aanet立体匹配安装中的坑
GPU:3060
第一次尝试
基础环境：python3.7.11 cuda10.0 pytorch1.2
在运行脚本./scripts/aanet_predict.sh
会出现以下问题：
1 卡死半天。
2 半天之后出现以下文字：
RuntimeError: CUDA error: no kernel image is available for execution on the device
第二次尝试
基础环境：python3.7.11 cuda10.0 pytorch1.4
在运行脚本./scripts/aanet_predict.sh
会出现以下问题：
_C.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe26detail36_typeMetaDa
第三次尝试
基础环境：python3.7.11 cuda10.1 pytorch1.4
在运行脚本./scripts/aanet_predict.sh
会出现以下问题：
ValueError: Unknown CUDA arch (8.6) or GPU not supported
参考https://blog.csdn.net/ng323/article/details/116940299
第四次尝试
基础环境：python3.7.11 cuda10.2 pytorch1.5
官网好想没有支持10.2的1.4 ,1.2 torch
在运行脚本bash build.sh
会出现以下问题：
error: command 'gcc' failed with exit status 1
第五次尝试
基础环境：python3.7.11 cuda11.0 pytorch1.7
官网好想没有支持10.2的1.4 ,1.2 torch
在运行脚本bash build.sh
会出现以下问题：
error: command 'gcc' failed with exit status 1
第六次尝试
基础环境：python3.6.9 cuda11.0 pytorch1.4
在运行脚本bash build.sh
会出现以下问题：
ValueError: Unknown CUDA arch (8.6) or GPU not supported
第七次尝试
root@2b8029e14ffa:/home/aiya/work/aanet-master# python3
Python 3.6.9 (default, Oct 8 2020, 12:12:24)
[GCC 8.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.version.cuda
'10.0.130'
>>> torch.__version__
'1.2.0'
>>>
root@2b8029e14ffa:/home/aiya/work/aanet-master# ./scripts/aanet_predict.sh
=> Loading pretrained AANet: pretrained/aanet_kitti15-fb2a0d23.pth
=> 1 samples found in the data dir
=> Inferencing 0/1
Traceback (most recent call last):
File "predict.py", line 199, in <module>
main()
File "predict.py", line 162, in main
pred_disp = aanet(left, right)[-1] # [B, H, W]
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 547, in __call__
result = self.forward(*input, **kwargs)
File "/home/aiya/work/aanet-master/nets/aanet.py", line 206, in forward
left_feature = self.feature_extraction(left_img)
File "/home/aiya/work/aanet-master/nets/aanet.py", line 133, in feature_extraction
feature = self.feature_extractor(img)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 547, in __call__
result = self.forward(*input, **kwargs)
File "/home/aiya/work/aanet-master/nets/resnet.py", line 178, in forward
x = self.conv1(x)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 547, in __call__
result = self.forward(*input, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/container.py", line 92, in forward
input = module(input)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 547, in __call__
result = self.forward(*input, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/activation.py", line 94, in forward
return F.relu(input, inplace=self.inplace)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/functional.py", line 911, in relu
result = torch.relu_(input)
RuntimeError: CUDA error: no kernel image is available for execution on the device