The top half of the table is something I put together BEFORE I started my 4x4-downsampling based optimizations. For anyone not familiar with profiling and optimizing this is a pretty crucial step [wink]
On paper the optimization should have improved performance, but I needed to make sure of this. With proper statistical information about the program before I got started it meant that I could later go back and run a comparison and know exactly how much difference I made.
The bottom line in the table indicates red for an area thats worse afterwards and green for areas that have improved.
Curiously the parts I intended to improve appear to be SLOWER (the "star filter" and "cross filter" columns) but the overall frame rate has gone up by 7.
I'd like to drill into that in more detail to work out why it's appeared that way, but I don't have time. The fact that it's working faster is all that matters really [grin]
I found out today that Erik Reinhard is giving a guest lecture at my university on Friday morning. That is, one of the researchers behind the paper that all my HDRI work is based on. Gonna be interesting to see what he's got to say [grin]