mikeash.com pyblog/friday-qa-2009-02-27-holistic-optimization.html comments

PatMcGee - 2009-09-20 00:54:31

Sun, 20 Sep 2009 00:54:31 GMT

Relevant book: "Discussion of THE Method", Billy Vaughn Koen.

The Engineering Method: The strategy for causing the best change in a poorly understood situation within the available resources.

I.e., if you understand the situation, you're doing technician work, not engineering.

I found it fascinating. Among other things, it's a collection of heuristics applicable to just about any engineering problem. Not to mention the discussion in which he claims that just about everything is engineering.

Pat

Phil Holland - 2009-03-07 03:45:30

Sat, 07 Mar 2009 03:45:30 GMT

A little late for your next Q&A, but at some point, I'd be interested in entries on the proper uses of libffi, Accelerate.framework, or the llvm.

mikeash - 2009-02-27 22:00:30

Fri, 27 Feb 2009 22:00:30 GMT

Yep, great point. For a game, "fun" is basically the only thing to optimize. It implies other optimizations (an action game that gets 5fps is not much fun!) but in the end that's all that matters. Maximum fun for the programmer time available.

millenomi - 2009-02-27 15:59:20

Fri, 27 Feb 2009 15:59:20 GMT

Let's not forget that you can optimize for things even less traditional than performance. Valve is famous for their build-run-playtest cycle they use to literally optimize for maximum fun (with Newell even advocating advanced forms of "profiling" such as biometrics taken from playtesters during play -- !!!).