python setup.py build develop
error: command '/usr/local/cuda/bin/nvcc' failed with exit code 1
nvcc fatal : Unsupported gpu architecture 'compute_86'
error: command '/usr/local/cuda/bin/nvcc' failed with exit code 1
环境:rtx3090,cuda11,torch1.7,python3.7
尝试换gcc版本,从7.5到9,没用
降低pytorch版本,没用
指定python加sudo,没用
这个也没用https://www.cnblogs.com/llfctt/p/13967898.html
1.A4000换torch1.4, 1.5,1.6 cu101没用,导致ValueError: Unknown CUDA arch (8.6) or GPU not supported,换torch1.8cu113没用
2.RTX3090 torch1.7cu110,导致
nvcc fatal : Unsupported gpu architecture 'compute_86'
error: command '/usr/local/cuda/bin/nvcc' failed with exit code 1
torch1.8.1cu111出现error: command '/usr/local/cuda/bin/nvcc' failed with exit code 1错误
torch1.8.1cu113解决
已解决
原因:显卡算力高于torch算力,如果cuda11.1最后配置cu113的
命令:conda install pytorch==1.8.1 torchvision==0.9.1 torchaudio==0.8.1 cudatoolkit=11.3 -c pytorch -c conda-forge
先找到python位置
whereis python#当前环境使用python版本
sudo python位置 setup.py build develop
环境:
cuda11.0 torch1.8.1 python3.8.2 RTX3090
1