nerfstudio-project / nerfacc

A General NeRF Acceleration Toolbox in PyTorch.
https://www.nerfacc.com/
Other
1.37k stars 113 forks source link

about the gpu memory #239

Open WBS111 opened 1 year ago

WBS111 commented 1 year ago

When I was debugging the code, I found a problem: by modifying target_sample_batch_size, init_batch_size, render_step_size does not change the GPU memory footprint, only by modifying the network size, such as lowering net_depth, net_width will the GPU memory footprint be reduced. Why does lowering the batch size not lead to a reduction in GPU memory usage?