Skip to content

Instantly share code, notes, and snippets.

@agostini01
Last active May 3, 2024 14:17
Show Gist options
  • Save agostini01/484bcfac8c5a99312e28fb6994284e9f to your computer and use it in GitHub Desktop.
Save agostini01/484bcfac8c5a99312e28fb6994284e9f to your computer and use it in GitHub Desktop.

Links for Nisght tools and tutorials

Download center

Installing Nsight compute from command-line:

# Download .run file from website and accept the terms and conditions.
sudo ./nsight-compute-linux-2020.1.1.8-28506821.run --nox11

Transitioning to Nsight Systems

ERR_NVGPUCTRPERM permission error

Profilling a remote machine with the GUI and instructions on root privillegis

Running the tools

Nsight Systems command-line example:

nsys profile --trace=cuda -o report-nsys-nbody /usr/local/cuda/extras/demo_suite/nbody -benchmark -numbodies=500224

Nsight Compute command-line example:

# Nsight Compute REQUIRES sudo privilleges
# Profilling integrateBodies kernel number 6 
sudo /usr/local/NVIDIA-Nsight-Compute/ncu -k integrateBodies -s 6 -c 1 -o report-ncu-nbody '/usr/local/cuda/extras/demo_suite/nbody' -benchmark -numbodies=500224
# More information - this will take a long time (5min) for a single kernel
sudo /usr/local/NVIDIA-Nsight-Compute/ncu -k integrateBodies -s 6 -c 1 -o report-ncu-nbody \
  --section ComputeWorkloadAnalysis \
  --section InstructionStats \
  --section LaunchStats \
  --section MemoryWorkloadAnalysis \
  --section MemoryWorkloadAnalysis_Chart \
  --section MemoryWorkloadAnalysis_Tables \
  --section Occupancy \
  --section SchedulerStats \
  --section SourceCounters \
  --section SpeedOfLight \
  --section SpeedOfLight_RooflineChart \
  --section WarpStateStats \
  /usr/local/cuda/extras/demo_suite/nbody -benchmark -numbodies=500224

Correlating source code with SASS in Nsight Compute

Creating custom report sections in Nsight Compute

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment