This is a project to create a benchmark suite along the lines of SPEC or Decapo, but oriented toward dynamic language systems. It will include some simple CPU benchmarks (including some from are-we-fast-yet, but also include garbage-collection and memory management benchmarks.
The ultimate goal is to allow people to say, "Our Smalltalk implementation gets a ZagMark of 7.4, but we're trying to get to CPython's Zagmark of 12.3".