[MENTION=887]ioFlow Studios[/MENTION] For the above frost scene, it takes the following time
Multi-Threaded:
With collision: 7.0 sec
Without collision: 5.1 sec
Single-Threaded
With collision: 12.8 sec
Without collision: 6.0 sec
This is for calculating 16x16x2 patches of voxels data, each having 32x32x32 cells within a patch. Each patch is sent to a thread to be calculated
The bottleneck now seems to be in the rendering (once the threads calculate the voxel data and all the vertex/index data) it queues it in the render thread to actually create the meshes. I didn’t get time to profile this further. I’m sure there’s more room for improvement
