http://ompldr.org/vZ2R6cw/fgesrc_nov.zip
I wonder if anyone think what Im done is slow and it can be faster. The link above is source code for my game engine. It demonstrate how to use the shadow system. It took 100% CPU usage on my SiS Mirage 3, and maybe somebody ever create faster than this. I'd like to know is this the maximum a graphic card can do or I make the slowest code.
I used RadASM for the IDE. It is recommended you open it using RadASM.
deleted
I often use AMD's CodeAnalyst (http://developer.amd.com/tools/heterogeneous-computing/amd-codeanalyst-performance-analyzer/). The statistical approach of the timer based method works for all systems (also Intel). This methode point out critical instructions/ and memory accesses.
I put a precompiled binary on the source above. I'd like everyone try it if they had time. The most consuming part is the shadow part since I did not making the shadow based on the sillhoutte but from every vertices that can be a shadow. It slow I know, but that is the limit of my ability.
I can't see a performance problem (3% CPU usage). Probably you have bad/old graphic card.
BTW, there are two message box coming up at start, but I'm not aware of the used language ;-)
Quote from: qWord on December 01, 2012, 11:19:42 AM
I can't see a performance problem (3% CPU usage). Probably you have bad/old graphic card.
BTW, there are two message box coming up at start, but I'm not aware of the used language ;-)
Yeah, I had a bad card. And for the two message box, it was a warning for some of feature from OpenGL that not exist on your card. Thanks for your feedback. I cant afford to buy a new system. Maybe for a couple next years so I stick with it.