jack - 2014-01-07

Ok, sounds good. I was thinking of implementing something in CUDA, with the added bonus of passing some of the workload from the cpu to the gpu.

BTW, I had tried updating only some of the boids in each Update call (meaning I place the boids into groups and take turns updating the velocity of one group per Update call, leaving the rest of the boids with their existing velocity). I had read that in a slide of yours (I forget where...). It certainly improved performance but the boid's behaviour became "jittery". I'd probably would have to tweak my steering code's parameters to fix that.

Thanks for the reply!