search

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )
http://pjreddie.com/darknet/
Other
21.75k stars 7.96k forks source link

GPU utilization is low #6372

Open jintengli opened 4 years ago

jintengli commented 4 years ago

Environment: Win10 OpenCV 4.4.0 CUDA 11.0 CUDNN 8.0.2 Hardware: GPU: RTX 2080 ti CPU: Intel Xeon Gold 6126 GPU utilization is very low(7-12%) when running YOLO v4-tiny with video input. Also, the memory usage for GPU is low(800 MiB/11264 MiB). Is there any way to increase the GPU utilization?

jintengli commented 4 years ago

And CPU utilization is also very low

Szamtu commented 4 years ago

GPU memory usage when running detector: Normal yolo v4 about 1 gb. Tiny yolo v4 about 300 mb. Both for input size 416x416

Is your darknet app detecting your gpu correctly? Check first lines after running darknet app ./darknet detector demo cfg/coco.data cfg/yolov4-tiny.cfg yolov4-tiny.weights 1.mp4

 CUDA-version: 11000 (11000), cuDNN: 8.0.0, CUDNN_HALF=1, GPU count: 1  
 CUDNN_HALF=1 
 OpenCV version: 4.4.0
Demo
 0 : compute_capability = 610, cudnn_half = 0, GPU: GeForce GTX 1060 6GB

If darknet app runs slowly, you can always try to run it from opencv Yolo demo. tutorial_dnn_yolo

yancccc commented 4 years ago

Environment: Win10 OpenCV 4.4.0 CUDA 11.0 CUDNN 8.0.2 Hardware: GPU: RTX 2080 ti CPU: Intel Xeon Gold 6126 GPU utilization is very low(7-12%) when running YOLO v4-tiny with video input. Also, the memory usage for GPU is low(800 MiB/11264 MiB). Is there any way to increase the GPU utilization?

I have the same problem. Have you solved it

jintengli commented 4 years ago

Environment: Win10 OpenCV 4.4.0 CUDA 11.0 CUDNN 8.0.2 Hardware: GPU: RTX 2080 ti CPU: Intel Xeon Gold 6126 GPU utilization is very low(7-12%) when running YOLO v4-tiny with video input. Also, the memory usage for GPU is low(800 MiB/11264 MiB). Is there any way to increase the GPU utilization?

I have the same problem. Have you solved it

Not yet, I use the same environment and hardware with you.

RohitSingh1226 commented 2 years ago

Any solution @jintengli @yancccc ?