A while ago, I discussed a bit about code optimisation using valgrind: Spot bottlenecks. At the end I obtained something like 1sec for about 4000 reflections. Meaning about 250-500us/refl. In the meantime, I needed a much faster version. I am … Continue reading →