[Haskell-cafe] Performance Issue

[Daniel Fischer]

On Saturday 22 May 2010 15:00:25, Thomas Schilling wrote:
> Actually, in this case it would be safe to do CSS.  Because
>
>  a) the function is strict in both arguments so GHC creates a worker
> which only uses unboxed types
>  b) this cannot cause any space leaks (it contains no pointers)
>
> The generated Core does look pretty weird, though:
>
> $wnewton =
>   \ (ww_s115 :: Double#) (ww1_s119 :: Int#) ->
>     case ww1_s119 of ds_Xr8 {
>       __DEFAULT ->
>         case ^_r11D
>                (case $wnewton ww_s115 (-# ds_Xr8 1)
>                 of ww2_s11d { __DEFAULT ->
>                 D# ww2_s11d          -- box the result of $wnewton
>                 })
>                lvl_r11B
>         of _ { D# x_avk ->            -- unbox it again

The boxing is due to the use of (^).
If you write x*x instead of x^2, it can use the primop *## and needn't box 
it.
As a side effect, the original time leak probably wouldn't have occured 
with x*x instead of x^2 because one would've made it
   let x = newton a (n-1) in (x*x +a) / (2*x)
instead of writing out newton a (n-1) thrice anyway, wouldn't one?

>         case $wnewton ww_s115 (-# ds_Xr8 1)
>         of ww2_s11d { __DEFAULT ->
>         /##
>           (+## x_avk ww_s115) (*## 2.0 ww2_s11d)
>         }
>         };
>       0 -> ww_s115
>     }
>