Running engine files in Deepstream slows down

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
Jetson
• DeepStream Version
5.0
• JetPack Version (valid for Jetson only)
4.4
• TensorRT Version
7.1
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs)
questions
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
After going out for a period of time, Xavier has been turned on during this period. When I returned and ran the engine files in Deepstream, the running speed was reduced to half of the previous speed. Some of these engine files were converted through TLT, and others were converted through other API interfaces. I have checked almost all configuration files. any idea?
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)

2

I have encountered before when the cooling system is turned off or not applied.

If the temperature reaches almost 90 degrees, the system will decrease the clock speed and run slower than before. In this situation, if I run the same configuration again, I get smaller FPS.

3

hi, mehmetdeniz, I have disconnected Xavier from the power supply for a few hours, I think it has cooled down enough, and when I switched it back on, the operating speed still did not improve.

4

A very special phenomenon, no matter what size yolov5 weights I run, the FPS is limited to about 6. Similarly, Faster-RCNN is limited to around 3.

5

I tried yolov3 example from objectDetector_Yolo and I got ~60 FPS with one source.
(JetPack 4.4; Deepstream 5.0.0)

Have you tried this example or could you give some details about your setup for yolov5 to help further?