[cairo] cairo recording surface purpose

[Baldeva, Arpit]

Hi,



We use cairo with pixman backend on our port of WebKit. We have been toying with the idea of using cairo recording surface to optimize performance. The basic idea is to use the recording surface in a frame and then fork the actual render operation on another surface in a different thread on a different CPU core. In our initial implementation, it seems like recording itself is a very expensive operation and did not yield much performance benefits. So my questions are



1.       Is it known that recording is actually fairly expensive operation so it is not meant to optimize? We went with this idea as our target hardware have lot of cores but each core has less power. So recording on one core and replaying on another seemed like a good idea.

2.       We have so far not seen any issues but is this thing considered safe in general (multi-threaded access)? Is there anything we should keep in mind?



Appreciate any answers.



--Arpit
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cairographics.org/archives/cairo/attachments/20140519/33cf7c38/attachment.html>