Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
en:services:application_services:high_performance_computing:performance_engineering_and_analysis:nsight_system_compute_graphics [2022/10/27 14:18] – jogajaen:services:application_services:high_performance_computing:performance_engineering_and_analysis:nsight_system_compute_graphics [2022/10/27 15:18] (current) – [Quick Start] jogaja
Line 1: Line 1:
 +==== Nsight ====
 +Nsight is a collections of GPU profiling tool set provided by Nvidia as part of developer tools. Nsight consist of three distinct tools, system, compute and graphics.
 +  * Nsight Systems: Used for analysing application algorithms system-wide
 +  * Nsight Compute: Used for CUDA kernels analysis
 +  * Nsight Graphics: For debugging and analysing graphics workloads.
 +
 +{{:en:services:application_services:high_performance_computing:nsight_flowchart_systemcomputegraphics.png?400|}}
 +
 +=== Quick Start in SCC ===
 +  * Nsight is included with the CUDA development package and to use Nsight in SCC, the prefered cuda module must be loaded first to set the environment correctly.
 +
 +<code>
 +(base) gwdu101:25 17:17:18 ~ > module show cuda
 +-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 +   /opt/sw/modules/21.12/cascadelake/Core/cuda/11.5.1.lua:
 +-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 +...
 +
 +(base) gwdu101:25 17:17:24 ~ > 
 +(base) gwdu101:25 17:18:33 ~ > module load cuda
 +(base) gwdu101:25 17:18:43 ~ >
 +</code>
 +
 +=== Nsight Systems ===
 +To identify the largest optimization opportunities a developer or user should start with a system-wide performance analysis of their application. This can be done using Nsight Systems. Nsight Systems provides developers and users a system-wide visualization of an applications performance from which they can optimize bottlenecks to scale efficiently across any number number of size of CPUs and GPUs from large servers to a SoC. Additional information can be obtained from [Nsight Systems](https://developer.nvidia.com/nsight-systems)
 +
 +=== Nsight Compute ===
 +As an interactive kernel profiler for CUDA applications, Nsight Compute provides detailed performance metrics and API debugging via a user interface and command line tool. Nsight Compute also provides customizable and data-driven user interface and metric collection that can be extended with analysis scripts for post-processing results. Additional information can be obtained from [Nsight Compute](https://developer.nvidia.com/nsight-compute)
 +
 +{{:en:services:application_services:high_performance_computing:kernel_profile_nsightcompute.png?400|}}
 +
 +=== Nsight Graphics ===
 +This is a standalone development tool for debugging, profiling and analysing graphics applications. Nsight Graphics allows optimization of performance of Direct3D 11, Direct3D 12, DirectX Raytracing 1.1, OpenGL, Vulkan, and KHR Vulkan Ray Tracing Extension based applications. Additional information can be obtained from [Nsight Graphics](https://developer.nvidia.com/nsight-graphics)
 +
 +==== Additional information ====
 +  * [Nsight](https://developer.nvidia.com/tools-overview)
 +