Intel CPU/iGPU Profiler

A comprehensive benchmarking and profiling suite for Intel CPUs and integrated GPUs using low-level Linux tools and custom C/C++ workloads.

Overview

This project analyzes instruction throughput, caching behavior, thermal limits, and power efficiency across various workloads and system configurations on Intel hardware.

Target System

Samsung Galaxy Book 5 Pro

• Processor: Intel® Core™ Ultra 7 258V @ 2.20 GHz
• Memory: 32.0 GB RAM

Project Structure

• benchmarks/ - C/C++ source files for CPU and GPU benchmarks
• scripts/ - Bash/Python automation for running tests and processing results
• data/ - Raw log files from benchmark runs
• results/ - Processed CSVs and visualization plots
• docs/ - Design notes and methodology documentation

Key Features

• Custom C/C++ benchmarks for evaluating low-level CPU/GPU behavior
• Real-time profiling with tools like perf, htop, vmstat, intel_gpu_top, and RAPL
• Logging of cycles, instructions, thermal behavior, and power usage
• Scripts for automated profiling and data collection

Tools & Technologies

• Languages: C, C++, Python, Bash
• Linux Tools:
  • perf - Performance counter statistics
  • mpstat - CPU utilization reporting
  • vmstat - Virtual memory statistics
  • turbostat - Intel CPU frequency and power data
  • intel_gpu_top - Intel GPU utilization monitoring
  • RAPL - Running Average Power Limit interface
• Visualization: Python + Matplotlib for analysis and reporting

Getting Started

1. Clone this repository
2. Build the benchmarks with cd benchmarks && make
3. Run benchmarks with scripts/run_bench.sh
4. Process results with scripts/parse_logs.py
5. Generate plots with scripts/plot_results.py

License

This project is licensed under the MIT License - see the LICENSE file for details.