YilingQiao
Segmentation fault (core dumped)
ubuntu20.04, nvidia a6000, cuda11.3
python ./scripts/exp_garden_ball.py --mode nerf --load_snapshot ./extra_data/nerf360/garden/35000.msgpack --width 800 --height 600 --gui
the output is :
config_path: ./scripts/exp/garden_ball.json 1200 800 10 read_obj_file!! shapes[s].mesh.num_face_vertices.size() 2 1 2 0 1 3 2 vec_obj_list size 2 size_obj_from_json 1 size_lightsrc_list 1 size_shadow_list 2 00:43:05 WARNING Vulkan: loader_icd_scan: Can not find 'ICD' object in ICD JSON file /usr/share/vulkan/icd.d/nvidia_layers.json. Skipping ICD JSON 00:43:05 WARNING Vulkan: Device Extension VK_NVX_binary_import is not supported by this layer. Using this extension may adversely affect validation results and/or produce undefined behavior. 00:43:05 SUCCESS Initialized Vulkan and NGX on device #0: NVIDIA RTX A6000 Loading snapshot ./extra_data/nerf360/garden/35000.msgpack 00:43:06 INFO Loading network config from: ./extra_data/nerf360/garden/35000.msgpack 00:43:07 INFO GridEncoding: Nmin=16 b=1.66248 F=2 T=2^19 L=16 00:43:07 INFO Density model: 3--[HashGrid]-->32--[FullyFusedMLP(neurons=64,layers=3)]-->1 00:43:07 INFO Color model: 3--[Composite]-->16+16--[FullyFusedMLP(neurons=64,layers=4)]-->3 00:43:07 INFO total_encoding_params=13623184 total_network_params=10240 Segmentation fault (core dumped)
also:
- without --gui, the output is the same
- i can build instant-ngp and run demo with both master and c4d622e branch
Are you using conda? I had the same issue. I disabled conda and made a venv instead, then rebuild and reinstall requirements.txt. After that it worked fine for me.
I ran into this issue on another computer where I used venv instead of conda already. It seems like in this case downgrading numpy solved the issue: pip install numpy==1.26.4
I ran into this issue on another computer where I used venv instead of conda already. It seems like in this case downgrading numpy solved the issue:
pip install numpy==1.26.4
It works. thanks