I am new to nsight eclipse IDE, I am looking for a starting point/tutorial in form of a video or blog for developing a C/C++ Makefile projects from a x86 linux(PC) on Aarch64 linux(Tegra Tx2) board via ssh.
Any help is appreciated, how to set up a remote ssh connect via nsight eclipse IDE to the tegra board.
Hi Ranjan,
I am in the same situation as you are. I am also trying to figure out the same.
I am trying to compile my code on eclipse for C/C++ with Cuda plugins and execute the code on Jetson nano.
I am able to compile the deepstream SDK sample code but unable to execute it on the nano.
Imported the deepstream-test1 project from /opt/nvidia/deepstream/deepstream-4.0/sources/apps/sample_apps/deepstream-test1
The Eclipse Nsight Project settings
[img]https://devtalk.nvidia.com/cmd/default/download-comment-attachment/82472/[/img]
When I try to run the project remotely I get "Exec format error"
```
Last login: Tue Dec 17 14:42:02 2019 from 192.168.1.56
echo $PWD'>'
/bin/sh -c "cd \"/tmp/cuda-debug\";export LD_LIBRARY_PATH=\"/usr/local/cuda-10.0/bin\":\${LD_LIBRARY_PATH};\"/tmp/cuda-debug/deepstream-test1-app\" /opt/nvidia/deepstream/deepstream-4.0/samples/streams/sample_720p.h264";exit
roy@jetson:~$ echo $PWD'>'
/home/roy>
roy@jetson:~$ /bin/sh -c "cd \"/tmp/cuda-debug\";export LD_LIBRARY_PATH=\"/usr/loocal/cuda-10.0/bin\":\${LD_LIBRARY_PATH};\"/tmp/cuda-debug/deepstream-test1-app\"" /opt/nvidia/deepstream/deepstream-4.0/samples/streams/sample_720p.h264";exit
/bin/sh: 1: /tmp/cuda-debug/deepstream-test1-app: Exec format error
logout
```
I get the same error if run as root on the nano
```
Last login: Tue Dec 17 15:47:24 2019 from 192.168.1.56
echo $PWD'>'
/bin/sh -c "cd \"/tmp/cuda-debug\";export LD_LIBRARY_PATH=\"/usr/local/cuda-10.0/bin\":\${LD_LIBRARY_PATH};\"/usr/local/cuda-10.0/bin/cuda-gdbserver\" --cuda-use-lockfile=0 :2345 \"/tmp/cuda-debug/deepstream-test1-app\" /opt/nvidia/deepstream/deepstream-4.0/samples/streams/sample_720p.h264";exit
root@jetson:~# echo $PWD'>'
/root>
root@jetson:~# /bin/sh -c "cd \"/tmp/cuda-debug\";export LD_LIBRARY_PATH=\"/usr/llocal/cuda-10.0/bin\":\${LD_LIBRARY_PATH};\"/usr/local/cuda-10.0/bin/cuda-gdbservver\" --cuda-use-lockfile=0 :2345 \"/tmp/cuda-debug/deepstream-test1-app\" /opt/nnvidia/deepstream/deepstream-4.0/samples/streams/sample_720p.h264";exit
Process /tmp/cuda-debug/deepstream-test1-app created; pid = 9790
Cannot exec /tmp/cuda-debug/deepstream-test1-app: Exec format error.
Child exited with status 127
gdbserver: Unable to determine the number of hardware watchpoints available.
gdbserver: Unable to determine the number of hardware breakpoints available.
No program to debug
Exiting
logout
</li>
I got hint/help from <a target='_blank' rel='noopener noreferrer' href='https://devblogs.nvidia.com/cuda-jetson-nvidia-nsight-eclipse-edition/'>https://devblogs.nvidia.com/cuda-jetson-nvidia-nsight-eclipse-edition/</a>
</ol>

Similar problem here, “Exec format error”. I got a great tip:
from linux command line, type
readelf -h
This showed me that the executable was compiled for 86_64, not aarch64. I can’t tell you what the solution is, but this makes the problem visible, i.e. the compiler did not CROSS compile, it compiled for its own architecture.
I don’t know, in the zillions of settings, where to change it to CROSS compile, so if anyone can help… added problem being that all the help is for NSight 11.2, not 10.2…