when installed all, I run python3 demo.py in terminal, nms/_ext/nms/_nms.so: undefined symbol: __cudaPopCallConfiguration
QianLingjun opened this issue · 4 comments
qlj@zyyl-PowerEdge-T630:~/projects/pytorch-mask-rcnn$ python3 demo.py
Traceback (most recent call last):
File "demo.py", line 10, in
import coco
File "/home/qlj/projects/pytorch-mask-rcnn/coco.py", line 50, in
import model as modellib
File "/home/qlj/projects/pytorch-mask-rcnn/model.py", line 26, in
from nms.nms_wrapper import nms
File "/home/qlj/projects/pytorch-mask-rcnn/nms/nms_wrapper.py", line 11, in
from nms.pth_nms import pth_nms
File "/home/qlj/projects/pytorch-mask-rcnn/nms/pth_nms.py", line 2, in
from ._ext import nms
File "/home/qlj/projects/pytorch-mask-rcnn/nms/_ext/nms/init.py", line 3, in
from ._nms import lib as _lib, ffi as _ffi
ImportError: /home/qlj/projects/pytorch-mask-rcnn/nms/_ext/nms/_nms.so: undefined symbol: __cudaPopCallConfiguration
what happen and how can address my problem.Thinks
请问_ext是怎么得到的
you can try a docker with cuda9.0 installed, for me, nvidia/cuda:9.0-cudnn7-devel works.
you can install torch0.4.0+cuda9.0, it works well on my machine, but you should be careful that your machine must have enough GPU memory.
Can I fix this by recompiling PyTorch？