Thanks, that is the kind of thing I was looking for (though a bit of a narrow test). Cool to hear that this gets a solid speedup from parallelism.