[Haskell-cafe] New Benchmark Under Review: Magic Squares

[Bulat Ziganshin]

Hello Malcolm,

Wednesday, July 5, 2006, 4:30:43 PM, you wrote:

> I note that your solution uses Arrays.  I have recently discovered that
> the standard array implementations in GHC introduce non-linear
> performance profiles (wrt to the size of the array).  None of the
> ordinary variations of arrays seemed to make any significant difference,
> but replacing Array with the new ByteString from fps brought my
> application's performance back down to the expected linear complexity.

are you give a chance to UArray? boxed arrays in ghc 6.4 may
sufficiently increase GC times


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