As I recall, these were microbenchmarks without the JIT in 2010 and 2011. I don't have that code anymore, so I'm going by memory. They were definitely hand-written PIR code, though.
If what I recall is true, then you can draw two conclusions. First, performance on a microbenchmark represents little more than performance on a microbenchmark. Second, features aren't free.
|Comment on Re^5: Perl 6 is going to get a lot faster in 2013|