PyCUDA ERROR: The context stack was not empty upon module cleanup

[JI0726]

After running ‘bash dev_scripts/w_scannet_etf/scane241_test.sh’ and 'bash dev_scripts/w_n360/chair_test.sh' I got this PyCUDA error:

Does anybody know how to solve it?

[zhoupeizhu]

same problem

[taekkii]

I have a same problem, too.

[hdzmtsssw]

same problem

[zhangchuanyi96]

similar problem, need help

[Ailon-Island]

same problem

[phillipinseoul]

same problem, has anyone found a solution?

[OvOtoQAQ]

same...

[OvOtoQAQ]

https://blog.csdn.net/Andrew_megemeta/article/details/115712499
hey guys, I find a blog which seems like be able to solve this problem, I tried a lot but still can't figure out where to pop, if anyone solve this one, please inform me.

[MobiusLqm]

Same problem!Does anyone know how to fix it??

[yjcaimeow]

try import pycuda.autoinit

[knightwzh]

Same problem

[Xharlie]

hi folks, it is wrong decision to build things in pycuda, and im sorry that i don't have enough time to rebuild everything in a fashion of pure cuda/pytorch.

[juliuskuehn1]

Hi, I have the same problem. Further I also encountered this when I was running the scripts for:
Test feed forward inference on dtu scenes like 'bash dev_scripts/dtu_test_inf/inftest_scan1.sh'. I was able to fix this by adding model.cleanup() and del model into the train_ft.py before exit() is called.
Unfortunately when trying to apply this to the test_ft.py script, I run into a Segmentation fault instead of the PyCuda ERROR...

[HanLingsgjk]

Hello, I would like to ask which parts are using pycuda. Perhaps I can rewrite these parts as cuda

[CTouch]

#82 (comment)
Try this method.

[lhp121]

我测试代码的时候把下面的代码加在test_ft_.py最后的main函数里面了，import写在这里，不在前面的头文件那，解决了！

if __name__ == '__main__':
    # 清除 CUDA context 栈
    import pycuda.driver as cuda
    import pycuda.autoinit
 
    cuda.Context.pop()
    main()