9.1 Proposal: Top five performance problems
erik at laptop.org
Fri Oct 24 15:50:51 EDT 2008
On Fri, Oct 24, 2008 at 01:56:47AM +0200, Marco Pesenti Gritti wrote:
> Problems and ideas in no particular order.
> * Sugar shell startup is too slow.
> Reduce dependencies, single process shell, modularize and delay
> initialization of components which are not immediately necessary,
> measure constantly to avoid regressions and monitor progress.
> * Icons rendering is slow and uses too much memory.
> Cache svg icon on disk pre-rendered and mmap them, render colored
> icons using a mask per color, user server side pixmaps to speed up
> rendering of some of the icons.
What about changing the kind of visual feedback we give. Instead of
pulsing icons what about icons with a string of dots beneath, a progress
bar, flashing, or another kind of overlay feedback which requires fewer
visual changes (frames) and/or could be overlaid on top of existing
icons without calculating a new animation for every icon?
> * The frame is too slow to appear and disappear.
> Experiment with pixmap caching. Is an empty frame fast enough?
> Composite the active window and the frame.
What about just turning on X Composite for everything? From the user's
perspective this wholly resolves a lot of issues associated with
rendering performance. From a programmer-resources perspective it's
also quite straightforward.
The problem is that we have fallout in terms of memory allocation, at a
cost of about 2mb per activity window. That said, switching between
activities is *instantaneous*. More problematically I see no clever
algorithm to figure out where a user is going to go in the activity
stack next, which might help us choose which windows to unmap.
Could we discuss this during the talk?
More information about the Devel