It means all vertices and textures have this size, there is no real reason of loading more to get faster…
Nevertheless, in fact there is. Here is my trick :
Something not optimized is the fact that the render time it cut between :
- CPU work (building BVH hierarchy, etc…)
- HARD DRIVE work (loading all textures from disk to GPU)
- GPU work (actual raytracing process)
On some big scene, first two steps can be as long as the third one, or even longer, and GPU compute power is pretty much “lost”. (inactive 50% to 60% of the time)
When you run 2 Blender instances rendering at same time, you don’t go “faster” because your single GPU is used by both Blender Instances at same time.
There is however a benefit in launching multiple Blender Instances.
Let say you want to render an animation.
You can set a frame step to 5 on your
.blend and :
- Launch the 1st Blender starting on frame 1
- Launch the 2nd Blender starting on frame 2
- Launch the 3rd Blender starting on frame 3
- Launch the 4th Blender starting on frame 4
- Launch the 5th Blender starting on frame 5
Basically what’s happening : 5 Blender Instances working at same time. “Theorically” Total time to render ALL frames would be the same. But in fact, all 5 blender instances will randomly shift between each other in the time (because of kinda random time to render a frame). The result is that, at pretty much any time, 1 instance will be building BVH, while another is loading textures, while again another will be raytracing, etc…
The result is that the GPU will work 100% full time, CPU 100% full time, and Hard Drive 100% full time. And the final total time can actually be cut by 2 or 3
See you ++