Performance Issue on DeepStream with RTX 4090 – Low GPU Utilization

Please provide complete information as applicable to your setup.

• Hardware Platform (GPU)
• DeepStream Version 7.0

**• TensorRT Version ** 24.08
**• NVIDIA GPU Driver Version ** CUDA Version: 12.2 -NVIDIA GeForce RTX 4090

Performance Issue on DeepStream with RTX 4090 – Low GPU Utilization

I’m encountering performance issues when running the DeepStream example on an RTX 4090 with 60 video streams. The GPU usage is quite low, around 15-20%, and the FPS is only around 30. I have already tried adjusting settings like NVDS_DISABLE_CUDADEV_BLOCKINGSYNC, but it doesn’t seem to fully utilize the GPU’s resources.

I’m running the example configuration source30_1080p_dec_infer-resnet_tiled_display_int8.txt, but I still can’t achieve near 100% GPU usage as mentioned in your page.

What can I do to increase GPU utilization and performance to match the potential of the RTX 4090? Any advice or specific settings I should consider?

image

Please refer to Performance — DeepStream documentation 6.4 documentation for how to configure for the performance test.

Please attach your deepstream-app configuration file.

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks