Training efficiency

[Linfeng Tang]

Hi

Thanks for the great work again.

When training SWIR with the KAIR toolbox, I found that the CPU utilization was particularly high, while the GPU was always idle. And the training is particularly inefficient. I wonder if the author would be so kind as to tell me the GPU and CPU configurations used, and the training time?

[Jingyun Liang]

Do you use DstributedDataParallel (DDP)?

[Linfeng Tang]

Thank you very much for your reply. Actually, I did not use DstributedDataParallel (DDP). And, I ran: "python main_train_psnr.py --opt options/swinir/train_swinir_sr_lightweight.json".

[Jingyun Liang]

This is the reason. DDP is much quicker than DP.