Recommended/Intended Cuda Development Toolchain for Linux (Ubuntu)

I’m looking at the current 10.1.130 Release Notes and noticed that “Nsight Eclipse Edition” is deprecated in 10.1. I always used emacs, but thought that was the goto tool in order to visualize/debug/profile cuda code. I was wondering what the intended tools are now to perform these tasks on a Linux system for development, I’m relatively new to Cuda dev myself.

Would love to hear generally what kind of toolchains professional Cuda Linux devs use as well.

When researching a replacement for profiling capability of compute 7.5 with CUDA 10, I found this blurb on the nVidia web site

Yes, nvprof profiling is not supported for device with compute capability > 7.2, we suggest to use our new tools Nsight Compute CLI (command line tool) and Nsight Compute (UI) for profiling on 7.2 onward. It is available under directory.

/usr/local/cuda/NsightCompute*

It is also documented under What’s New
https://docs.nvidia.com/cuda/profiler-users-guide/index.html#whats-new

For more details about Nsight Compute, please check

https://docs.nvidia.com/nsight-compute/NsightCompute/index.html
https://docs.nvidia.com/nsight-compute/NsightComputeCli/index.html

Please let us know the feedback.
docs.nvidia.com
Profiler :: CUDA Toolkit Documentation
The user manual for NVIDIA profiling tools for optimizing performance of CUDA applications.
docs.nvidia.com
Nsight Compute :: Nsight Compute Documentation
The user manual for NVIDIA Nsight Compute.
docs.nvidia.com
Nsight Compute CLI :: Nsight Compute Documentation
The user manual for the NVIDIA Nsight Compute Command Line Interface.