Nvidia Training Course : Find the Bottleneck – Optimize AI Pipelines With Nsight Systems

sm-nfb
Durée : 0.5jour
Tarif : 650€ HT

97%

Customer satisfaction rating
(based on 1838 reviews from 05/19/21 to 10/07/26)

See reviews

Objectifs

By participating in this workshop, you will be equipped to:

Profile AI applications using NVIDIA Nsight Systems to identify performance bottlenecks.

Annotate code with NVTX (NVIDIA Tools Extension) for better visualization and analysis.

Analyze and optimize data transfers between host CPU and GPU memory.

Leverage GPU-accelerated libraries to eliminate unnecessary data movement.

Use Nsight Systems plugins and multi-report analysis for advanced performance insights.

Apply optimization strategies to real-world AI pipelines for significant speedups.

Prérequis

Basic Python programming experience and familiarity with Jupyter notebooks.

Understanding of GPU computing concepts (CUDA knowledge helpful but not required).

Familiarity with deep learning frameworks, particularly PyTorch.

Basic understanding of video processing concepts.

Public

IT professionals

Dernière mise à jour

Programme mis à jour le01 October 2026

Cette formation vous intéresse ? Contactez-nous

Bon à savoir

Assess your level

To help you choose the right training course, we offer either an interview with the instructor or an assessment test. This ensures you have the necessary knowledge to complete the training under optimal conditions.

Our services

We provide certain IT services for you: custom development, software redesign, application maintenance, ERP interfacing.

Practical exercises

Our training courses include numerous practical exercises for enhanced learning (60% hands-on). We also offer the opportunity to work with your own data for a more personalized experience.

Guaranteed sessions

The majority of our remote sessions are guaranteed. They can be recorded upon request.

Télécharger la fiche de cette formation

Les Modules
de formation

Module1

Introduction

Understand the video segmentation pipeline for background blurring.

Learn about the components: ffmpeg, OpenCV, and PyTorch.

Review the optimization workflow with Nsight Systems.

Module2

First Optimization Iteration

Annotate applications with NVTX for better profiling visibility.

Generate baseline profiles using Nsight Systems.

Navigate the Nsight Systems GUI and timeline views.

Identify initial performance bottlenecks.

Module3

Optimize Data Transfers

Analyze CPU-GPU data transfer patterns and bottlenecks.

Implement GPU-accelerated video decoding/encoding with PyNvVideoCodec.

Eliminate unnecessary host-device memory copies.

Verify optimizations through profiling.

Module4

Nsight Systems Plugins

Explore Nsight Systems plugin architecture.

Use built-in collector plugins for extended metrics.

Understand how to create custom data collectors.

Module5

Advanced Analysis

Learn about multi-report analysis for parallel applications.

Use recipe scripts for statistical analysis across reports.

Identify load balancing issues and communication patterns.

Generate automated performance reports.

Module6

Multi-Node Performance

Profile applications in distributed computing environments.

Use Nsight Systems with cluster schedulers like Slurm.

Apply data collection strategies for large-scale applications.

Analyze performance across multiple nodes and ranks.

Les prochaines
sessions de formation

Sur demande

Vous souhaitez organiser cette formation à une date spécifique ?Contactez-nous en remplissant le formulaire ci-dessous

Interested in this training? Contact us!