Thanks, yes the difference in both machines is clear, but there still persists the "warming up" which I also thought it could be caused by the JIT, but if that were to be true, why aren't compiled objects reacting in the same way?
Compiled:
Non-Compiled:
Machine Specs:
Core: i7-4510U @ 2.00GHz
Ram: 12 GB
OS: Win 8.1 64x
GPU: GeForce 840M, Graphical Clock: 1029 MHz
Java: 7, 64x
its clear that in both machines (even with different specs) we still have the warm up, but although its just a short initial period, it will happen everytime a different object comes into play (rendered) which is a bit weird, because when using compiled objects such behaviour is nonexistent.
I thought this could besome kind of caching on the application side once we are using computer memoty to store the objects to be sent to the pipeline every frame, so that could lead to the warm up, but could it be optimized on the jPCT framework? Or is it really java's only fault here?
Anyway, I tried the VM params (-server -Xcompile) you used and "Config.glVertexArrays=false;", and it got a little better:
but on bigger scenes with a lot more dynamic stuff happening, I still expierience the warm up everytime a new object comes into play.
Anyway, even if this "issue" got somehow better, there still be performance issues when expanding the scene, as we can see on the "non-compiled" case. Thereforce I'll bet on the compiled scene, so I gotta hit the road on learning GLSL so I can use more dynamic lighting and stuff.
thanks a lot for your time on this matter.
