[cairo-bugs] [Bug 43064] stroke performance is super-linear in number of points in the path

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=43064

--- Comment #7 from Chris Wilson <chris at chris-wilson.co.uk> 2011-11-18 12:55:15 PST ---
That particular benchmark is indeed close to one type of pathological case for
the scan-line rasteriser: a block of nearly vertical lines. There is a
micro-benchmark for that in the shape of vlines. The obvious optimisation in
this case is to render using a vertical sweep line, which can be easily done in
the application by rotating the lines by 90 degrees rendering into a temporary
buffer, and then rotating back through 90 degrees on compositing onto the
output.

I've considered doing so automatically if sum_dx << sum_dy, but I've seen no
real world cases where there was not a better alternative method of rendering
or at a last resort could optimise its own rendering far better than cairo
could.

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA Contact for the bug.