search

eriklindernoren / PyTorch-YOLOv3

Minimal PyTorch implementation of YOLOv3
GNU General Public License v3.0
7.31k stars 2.63k forks source link

is woring in 3090,lower fps! #837

Open 827379852 opened 1 year ago

827379852 commented 1 year ago

inference video in RTX 3090,max fps is 16, and my computer GPU is 1050ti, max fps is 10 what should i do?

827379852 commented 1 year ago

both cpu and gpu are in normal state

Flova commented 1 year ago

Hard to say where the bottleneck is. I get significantly more fps on my 4090, 2080ti and 1080. Can you please provide the info required in the issue template? You can also attach screenshots of htop and nvtop to show the system utilization. The bottleneck can also be an io one, so where are the images stored and what's the utilization of this device. Also what images are used? Can you try to run it on coco or the example images from this repo? What's your batch size? Do you run the model using the provided cli or some custom code?