Installing Nsight system in Jetson Nano - Jetpack 4.6.1

[srua]

Hi everyone.

I need help to install nsight system in my jetson nano with jetpack 4.6.1. My jetson nano is connected in a remote location, so I can not use visual sdk manager to install it. I have connecttion to my jetson via ssh. Do you have some manual to install nsight system via terminal to the jetson??

Can I download a .deb? I was trying to install some of this and didnt work

Gameworks Download Center

At the end I just want to run the application, generate the file.nsys-rep and copy that remote to my local machine, where I can read it.

[hwilper]

@Andrey_Trachenko can you find someone to answer this question.

[srua]

Hi.

I little update. I found this repo today:

https://repo.download.nvidia.com/jetson/

I try installing the following three version:

1. nsight-systems-cli-2021.2.3_2021.2.3.8-73c8c79_arm64.deb
2. nvidia-nsight-sys_4.6.3-b17_arm64.deb
3. nsight-systems-linux-tegra-public-2021.5.4.19-e642d4b_2021.5.4.19-e642d4b_arm64.deb

Once I installed by doing ‘nsys status -e’, I get

Timestamp counter supported: Yes
Sampling Environment Check
Linux Kernel Paranoid Level = 2: OK
Linux Distribution = Ubuntu
Linux Kernel Version = 4.9.253-tegra: OK
Linux perf_event_open syscall available: OK
Sampling trigger event available: OK
Intel(c) Last Branch Record support: Not Available
Root privileges: Yes
Kernel module: Available
Sampling Environment: OK

Which seems that is ok. Now following this forum: Nsight Systems Profiling in L4T Container?

,I launch my container with the following command

sudo docker run
-it
–rm
–privileged
–net=host
–runtime nvidia
-w /opt/nvidia/deepstream/deepstream
-v /tmp/.X11-unix/:/tmp/.X11-unix
-v ~/elvistech/jetson-app/deepstream/app:/home/app
-v /opt/nvidia/nsight-systems-cli/:/opt/nvidia/nsight-systems-cli/
-w /home/app/
deepstream-python:v5

Inside the container I launch the profiler with the following command
/opt/nvidia/nsight-systems-cli/bin/nsys profile -t cuda,osrt,nvtx,cudnn -y 40 -d 20 python3 app.py

This app.py is a deepstream application

A .nsys-rep is generated but it doesnt collect any traces. This is the report.