NVIDIA Nsight Compute

Nsight Compute 2024.1 is now archived

Get the latest version

Nsight Compute allows profiling on x86_64 Windows, Linux, PowerPC, and Arm Server Based System Architecture platforms locally or from Windows, Linux, or MacOS hosts. See System Requirements for more information.



Download
Latest
Release
Nsight Compute 2024.1 (Windows)

Download
Latest
Release
Nsight Compute 2024.1 (Linux Desktop)

Download
Latest
Release
Nsight Compute 2024.1 (MacOS)

Download
Latest
Release
Nsight Compute 2024.1 (PowerPC)

Download
Latest
Release
Nsight Compute 2024.1 (ArmServer)

Download
Latest
Release
Nsight Compute is available in the CUDA Toolkit bundled in the JetPack SDK.

Download
Latest
Release
Nsight Compute is available as part of DRIVE OS for development and deployment on DRIVE AGX-based autonomous vehicles.

Download NVIDIA Nsight Tools JupyterLab Extension:
The Nsight Tools JupyterLab Extension integrates Nsight Compute's CUDA kernel profiling functionality in JupyterLab.
Download Nsight Tools JupyterLab Extension (PyPI)


Nsight Compute is also available as part of the CUDA Toolkit

Read Nsight Compute 2024.1 documentation


System Requirements

Windows Linux Mac DRIVE OS
Host Windows x86_64[1] Linux x86_64[1]
Linux aarch64 sbsa[1]
Linux aarch64 (L4T)[2]
MacOS[1] -
Target Windows x86_64[1] Linux x86_64[1]
Linux PowerPC[1]
Linux aarch64 sbsa[1]
Linux aarch64 (L4T)[2]
- DRIVE OS QNX aarch64[2][3]
DRIVE OS Linux aarch64[2][3]

Host platforms support the Nsight Compute UI for viewing reports, interactive profiling and remote connections. Applications are profiled on target platforms, which also support the Nsight Compute command line interface.


Supported NVIDIA GPU architectures

  • Ada: AD10x
  • Ampere: A100 with Multi-Instance GPU, GA10x
  • Hopper: H100 with Multi-Instance GPU
  • Turing: TU1xx
  • Volta: GV100[1], GV10B[2]

[1] available in this download and the CUDA Desktop Toolkit

[2] available in the Embedded or Drive toolkits only

[3] Only the command line interface (CLI) is provided for these platforms. There is no Nsight Compute GUI application for these platforms


Recommended Drivers

  • NVIDIA Windows display driver - 551.78 or newer
  • NVIDIA Linux display driver Linux - 550.54.15 or newer

We recommend using drivers provided with the most recent CUDA Toolkit production release or a newer version. Older driver versions are also supported.




Release Notes

2024.1 Update 1 - 2024/04/03

2024.1 - 2024/03/05

  • Support for the CUDA Toolkit 12.4.
  • Added support for CUDA Green Contexts.
  • Added SASS view and Source Markers support in Source Comparison.
  • Improved Source Comparison diff visualization.
  • Added a Launch Details tool window.
  • Added a new optional PmSampling_WarpStates section.
  • Added a new rule for detecting load imbalances.
  • Various performance improvements and bug fixes.
  • View full release notes

2023.3 Update 1 - 2023/11/15

  • Support for the CUDA toolkit 12.3 Update 1.
  • Switched to using OpenSSL version 1.1.1w.
  • Improved the speedup estimates for rule IssueSlotUtilization as well as its child rules.
  • Updated report files and documentation for the samples in this release.
  • View full release notes

2023.3 - 2023/10/19

  • Support for the CUDA Toolkit 12.3.
  • Adds periodic metric sampling.
  • Adds side-by-side profiled source comparison.
  • Improved accuracy of estimated speedups.
  • Expanded OS support for WSL2 profiling.
  • Various bug fixes.
  • View full release notes

2023.2 Update 2 - 8/29/2023

2023.2 Update 1 - 7/25/2023

2023.2 - 6/26/2023

  • Support for the CUDA Toolkit 12.2.
  • Adds rules to show potential performance improvement estimates for prioritization.
  • Workflow improvements and bug fixes.
  • View full release notes

2023.1 Update 1 - 4/18/2023

  • Support for the CUDA Toolkit 12.1 Update 1.
  • New configuration options to set the default view mode and precision for the Source page.
  • Workflow improvements and bug fixes.
  • View full release notes

2023.1 - 2/28/2023

  • Support for the CUDA Toolkit 12.1 and OptiX 7.7.
  • New application range replay mode<./li>
  • Added sharedBankConflicts sample CUDA application.
  • CPU and GPU NUMA topology metrics and NUMA Affinity sections.
  • Performance improvements and source file re-resolve on the source page.
  • View full release notes

2022.4 Update 1 - 1/30/2023

  • Support for the CUDA Toolkit 12.0 Update 1.
  • Support for the latest ADA GPUs, including AD104, AD106, and AD107.
  • View full release notes

2022.4 - 12/8/2022

  • Support for the CUDA Toolkit 12.0.
  • Profiling of complete CUDA graphs and device-sided graph launches.
  • OptiX resource tracking, export and Acceleration Structure viewer enhancements.
  • View full release notes

2022.3 - 08/03/2022

  • NVIDIA Ada Lovelace GPU architecture support
  • NVIDIA Hopper GPU architecture support
  • Acceleration Structure viewer supports multiple traversable handles and uses OptiX naming for displayed elements.
  • View full release notes

2022.2 - 05/11/2022

  • Improved performance for profiling and metric query
  • Expanded Resources tool window
  • Acceleration Structure viewer for debugging and optimizing Optix API ray tracing.
  • View full release notes



Archives

View the full Nsight Compute Release History.




Resources

Nsight Compute Documentation


Release Notes

Release notes, including new features and important bug fixes. Supported platforms and GPUs. List of known issues for the current release.


Nsight Tools Tutorials

Access the latest resources to get started with Nsight Compute.


Kernel Profiling Guide

Kernel Profiling Guide with metric types and meaning, data collection modes and FAQ for common problems.


Nsight Compute

NVIDIA Nsight Compute User Interface (UI) manual. Information on all views, controls and workflows within the tool UI. Transitions guide for Visual Profiler.


Nsight Compute CLI

NVIDIA Nsight Compute Command Line Interface (CLI) manual. Information on workflows and options for the command line, including multi-process profiling and NVTX filtering. Transitions guide for Nvprof.

Documentation



Tutorial Sessions

Check out a catalog of Nsight Compute training videos.

Become Faster in Writing Performant CUDA Kernels using the Source Page in Nsight Compute

Learn how to make the most of the Source Page in Nsight Compute to quickly pinpoint and resolve bottlenecks in your CUDA kernels.

Watch

From the Macro to the Micro: CUDA Developer Tools Find and Fix Problems at Any Scale

Understand how your multi-node CUDA workload is scaling across machines and how a GPU assembly instruction is moving through the pipeline

Watch

What's New in CUDA Developer Tools: Profiling NVIDIA Hopper and workflow enhancements

The latest updates to CUDA developer tools include workflow enhancements, new supported environments, and profiling features for the latest NVIDIA Hopper-based platforms.

Watch

Optimizing CUDA Machine Learning Codes with Nsight Profiling Tools

This lab teaches how to use NVIDIA's Nsight tools for analyzing and optimizing CUDA applications.

Enroll

Guided Analysis with Nsight Compute

Guided analysis is the set of features in NVIDIA Nsight Compute that provides expert analysis of collected profile data, including insights into performance issues and options to fix them.

Watch

Tools and Techniques for Making Efficient Use of GPUs

We look at programming GPUs and some of the technologies available to minimize bottlenecks both within and across nodes. In addition, some tools for scientific visualization of data will be presented.

Watch



Support

Nsight Compute forum

To provide feedback, request additional features, or report Nsight Compute issues, please use the Developer forums.


Training

NVIDIA Nsight Compute blogs, videos, samples and tutorials.