gnash-commit
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnash-commit] gnash ChangeLog backend/render_handler_agg.cpp


From: Bastiaan Jacques
Subject: Re: [Gnash-commit] gnash ChangeLog backend/render_handler_agg.cpp
Date: Tue, 6 Nov 2007 04:35:51 +0100 (CET)



On Tue, 6 Nov 2007, Bastiaan Jacques wrote:
From my own testing I conclude the same; the specialized clipping class
impacts on performance negatively. Interesting to note that this is true
even when testing movies with complex clipping regions.
So I think that the clipping patch should probably be reverted.


AGG docs on this class:

"Class renderer_mclip performs clipping to multi-rectangle clipping
areas, but its performance depends on the number of clipping rectangles."

"The cost of creation is minimal, it's just initializing of the class
member variables. But renderer_mclip allocates memory when you add new
clipping rectangles and deallocates it when destroying. It uses the
pod_deque class that allocates blocks of memory of equal size and never
reallocates it. When you reset clipping, the memory is not deallocated,
it's reused. The renderer_mclip deallocates memory only when destroying.
This technique is widely used in AGG and prevents from deep memory
fragmentation."

Bastiaan




reply via email to

[Prev in Thread] Current Thread [Next in Thread]