[Haskell-cafe] Performance: MD5

Bulat Ziganshin bulat.ziganshin at gmail.com
Sun May 18 05:13:41 EDT 2008


Hello Neil,

Sunday, May 18, 2008, 12:54:30 PM, you wrote:

> One day, I hope that high-level Haskell will outperform C all the
> time. It's already true with some of the ByteString stuff, but

i don't think so. all the tests where Haskell "outperform" C i've seen
was just unfair - for example they compare results of
MANUALLY-unrolled Haskell loops with rolled C ones - totally ignoring
the fact that gcc can unroll loops just by enabling option while with
ghc this can be done only by hand. many tests that show "equal
performance" just limited by memory bandwidth. and so on

it's hard to write program that outperforms gcc-generated code even
using assembler, and ghc is limited to this fact, too :)


-- 
Best regards,
 Bulat                            mailto:Bulat.Ziganshin at gmail.com



More information about the Haskell-Cafe mailing list