NVIDIA / NVIDIA/cuda-python

[BUG]: Replace obsolete Nsight GUI utility names with nsys-ui and ncu-ui

Open
#2,654 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

triage
Dominant language
Cython
Stars
3.4k
Forks
329
Avg merge
1d 23h
Merged PRs (30d)
116

Description

Problem

cuda.pathfinder.SUPPORTED_BINARY_UTILITIES currently includes the legacy GUI names nsight-sys and nsight-compute, but omits the executable names used by supported Nsight releases: nsys-ui and ncu-ui.

Because find_nvidia_binary_utility() treats the requested name literally (and appends .exe on Windows), calls using the legacy names return None on a current standalone Nsight installation even though the corresponding GUI is installed.

The CLI entries nsys and ncu are correct and should remain unchanged.

Evidence

These legacy names predate the CUDA Toolkit majors currently covered by cuda.pathfinder's “two most recent major versions” support policy.

Proposed change
  1. Replace nsight-sys with nsys-ui and nsight-compute with ncu-ui in the supported binary descriptors.
  2. Preserve nsys and ncu as the CLI utility names.
  3. Add standalone Windows discovery through the existing Nsight registry roots:
    • Nsight Systems: locate nsys-ui.exe under the host directory (for example, host-windows-x64).
    • Nsight Compute: locate the root shortcut and/or the actual executable under host/windows-desktop-*/ncu-ui.exe.
  4. Add tests for wheel/Conda layouts and standalone Windows layouts.

If removing names from the public supported tuple is considered too disruptive, a compatibility alternative is to retain the old inputs temporarily as deprecated aliases that resolve to the current UI executable, while documenting that the returned basename differs.

Expected behavior
find_nvidia_binary_utility("nsys-ui")  # absolute path to nsys-ui[.exe]
find_nvidia_binary_utility("ncu-ui")   # absolute path to ncu-ui[.exe]
find_nvidia_binary_utility("nsys")     # unchanged
find_nvidia_binary_utility("ncu")      # unchanged

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with cuda_pathfinder/cuda/pathfinder/_binaries/supported_nvidia_binaries.py and find_nvidia_binary_utility.py to trace supported names and platform-specific lookup behavior. Add coverage for wheel, Conda, and standalone Windows layouts, then verify that nsys-ui and ncu-ui resolve to absolute executable paths while nsys and ncu remain unchanged.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
tooling
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
55/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.