Built for Intel Core Ultra Processors

OpenVINO NPU Inference
Benchmark Suite

Unlock the full potential of your AI PC. A professional-grade benchmarking framework designed to validate, test, and optimize AI inference performance on Intel NPU, CPU, and GPU.

8.9x
Max NPU Speedup
5.3x
Avg NPU Speedup
1.1ms
Fastest Latency
17
Models Tested

Everything You Need for AI Benchmarking

Comprehensive tools to measure, compare, and optimize AI inference across all Intel hardware.

🚀

Multi-Device Support

Seamlessly benchmark across Intel CPU, Arc/Integrated GPU, and AI Boost NPU. Direct side-by-side performance comparisons.

📊

Interactive Dashboard

Stunning glassmorphic web UI to run tests and visualize real-time performance with beautiful animated charts.

📑

Professional Reports

Generate detailed HTML reports with hardware specifications, speedup metrics, and downloadable results.

🔄

Model Zoo

Curated collection of industry-standard models including ResNet, YOLO, BERT, and Vision Transformers. All pre-configured for NPU.

INT8 Quantization

Compress models for faster NPU inference with negligible accuracy loss. Up to 4x compression ratio.

🎯

Batch Sweeps

Automatically find the optimal batch size for maximum throughput on each device type.

Interactive Web Dashboard

Launch the beautiful web interface with a single command. Run benchmarks, visualize results, and download reports all from your browser.

  • Real-time benchmark progress tracking
  • Interactive performance charts
  • Hardware specifications display
  • One-click professional HTML reports
  • Model selection with search & filters
$ npu-benchmark web
⚡ NPU Benchmark Dashboard

127.0.0.1:5000

5.2x
NPU Speedup
12ms
Avg Latency

Real Benchmark Results

Actual performance measurements from an Intel Core Ultra 7 255H system with Intel AI Boost NPU.

💻
Processor
Intel Core Ultra 7 255H
16 cores, 16 threads
🧠
NPU
Intel AI Boost
FP16 ✓ INT8 ✓
💾
Memory
30.9 GB RAM
System Memory
🖥️
Platform
Windows 11
Python 3.13
8.9x
Max NPU Speedup
ResNet-50
5.3x
Avg NPU Speedup
All Models
1.1ms
Fastest Latency
MobileNetV3
17
Tests Completed
Classification Models

📊 Performance Comparison (Latency in ms)

Model CPU (ms) NPU (ms) Speedup
ResNet-50 73.1 8.5 8.6x
EfficientNet-B0 17.8 3.5 5.1x
MobileNetV3-Small 3.6 1.2 3.0x
* Results from Intel Core Ultra 7 255H • Lower latency is better • NPU shows significant efficiency gains
CPU
NPU

Pre-configured AI Models

Industry-standard models automatically downloaded, converted, and optimized for NPU.

👁️ Computer Vision
ResNet-50 Classification
MobileNetV3 Edge Optimized
EfficientNet-B0 Balanced
🎯 Object Detection
YOLOv8n / YOLOv8s Real-time
YOLO11n / YOLO11s Latest
📝 Natural Language
BERT-Base Transformers
DistilBERT Efficient
ViT (Vision Transformer) Hybrid

Get Up and Running in Minutes

Simple installation with pip. Works on Windows 10/11 and Linux with Python 3.10+.

1

Clone the Repository

git clone https://github.com/singhraghvendra2104/OpenVINO-NPU-Inference-Benchmark-Suite.git
cd OpenVINO-NPU-Inference-Benchmark-Suite
2

Install the Package

pip install -e .
3

Launch the Dashboard

npu-benchmark web

Open http://127.0.0.1:5000 in your browser

Ready to Benchmark Your AI PC?

Discover the true AI performance of your Intel Core Ultra processor. Start benchmarking today with our open-source suite.