Hi, I’m having an issue adding a new remote connection within nsight compute. I’m running nsight on ubuntu 18.04 and trying to add a remote connection which is a jetson xavier. When I use the connection menu to add a new embedded linux remote connection, and hit “Add” after filling out the IP address etc, nothing seems to happen. There is no new connection added to the dropdown menu in the “Remote Launch” tab, and no error appears. Not sure what’s happening here. Any advice? Thank you
Noting that I’m able to add a new remote connection if I use the hostname instead of the device’s IP address
There was an issue that creating a connection with the same (host) name didn’t work, if it existed for any target platform already. Instead, it would fail silently. In the latest version, this should already be fixed and you should see a dialog notifying you about overriding another connection. Is it possible that a connection with the same IP address as host name already existed?
Could you let us also know which exact version of Nsight Compute you are using (as shown in the About dialog for example)?
Ah thanks for letting me know, that is probably what’s happening. The IP address is also present in a non-embedded target. I’m using Nsight Compute v 1.0. To use the latest version, do I upgrade via jetpack/sdkmanager or install the host nsight compute app directly?
I queried internally, but I haven’t heard back yet. Still, using jetpack to update should be the correct way, yes. Note that Nsight Compute is shipped as part of the CUDA toolkit within jetpack/sdkmanager, so that’s likely the package you need to upgrade. Let us know if there are further questions.