
I have optimized the code already, and cann''t figure where i could optimize now :
- I use vertex buffer in place of "direct mode"
- I don''t create new particles, instead i create a particle array at the initialisation and then reuse dead particles (so no new/delete)
- i have a particle manager class that does all the redundant stuffs that each system needs (calculating the modelview matrix invert for example) so those stuffs are calculated only once instead of one per system.
But obviously there''s still something wrong
