deepstream-profile-pipeline

Profile NVIDIA DeepStream pipelines with Nsight Systems and nvidia-smi.

Updated Nov 21, 2025

Search Tags:#benchmarking #profiling #performance-tuning #nsight-systems #video-analytics #deepstream

Ready to run

npx skills add https://github.com/olibartfast/deep-infer --skill deepstream-profile-pipeline

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: deepstream-profile-pipeline
Source: https://github.com/olibartfast/deep-infer/tree/main/skills/deepstream-profile-pipeline
Command: npx skills add https://github.com/olibartfast/deep-infer --skill deepstream-profile-pipeline

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill removes the guesswork from performance tuning by providing a structured, measurement-based approach to identifying bottlenecks in NVIDIA DeepStream video analytics pipelines.

Core Features & Use Cases

Bottleneck Diagnosis: Automatically classifies pipeline performance as decode, compute, memory, or tracker-bound using Nsight Systems and nvidia-smi metrics.
Config Derivation: Applies closed-form rules to set optimal batch sizes, memory types, and inference parameters based on measured hardware ceilings.
Use Case: When you need to determine the maximum number of RTSP streams your current GPU can handle at 30 FPS, this skill profiles your pipeline and provides a definitive capacity report with actionable hardware upgrade paths.

Quick Start

Use the deepstream-profile-pipeline skill to profile my current pipeline and report the maximum stream capacity for this GPU.

Frequently Asked Questions about deepstream-profile-pipeline

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I find the maximum RTSP stream capacity for my DeepStream pipeline?▼

To find maximum stream capacity, profile your DeepStream pipeline using Nsight Systems and nvidia-smi to measure hardware ceilings and generate a definitive capacity report with actionable upgrade paths.

How do I diagnose bottlenecks in an NVIDIA DeepStream video analytics pipeline?▼

Diagnose DeepStream pipeline bottlenecks by classifying performance as decode, compute, memory, or tracker-bound using Nsight Systems metrics and nvidia-smi data to isolate the limiting hardware resource.

Do I need Nsight Systems and nvidia-smi to tune DeepStream performance?▼

Yes, you need Nsight Systems and nvidia-smi available within a compatible container environment alongside the DeepStream SDK to measure hardware metrics and derive optimal pipeline configurations.

What is the best way to configure batch size and memory type for DeepStream inference?▼

The best way to configure batch size and memory type is applying closed-form rules based on measured hardware ceilings derived from Nsight Systems profiling and nvidia-smi metrics.

Can I use this profiling approach for capacity planning across different GPU architectures?▼

Yes, you can use this measurement-based profiling approach for capacity planning and benchmarking of inference pipelines across various hardware architectures to determine hardware upgrade paths.

Why does my DeepStream pipeline drop frames during video analytics processing?▼

Frame drops occur when pipeline performance hits hardware limits; profiling with Nsight Systems identifies whether decode, compute, memory, or tracker bottlenecks cause the constraint.