• Hardware Platform (Jetson / GPU)
RTX 5000, x86-64 RHEL • Nsight Systems Version
2021.4 • DeepStream Version
5.1 (nvidia_deepstream:5.1-21.02-devel) • NVIDIA GPU Driver Version (valid for GPU only)
470.57.02 • Issue Type( questions, new requirements, bugs)
Question • How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
nsys profile --trace=cuda,cudnn,nvtx,osrt,opengl --delay=3 --duration=30 …/deepstream-app/deepstream-app -c deepstream_app_config.txt
The latest nsys CLI outputs profile results in a new format (report.nsys-rep) rather than the previous .qdrep output format. I can see the .qdrep intermediate file getting generated and then delted as part of the profiling of the application. How should we view this new format in the Nsight GUI? Is there any way to preserve the .qdrep file for export/sharing? I would like to run a profile of an application using nsys CLI and then transfer the profile output to a different system analysis using the Nsight Systems GUI
I was able to fix this by ensuring the version of Nsight Systems Windows GUI matched the 2021.4 nsys CLI version I used for profiling on the remote system.
I was able to fix this by ensuring the version of Nsight Systems Windows GUI matched the 2021.4 nsys CLI version I used for profiling on the remote system.
Please take a look at the section “Adding Nsight Systems to your Existing Docker Container” → “Dockerfile code example to add Nsight Systems to a Debian-based container image” on Using NVIDIA Nsight Systems in Containers and the Cloud.
For the CLI, on the latter apt-get install command you would have to replace nsight-systems-2020.2.1 with nsight-systems-cli-2021.4.1