对于x86架构的GPU服务器，我想请教一下超多100路视频流时时分析，应该如果处理，用什么样的GPU，有没有负载均衡解决方案
Please provide the details of your pipeline.
- What types of video streams? USB cameras, CSI cameras, RTSP streams,…?
- The format of the videos? MJPEG, H264, H265, …? The resolution and FPS?
- What kind of inferencing? Which model? The model performance?
- The pipeline you want?
Different pipeline(app) may needs different hardware resources - ethernet bandwidth, USB throughput, GPU loading, CPU loading, decoder loading,…
The solution is based on your app details.
Please discuss in English since the forum is for the global users and global engineers. Thank you!
rtsp h264或者h265视频流这种，就是简单的目标检测和目标跟踪，我这边主要想了解大致的解决方案，而不注重具体的GPU选型
DeepStream pipeline supports multiple streams inferencing. Can you tell us what kind of information do you want us to provide?
运行一个deepstream，能支持100路rtsp么，或者一个GPU服务器应该怎么负载均坑调用gpu，nvidia又这样的官方解决方案么，一个gpu满载后，调用下一个
To answer this question, we need your answers for the questions about the videos.
What kind of GPU loading? Graphics, model inferencing, decoding,…?
您好，因为我这边不是做算法的，所以不能用专业的术语答复您算法问题，简单说不管是推理还是渲染，导致的一块GPU满载了，但是还需要增加rtsp视频流处理，nvidia有没有负载均衡解决方案去分配GPU，还是需要用户自己做调度处理
There is no solution for such case. The low level driver can’t know the behavier of uper level application. The application should take the responsibility.
好的明白了，谢谢
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.