Instant-NGP speed/memory

[wangjksjtu]

Thanks for the amazing projects!

When I run the instant-ngp via xrnerf, the memory is keep growing and will get OOM error. Also, seems the training is very slow (I checked the GPU utilization rate and it is around 10%). Did I mess up something?

My command:

python run_nerf.py --config configs/instant_ngp/nerf_blender_local01.py --dataname lego

Thanks!