I have a Mac with an unsupported GPU...but I also have a Windows 10 machine running SSH that does have a supported GPU (and fully setup to run CUDA).
I'm trying to set up the remote target system on my Mac within Nsight Eclipse Edition. I can add the remote connection to the Windows machine without issue, but when I try to configure the toolkit for the remote connection I run into problems.
I can enter (or browse on the remote Windows computer) the path to the Toolkit binaries. No matter what I enter, it will always says something like '/C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v9.2/bin/nvprof' does not exist'.
If I click on the 'Detect...' button, it will try to detect available toolkits, but will then pop an error dialog saying "Unable to execute shell commands. Failed to execute "bash -i which nvcc". Obviously that won't work on Windows.
What this is telling me is that a Windows machine cannot be used as a remote target. Is this correct?
What this is telling me is that a Windows machine cannot be used as a remote target. Is this correct?
yes, that is correct.
Referring to the Nsight EE Documentation:
Section 3.6 indicates that for development, the remote system must be linux.
The remote target system can be a Linux x86 or POWER system with an NVIDIA GPU or an Tegra-based ARM system.
Section 3.7 indicates that for debugging, the remote (target) must be linux.
Only Linux targets are supported.