive conclusion)
a. In this case Java is faster than C++ because its garbage collector is outperforming the manual memory allocation in C++. Given that the code itself runs very fast in the two languages, the gc is suddenly important.
b. In this case Python is a little slower than ruby because its attribute access could be slower

Given the fact that I couldn't attach a high supporting evidence to these hypothesis, I chose to not very explicitly state it. But a post like this can still help people understand a little bit more within a specific context.

It is quite helpful when someone explains their votes. Thank you for your vote and comment.
Dhananjay

You may be surprised, because often what one intends and what another perceives out of a blog post are not exactly the same.

I realize you may have intended to ensure that the above two observations of mine are in your blog, but the result is different. Those two important observations are in a "note" as an "aside".

You then go on, armed with bullet-points and bold letters which strikes the reader as "important" :), to say some over-generalizations that can easily be misconstrued, especially since they are based on a very focused, and probably unoptimized, set of code.

I totally realize I'm being picky, but that's why I voted down.

Understand you much more clearly now. I'll take that as a useful learning.

nice work

Really appreciate the effort put into this.