[nivloc]

Benchmarks don't mean much until it's released, if ever. But I was playing around with it, and it's hard not to get excited about performance!

      |  mem | time |
  p1  | -24% |  -3% |
  p2  |  -8% | -12% |
  p3  | -31% | -12% |
  avg | -21% |  -9% |

(over 1.9.3)

Those are a couple workers I fired up for an hour. p2 relies on a C library, which might account for the smaller drop in memory usage. CPU needs investigation.

The numbers are kinda useless, but look good anyway.