Question about buffer-pool-size value in nvstreammux

[DW02689]

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) AGX Jetson
• DeepStream Version 5.0
• JetPack Version (valid for Jetson only) 4.4
• TensorRT Version 7.1.3
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) question

I have a question regarding choosing the buffer-pool-size value in nvstreammux. Our products are running a gst pipeline(gst-launch-1.0) with deepstream from UDP cameras, but we found that our pipeline can easily see high or full RECV-Q value which means buffers are all filled but senders are keeping sending. After specifying buffer-pool-size in the nvstreammux’s property, this backed-up pipeline issue is gone.

In this attempt, I set this buffer-pool-size to 16 randomly but I wonder if there is any way to determine a more appropriate buffer-pool-size scientifically? I searched for it online but nothing related has been found.

Thanks in advance!

[Fiona.Chen]

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

@DW02689

In this attempt, I set this buffer-pool-size to 16 randomly but I wonder if there is any way to determine a more appropriate buffer-pool-size scientifically?

The pool size will impact the processing latency and the upstream/downstream element processing if they are all use the shared GstBuffers. If the downstream elements uses the shared GstBuffers, the pool size should consider the upstream/downstream elements buffer usage too. There is no determine way to know the appropriate value without enough information for the upstream/downstream elements.