Thanks for the reply. I think that you are right that laziness and too much abstraction are the main reasons for low performance.
I've actually received some patches from a Mr. Lemmih, who converted some of the code to use the ST monad, and managed to get results with very good performance. I am confident that with work, it should be possible to complete the omegagb project and have a working game boy emulator written in pure haskell. On the other hand, I would like to mention that I have reimplemented most of the omegagb code in C, just for fun. Obviously it runs much faster then the haskell version, by a factor of about 20. More interestingly though, is that the C version isn't longer, it's not more verbose, and it's not any harder to read then the haskell version. It doesn't have any of the monad goodness of the haskell version, and it would probably be harder to proove things about the code, but there is something to be said for lean and mean C code. The C language is a very good fit for this problem domain. If I had to do a real world project for a similar problem, I'm not sure that haskell would have that much of an advantage over C, even if performance wasn't an issue. Then again, I imagine that If I were to try to attempt to make an emulator for a computer more complicated then the game boy, then a hybrid C/Haskell approach might be very attractive. I'm putting OmegaGB on the back burner for now and am moving on to other projects. Peace, Bit _______________________________________________ Haskell mailing list Haskell@haskell.org http://www.haskell.org/mailman/listinfo/haskell
