search

open-mmlab / mmsegmentation

OpenMMLab Semantic Segmentation Toolbox and Benchmark.
https://mmsegmentation.readthedocs.io/en/main/
Apache License 2.0
8.26k stars 2.62k forks source link

Running time problem #3170

Open lxj1999 opened 1 year ago

lxj1999 commented 1 year ago

I use 74000 iterations(100 epochs) and 8 batch sizes, with a dataset of 5920 training images, which will use 24G for one GPU training, and the estimated running time is 17 hours. However, when I use 'bash tools/dist_train.sh MyConfigs/Upernet_Swin.py 2' to run on two GPUs, the estimated running time is even a little higher, up to 18 hours. Is it running for 200 epochs, or is there a running time error?

wujiang0156 commented 1 year ago

I have the same question, can anyone have? how to sovle?thank you