devel버전을 사용...

https://github.com/google-research/Splat-SLAM/issues/7

cuda 11.7이 4090을 지원 안 한다고 하는데 4070super도 지원 안 하는 것인가? 아무튼 똑같은 에러가 발생해서
https://hub.docker.com/layers/nvidia/cuda/11.8.0-cudnn8-devel-ubuntu22.04/images/sha256-bd746eb3b9953805ebe644847a227e218b5da775f47007c69930569a75c9ad7d
으로 변경

apt-get update && apt-get -y install sudo

conda install nvidia/label/cuda-11.8.0::cuda-toolkit

Process Process-2:

Traceback (most recent call last):

File "/root/anaconda3/envs/splat-slam/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap

File "/root/anaconda3/envs/splat-slam/lib/python3.10/multiprocessing/process.py", line 108, in run

self._target(*self._args, **self._kwargs)

File "/home/Splat-SLAM/src/slam.py", line 97, in tracking

File "/home/Splat-SLAM/src/tracker.py", line 78, in run

self.online_ba.dense_ba(2)

File "/root/anaconda3/envs/splat-slam/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context

return func(*args, **kwargs)

File "/home/Splat-SLAM/thirdparty/glorie_slam/backend.py", line 76, in dense_ba

n_edges = self.ba(t_start, t_end, steps, graph, nms, radius,

File "/root/anaconda3/envs/splat-slam/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context

return func(*args, **kwargs)

File "/home/Splat-SLAM/thirdparty/glorie_slam/backend.py", line 51, in ba

graph.update_lowmem(

File "/root/anaconda3/envs/splat-slam/lib/python3.10/site-packages/torch/amp/autocast_mode.py", line 44, in decorate_autocast

return func(*args, **kwargs)

File "/root/anaconda3/envs/splat-slam/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context

return func(*args, **kwargs)

File "/home/Splat-SLAM/thirdparty/glorie_slam/factor_graph.py", line 322, in update_lowmem

self.video.ba(target, weight, damping, self.ii, self.jj, t0, t1,

File "/home/Splat-SLAM/thirdparty/glorie_slam/depth_video.py", line 306, in ba

success = self.dspo(target, weight, eta, ii, jj, t0, t1, iters, lm, ep, motion_only,opt_type)

File "/home/Splat-SLAM/thirdparty/glorie_slam/depth_video.py", line 280, in dspo

poses, disps, wqs = thirdparty.glorie_slam.geom.ba.BA_with_scale_shift(

File "/root/anaconda3/envs/splat-slam/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context

return func(*args, **kwargs)

File "/home/Splat-SLAM/thirdparty/glorie_slam/geom/ba.py", line 220, in BA_with_scale_shift

dwq, dz = schur_solve(H, E, C, u, w, ep, lm)

File "/home/Splat-SLAM/thirdparty/glorie_slam/geom/chol.py", line 88, in schur_solve

S = H - torch.matmul(E, Q*Et)

커뮤니티