Have you been following Tim Bray’s Wide Finder project? It’s an exercise to express a fairly simple task in a manner that will scale across multiple CPU cores. Some of Tim’s initial progress with Erlang, and other contributors in different languages, is quite fascinating.
Like Tim I was also amused at Pete Kirkham’s C++ implementation which was purported to be shorter than an initial attempt by an Erlang expert (in Erlang obviously). However on closer examination it seems that Pete’s C++ implementation was simply handling the I/O portion with simplified parsing and not the subsequent sorting.
So as another C++ coding Kata I decided to have a go. Whereas Tim’s goal was to evaluate different methods of expressing algoritms for parallel computation, mine was a lot more modest: just get it running concisely in C++ and compare performance with the raw Ruby version. Here’s what I came up with.
