Home » forums » LtU Forum

The Language Wars are Dead; Long Live the Language Wars!

The [New] Great Computer Language Shootout


This has probably already made the rounds of everyone's favorite language-specific mailing lists, but I hadn't seen it show up on LtU. One of the Debian developers has taken it upon himself to re-implement the infamous "shootout", which hasn't been updated since 2001. Most included languages have had compiler/runtime/interpreter upgrades since the original, so there is definitely some new data lurking around in all those familiar old microbenchmarks.

By lennon at 2004-07-01 22:23 | LtU Forum | previous forum topic | next forum topic | other blogs | 5628 reads

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Scutigena

My friend Salvatore and I tried to launch a similar project, at http://scutigena.sf.net but I guess with these out, we don't have much hope of attracting much of a user/developer base.

By David N. Welton at Fri, 2004-07-02 13:02 | login or register to post comments

And now for something completely different?

Well you could probably attract a base if you were to do something different with it. I took a quick look at it, and so far it seems to be little more than a cut down version of Bagley's shootout at the moment (and frankly, who needs to benchmark Hello world more than once -- if that many times?).

You should try to find a different niche in the benchmarking arena and try to fill it. I know I would be more interested in seeing a benchmark/code repository of slightly larger problems -- e.g. doing a Burrows-Wheeler transform followed by a Huffman encoding on a large text file, computing something like a multi-resolution wavelet transform on an image, or an implementation of something like AES or SEAL (and I'm sure there are many others), than yet another implementation of Ackermann's function, reversing a file, or the eight queens problem.

I'd also be more interested in a collection of data structures with a series of tests that use them for benchmarking, e.g. implement a heap, and use it to do heapsort (instead of the standard in-place array version), or something more interesting. It's early, and my coffee hasn't kicked in yet, but I'm sure you get the idea.

-30-

By William D. Neumann at Sat, 2004-07-03 15:12 | login or register to post comments

Forum vs. Home page

Lennon, recall that you can post home page items instead of forum item, when you think something is important enough to warrnat it. Use the create content option, and click on "story".

By Ehud Lamm at Fri, 2004-07-02 13:27 | login or register to post comments

comparisons are tricky

I looked at just the C and Lisp versions of the prime number sieve, and it shows how tricky some comparisons can be.

For example, the C version appears to allocate the array in the compile phase, while it's done at runtime in the Lisp version. The Lisp is easily extensible, the C is not, and so on.

These benchmarks are neat, and represent lots of work, but may point out not so much that one language is "faster" than another, but that one algorithm doesn't translate well to the strengths of each language.

I'd like to see (in addition to previously mentioned comparisons of data structures like heaps) the best algorithm possible in each language, rather than what is substantially a compiler comparison.

By ericmoss at Mon, 2004-07-05 21:52 | login or register to post comments