NvMultiurisrcbin doesn't set gpu-id

• Hardware Platform (Jetson / GPU)
dGPU
• DeepStream Version
Deepstream 6.3

I’ve noticed that nvmultiurisrcbin doesn’t work with anything other than gpu-id 0.
The pipeline below runs without errors
gst-launch-1.0 nvmultiurisrcbin port=9000 ip=localhost max-batch-size=1 uri-list=file://CH2.mp4 width=1920 height=1080 ! nvmultistreamtiler ! nvvideoencfilesinkbin output-file=out.mp4
but the pipeline below
gst-launch-1.0 nvmultiurisrcbin port=9000 ip=localhost max-batch-size=1 gpu-id=1 uri-list=file://CH2.mp4 width=1920 height=1080 ! nvmultistreamtiler ! nvvideoencfilesinkbin output-file=out.mp4
gives this error
ERROR: from element/GstPipeline:pipeline0/GstDsNvMultiUriBin:dsnvmultiuribin0/GstBin:dsnvmultiuribin0_creator/GstNvStreamMux:src_bin_muxer: Memory Compatibility Error:Input surface gpu-id doesnt match with configured gpu-id for element, please allocate input using unified memory, or use same gpu-ids OR, if same gpu-ids are used ensure appropriate Cuda memories are used

This is a issue for nvmultiurisrcbin, and we will fix it in the next version. Thanks

1 Like

@yuweiw Is there an estimated timeframe for addressing this bug? Its impact on my system’s production deployment is significant. I would greatly appreciate it if you could expedite the resolution of this issue. Thank you kindly

It’s resolved on the DeepStream 6.4. You can try that. Thanks

1 Like

In deepstream 6.3 Medium nvurisrcbin will there also a version problem?
ERROR from element streammux_1: Memory Compatibility Error:Input surface gpu-id doesnt match with configured gpu-id for element, please allocate input using unified memory, or use same gpu-ids OR, if same gpu-ids are used ensure appropriate Cuda memories are used
Error details: gstnvstreammux.cpp(1691): copy_data_cuda (): /GstPipeline:ds-pipeline/GstNvStreamMux:streammux_1:
surface-gpu-id=0,streammux_1-gpu-id=1

Please upgrade DeepStream to our latest version DeepStream 7.1. Thanks

We need to make minimal changes to the environment. Does version 6.4 have this issue

No. But we still recommend that you update to the latest version. This part is open source now and you can customize that yourself.