Cuda different errors CudaErrorIllegalAdress, cuda segmentation fault, error: destroying cuda device: 0

I am working on the following machine:

Processor: Intel Core i7-8750H 
Memory: 16 GB RAM 
1000 GB | 5400RPM | SATA | -,
256 ssd
GPU: NVIDIA GeForce GTX 1060

My deepstream version

DS: 6.1
Driver Version: 515.76
CUDA Version: 11.7

I’m working with deepstream imagedata multistream with adding some integration with redis to add or delete the RTSP streams (using add-remove streams sample) and kafka-python,

The models been used are (primary: resnet10 and Nvds tracker)

After running the code with 1 RTSP stream for X time (where X is different each time but it’s usually more than half hour), I got those different errors at the below screenshoots, kindly advice me what should we do to make that code running 24/7 with 1:30 streams

Hi @ayanasser,

Does this project utilize OptiX? At first glance, this looks like something that might be better addressed in one of the CUDA channels


1 Like

No I don’t use OptiX utilization

How much memory is on your GTX 1060? Can you also share the output of “nvidia-smi” when running your program?

I mention the specs above, also if you mean the Vram is 6GB

The output of nvidia-smi while it’s working smoothly

Can any DeepStream sample applications run without any error in your machine? Can you monitor the GPU status and usage with “nvidia-smi dmon” command during your application running?

I am working on imagedata sample combining with (Add-delete-sources) sample, but with some modifications and integration with redis and kafka as I mentioned,
and again this happens after N working hours (where N is unknown number).

the other Deepstream samples are working fine but I only tested them with 5 minute video or smthing like that (not a long time)

here’s the output of nvidia-smi demon as the deepstream app is working fine.

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.

You need to monitor the GPU and memory status until the error happens.

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