Can this project work with CUDA 11 and Pytorch 1.9.0?

facebookresearch / maskrcnn-benchmark

Fast, modular reference implementation of Instance Segmentation and Object Detection algorithms in PyTorch.

MIT License

9.31k stars 2.49k forks source link

Can this project work with CUDA 11 and Pytorch 1.9.0? #1322

Open ghfdjsk opened 3 years ago

ghfdjsk commented 3 years ago

❓ Questions and Help

Currently my machine has Cuda 11 and Pytorch 1.9.0. Due to the dependencies of other projects I am working on I can not change the version of these. However, I want to run this project. Is this project compatible with my system's version of Cuda and Pytorch? If so, then what changes should I make to run it? If not, how do I resolve this issue?

Dawn-LX commented 3 years ago

Pytorch CUDA error: no kernel image is available for execution on the device

qijithu commented 2 years ago

same question

csguanhu commented 2 years ago

mine env: cuda11.6+pytorch1.10.2，it works quite well... find those question and fix them

litingfeng commented 2 years ago

@csguanhu Could you please share your gcc version as well? I've encountered this error error: command '/usr/bin/g++' failed with exit code 1 when using pytorch-nightly 1.12, cuda 11.3, gcc 9.

Thanks in advance.

Greatuso commented 2 years ago

mine env: cuda11.6+pytorch1.10.2，it works quite well... find those question and fix them

Could you tell me how to use 'apex'?When I install apex,it always told me RUNtimerror,and I use Windows. Thank you.