Re: tree walker evaluator

[David Bateman]

John W. Eaton wrote:
> On  6-Jan-2009, John W. Eaton wrote:
>
> |   * I haven't tried to do any timings, so I don't know how this change
> |     affects performance compared to the current eval methods in the
> |     tree classes.
>
> Here are some timings.  I ran make check on my system with the old
> evaluator:
>
>   77.56user 18.73system 1:37.80elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k
>   29074inputs+5928outputs (112major+5161147minor)pagefaults 0swaps
>
> and with the new evaluator (using the patch I posted yesterday, no
> other changes):
>
>   75.61user 18.78system 1:37.86elapsed 96%CPU (0avgtext+0avgdata 0maxresident)k
>   63288inputs+5896outputs (345major+5137036minor)pagefaults 0swaps
>
> So if anything, there is a minor improvement, and perhaps we can even
> do better than this with a few more changes.
>
> Does anyone object to me applying this set of changes?
>
> jwe
>
>   
Its the way we should go, so I'd say go ahead. I'd say next we should 
probably add an instrumented version first to allow the implementation 
of the profiler. Do the rvalue methods need to become part of the tree 
walker class first to allow that?

Regards
David

--
David Bateman                                address@hidden
35 rue Gambetta                              +33 1 46 04 02 18 (Home)
92100 Boulogne-Billancourt FRANCE            +33 6 72 01 06 33 (Mob)