After hearing about all the hype surrounding node.js, I decided to test it's raw power by porting the code I wrote to solve the minimum enclosing circle (MEC) problem in Ruby a couple of years ago (Solution in Ruby to the MEC Problem). I also benchmarked the MEC code against various Ruby implementations.
Here are the results:
It is probably my crappy implementation, but for long running cases, the code really breaks down. It tracks the Ruby 1.9.1 performance through 1000 points or so, then falls off a cliff. I assume my implementation is creating too much garbage but haven't really had time to dig in to figure out why.
This benchmark is really garbage anyway for node.js -- heavily CPU intensive tasks are not what it was designed for. Most web applications involve blocking IO operations and the evented model of node.js will really shine in that environment.