两块Tesla P100训练yolov5时，batch size大小测试

两块Tesla P100训练yolov5时，batch size大小测试

 

#300 epochs completed in 4.716 hours.
 batch-size 2  epochs 300 yolov5m.yaml  device  0,1

#300 epochs completed in 2.839 hours.
 batch-size 4  epochs 300 yolov5m.yaml  device  0,1

#300 epochs completed in 2.279 hours.
 batch-size 16 epochs 300 yolov5m.yaml  device  0,1

#360 epochs completed in 2.413 hours
 batch-size 16  epochs 360  yolov5s.yaml  device  0,1

#RuntimeError: CUDA out of memory. Tried to allocate 16.00 MiB (GPU 0; 11.91 GiB total capacity; 11.07 GiB already allocated; 16.06 MiB free; 11.09 GiB reserved in total by PyTorch)
batch-size 32  epochs 360  yolov5x.yaml  device  0,1

 

 

 

 

########################