Re: [GEM-dev] performance? was: same patch different ...