Training Memory Consumption

pkuCactus / BDCN

The code for the CVPR2019 paper Bi-Directional Cascade Network for Perceptual Edge Detection

MIT License

341 stars 71 forks source link

Training Memory Consumption #20

Closed huberthomas closed 5 years ago

huberthomas commented 5 years ago

Is it possible to train with a Nvidia GTX 1080 Ti with 11GB GDDR RAM? I always run out of memory but installed your recommended software via conda. Without CUDA support my system has peaks around 20GB RAM and I don't know if this is also too much. If you use your graphic card for learning can you tell me your configuration?

My settings: Ubuntu 16.04 LTS, Nvidia Driver: 418.67, Cuda: 9.2

Big thanks in advance!

jannctu commented 5 years ago

hi @huberthomas

I run on the similar environment as yours and it's work fine.

Ubuntu 16.04.5 LTS Nvidia GTX 1080 Ti with 11GB Memory NVIDIA Driver 410.48
CUDA 9.0

huberthomas commented 5 years ago

Thank you very much, I played around with my configuration and found a setting that is working:

Ubuntu 16.04 LTS, Nvidia: 418.67, cuda: 9.2, cudnn: 7.6 conda install ninja pyyaml mkl mkl-include setuptools cmake cffi typing torchvision scipy numpy pillow opencv matplotlib pytorch=0.4.1 cuda92 python=2.7 -c pytorch