Re: [pypy-dev] MalGen as a benchmark?