Patch review for optimization in paintop

[Boudewijn Rempt]

On Wednesday 27 September 2006 14:24, Cyrille Berger wrote:
> Hi
>
> Can anyone have a look at this patch ? It optimized greatly paintop
> (compare drawing at zoom = 10% with the patch and without), I didn't
> noticed any side effect, but I don't want to introduce regression in 1.6.
> Now, computeDab is still the expensive function for paintop (with 40% of
> its cost coming from creating/destroying iterators :'( ).

Shouldn't the dab be cleared after use if you want to reuse the same paint 
device over and over again? In any case, 1.6 is already in rc mode, I would 
be hesistant to add optimizations. But, for 1.6, that's your call.

> PS: sometimes, I have discovered sysprof
> (http://www.google.com/search?q=sysprof&ie=UTF-8&oe=UTF-8) for profiling,
> it's similar than oprofile, a little bit less powerfull, but much more
> easier. And unlike cachegrind, there is nearly no speed decrease.
> I have been very successfull to find where to optimize with sysprof, I
> strongly advise you to test it :)

Thanks for the tip!

-- 
Boudewijn Rempt 
http://www.valdyas.org/fading/index.cgi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: not available
Url : http://mail.kde.org/pipermail/kimageshop/attachments/20060927/4b2e6d14/attachment.pgp