Humn... I agree with both of you, Albert and Tom. I started it from the beginning, using map and don't using reverse anymore. But the C program is still 7x faster than the Haskell one. Here is the code of the Haskell program:
main :: IO () main = do bmFile <- openFileEx "in.txt" (BinaryMode ReadMode) bmString <- hGetContents bmFile writeFile "out.txt" (map inc bmString) hClose bmFile
inc :: Char -> Char inc a = toEnum ((fromEnum a) + 1)
Well, in Haskell each character of the string takes 20 bytes: 12 bytes for the list cell, and 8 bytes for the character itself - the memory used by the character will be recovered at GC time though, as long as the character is < chr 256. The map operation also allocates a further 28 bytes per character: list cell + thunk(8) + character, assuming the inc operation is suitably optimised not to do any extra allocation. That's a total of 48 bytes per character. The C code, by comparison, doesn't do any dynamic allocation at all. To really match the C program, you need to use IOExts.hGetBuf and IOExts.hPutBuf, and do the operations on raw characters in memory. Using a UArray of Word8 would be better, but there aren't any operations to do IO to/from a UArray yet (actually I've written these, but they aren't in the tree yet). You should find that the IO library in GHC 5.02 is slightly faster than the one in 5.00.2. Anyway, I hope all this helps to explain why the Haskell version is so slow. Cheers, Simon