search

irfanICMLL / structure_knowledge_distillation

The official code for the paper 'Structured Knowledge Distillation for Semantic Segmentation'. (CVPR 2019 ORAL) and extension to other tasks.
BSD 2-Clause "Simplified" License
708 stars 103 forks source link

__ext.so error: undefined symbol: __cudaRegisterFatBinaryEnd #61

Open RonDen opened 3 years ago

RonDen commented 3 years ago

Hi, I have encountered the same problems mentioned in this issue. Can you give me some detail on how to solve this?

I have tried to create a new conda evironment with python=3.5, torch==0.4, and successfully build the bn.o and ext.so files. But when I use command python test, I got the same error. I tried to delete the ext.so file and rebuild, but the same error again and again.

What you mean about the multiple environments, and how to clean the environment?

aPlutoBoy commented 2 years ago

Hi, I have faced the same error and I finally find that this is mainly because the version of CUDA and your torch are not match. As for me, using torch==0.4.1 and cuda = 9.0 is fine.