RandLA-Net icon indicating copy to clipboard operation
RandLA-Net copied to clipboard

How to speed up the training?

Open ZouhairBALLOUCH opened this issue 4 years ago • 3 comments
trafficstars

Hi ; I just want to know if there is a possibility to speed up the training because for me each epoch requires almost 4 hours of calculation Thanks in advance

ZouhairBALLOUCH avatar Aug 12 '21 12:08 ZouhairBALLOUCH

Are you using GPU/CPU?. With Semantic KITTI on CPU it took around 4 hours per epoch but in GPU it was 30 min. check whether the code is using GPU or not.

loki-veera avatar Oct 19 '21 09:10 loki-veera

I am unable to mount the process to GPU. Can you suggest to me the necessary changes so that I can mount it on the GPU??

ARITRA2296 avatar May 16 '22 06:05 ARITRA2296

@ARITRA2296 Did you install tensorflow or tensorflow-gpu. tensorflow-gpu should pick the GPU automatically. You can use something like this is https://stackoverflow.com/questions/38559755/how-to-get-current-available-gpus-in-tensorflow. To check your GPUs in the code.

loki-veera avatar May 16 '22 08:05 loki-veera