Instead of bricks pointing to indices of material data, look into making brick point directly at material data. This will increase memory usage, but potentially make cache coherence better since we no longer read two buffers in the event of a voxel hit. This is not guaranteed to increase performance in any way and should therefore be tested because it is guaranteed to increase memory usage overall. It will also make material data more flexible since we are no longer limited to 255 unique materials.
Instead of bricks pointing to indices of material data, look into making brick point directly at material data. This will increase memory usage, but potentially make cache coherence better since we no longer read two buffers in the event of a voxel hit. This is not guaranteed to increase performance in any way and should therefore be tested because it is guaranteed to increase memory usage overall. It will also make material data more flexible since we are no longer limited to 255 unique materials.