Re: [PD] abstraction penalty benchmarks