Profiling Cycles?


With relatively big scene I would like to understand what’s eating the memory and where the bottlenecks are. Is it possible to somehow enable logging of that data as I assume it might have been needed for development.

Thank you!

I don’t know if there really is a way to do this.

Generally, the things that eat the most memory are high numbers of hair particles and high resolution displacements while the biggest performance eaters tend to be deformation blur, hair primitives set to use smooth curves rather than segments, and heterogeneous volumes.

I too would be very interested in a profiling option for Cycles (memory, CPU, etc.) I have a scene in which Cycles GPU rendering is 1/10 the speed of CPU rendering on a system where it is typically anywhere from on par with up to 3x the performance of CPU rendering. Trial and error is a very inefficient way to understand situations like this… if higher level profiling (i.e. higher level than something like gdb) doesn’t currently exist, does anyone know it is on the radar for a future release?