ImportError: DLL load failed while importing cd:

[lileitech]

module = importlib.util.module_from_spec(spec)

ImportError: DLL load failed while importing cd: The specified module could not be found.

After fixing many previous bugs, I get a new bug....it is quite annoying~~~is it due to the difference of pytorch/python/cuda version?

[otaheri]

Hi @Marie0909, sorry for any inconveniences. I have never faced the issues that you mention and tried the repo with many torch GPU versions, they work fine. You can quickly check the Colab notebook here. Please make sure you are using a torch version with cuda compilation. Another alternative would be to use the Pytorch3D package, which already has the chamfer distance package built-in.

I hope this helps.

[lileitech]

Hi @otaheri, thanks so much for your sweet reply.
I tried Pytorch3D package, which seems need to compile before using in win10, and also failed to work in my PC. My currrent environment is "python 3.8.8, pytorch 1.9.1, cuda 11.1".

Also, I tried to use pytorch version (without C++) (link: https://gist.github.com/WangZixuan/4c4cdf49ce9989175e94524afc946726), which is quite slow and requires large GPU memory. I will let you know if I could figure out this (fingers crossed).

Kind regards,
Lei

[lileitech]

BTW, I tried in another environment and the error dispear, but new bugs appear +_+

[otaheri]

Hi @Marie0909, can you post the new bugs here if you still didn't solve them? If you solve it, the bug and the solution would be very helpful for others :)

[lileitech]

Hi @Marie0909, can you post the new bugs here if you still didn't solve them? If you solve it, the bug and the solution would be very helpful for others :)

I finaly solve this issue by using it on linux system instead of windows :)

[doublexbepa]

Hi, I also encountered this problem. After I solved many of the previous bugs, I encountered this problem: ImportError: DLL load failed while importing cd: The specified module could not be found. I haven't found a solution yet. If there is If so, please let me know, thanks

[otaheri]

Please try the solution here if you still face the issue.