Maximum fps on a dgpu?

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) DGPU 3080 TI
• DeepStream Version 6.1
• JetPack Version (valid for Jetson only)
• TensorRT Version 8.2
• NVIDIA GPU Driver Version (valid for GPU only) latest
• Issue Type( questions, new requirements, bugs)
• 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)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)

While our models run on NX’es, we decided to test the DeepStream performance on a Dgpu (rtx3080 ti, i5, 32gb ram). We ran the Default DeepStream-app with the source30-1080p config, which runs multi-ur on a single mp4 file.

When adding more streams (50 instead of the default 30), the frame rate dropped to 16-ish fps. buffers were set accordingly to the amount of streams.
The GPU was barely working at max 13% and the cpu was at 25-35%. Memory seems to be enough and fast enough. So the system is just getting started, but it seems 800-ish fps in total seems to be the maximum.

Can my system achieve more fps and am I missing a bottleneck, or is 800fps just the max?

Have you monitored the GPU performance with “nvidia-smi dmon” while running your case? What is the codec(video decoder/video encoder) performance with 50 sources? Have you set the “sync=0” parameter for [sink] group in deepstream app configuration file?

Hi Fiona, I will try with don this weekend. sync=0 indeed. And buffers were set according to nr of streams

Sorry for the wait. Nvidia-smi dmon showed 100% on the decoder. Gpu is 14% max. Is there a method to get the decoder utilization down, or am I at the limit?

GPU video codec (encoder/decoder) capability is listed here Video Encode and Decode GPU Support Matrix [NEW] | NVIDIA Developer

Your case reaches to the hardware limitation. You may need to use software decoder for some of your streams since there is still some CPU capability to be used.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.