• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version 5.0
• TensorRT Version 7.1.3
• NVIDIA GPU Driver Version (valid for GPU only) CUDA 10.2
Im facing a problem with Nvmultistreamtiler. When i add tiler component into my pipeline it run successfuly. however after some minute it raise a error like below:
But when I remove tiler in my pipeline and there isnt any error.
How can i fix it? thanks a lot.
Hey， can you share your pipeline with me?
In addition, pls make sure you can run deepstream-app or deepstream-test3 with multiple sources well
The problem boils down to the combination of the muxer and demuxer as I described in the Topic: Nvmultistreamtiler - Segmentation fault
To add the tiler or leave it out creates just another symptom of an underlying problem. Most likely within the metadata created by the muxer that are not correctly handled by the demuxer.
Furthermore, I like to ask the NVIDIA Dev Team to have a deeper look into the diff files, I’ve provided, to recreate the issue.
It is even easier to just combine a muxer and demuxer directly (batch size of 1) to bump into the same issue.
Thanks for your help!
@sh2222 We had fixed a demux issue recently and had shared the library in Nvmultistreamtiler - Segmentation fault , let’s track your issue in that topic, I don’t think you are facing the same issue.
this is my pipeline, when i remove tiler it work, but i add tiler it raise above error.
I replaced file that you suggested. However, I still see the error. I checked file again but I replaced correctly.
Can you try deepstream-test3 with multiple sources, I can run test3 well, could you confirm if you can run in your device?
In addition, the tiler should be before the nvvidconv in your pipeline, can you refer test3’s pipeline?
Thanks a lot. I saw that my problem is allocate memory in user meta data.
Have nice day!