[cairo] Survey of polygon rasterization techniques

[Zack Rusin]

On Wednesday 01 August 2007 06:45:55 am Kiia Kallio wrote:
> I have to disagree about the relevance of this test method if you want
> to test "raw rendering speed". Especially with graphics HW the whole
> system is built so that the application should keep as much of the
> data as possible on the HW side. 

That's exactly what I didn't want to happen in this test (not mentioning that 
none of the frameworks we have does this, but it's the general concept of 
caching that I wanted to make sure doesn't happen) because it'd be pretty 
funny to try to test tessellation and intersection detection by allowing 
frameworks to skip intersection detection.  But like I said, my tests are 
completely irrelevant for this discussion since they had a completely 
different goal.

> If the application re-feeds data that is 
> not really modified every frame, it's just doing things simply wrong and
> causing unnecessary overhead everywhere.

Unfortunately this is also a case we have to make work and work well. 

z