CPU Stress Tester

A Python-based stress testing tool for both CPUs and GPUs with real-time performance monitoring. Test your CPU with intensive calculations showing MFLOPS/GFLOPS, or stress test your GPU with matrix operations displaying TFLOPS.

Features

• Multi-core CPU support: Utilize all or a specific number of CPU cores
• GPU support: CUDA-based GPU stress testing with PyTorch or CuPy
• Configurable duration: Run tests for any length of time or continuously
• Real-time FLOPS monitoring: See performance metrics updated every second
• Colorized output: Easy-to-read color-coded performance data
• Continuous mode: Run indefinitely until manually stopped
• Cross-platform: Works on Windows, Linux, and macOS
• Minimal dependencies: CPU mode uses only Python standard library

Requirements

For CPU testing:

• Python 3.6 or higher

For GPU testing:

• Python 3.6 or higher
• NVIDIA GPU with CUDA support
• One of the following:
  • PyTorch with CUDA: pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
  • CuPy: pip install cupy-cuda11x (replace 11x with your CUDA version)

Usage

Basic Usage

CPU Testing (default):

python cpu_stress_test.py

GPU Testing:

python cpu_stress_test.py --mode gpu -d 60

Continuous mode (run until Ctrl+C):

python cpu_stress_test.py -d 0

Command-line Arguments

• -c, --cores <number>: Number of CPU cores to use (default: all available) - CPU mode only
• -d, --duration <seconds>: Duration in seconds to run the test (default: 60, use 0 for continuous)
• -v, --verbose: Show per-core performance breakdown (CPU) or detailed GPU info
• --mode <cpu|gpu>: Test mode - 'cpu' for CPU testing (default) or 'gpu' for GPU testing
• -h, --help: Show help message

Examples

Test all cores for 1 minute (default):

python cpu_stress_test.py

Extended stress test (10 minutes):

python cpu_stress_test.py -d 600

Verbose mode with per-core breakdown:

python cpu_stress_test.py -v -d 30

How It Works

CPU Mode

The stress tester spawns multiple worker processes (one per CPU core by default) that perform intensive mathematical calculations including:

• Square root and power operations
• Factorial calculations
• List comprehensions with exponential operations

Each worker runs continuously for the specified duration (or indefinitely in continuous mode), reporting real-time FLOPS (Floating Point Operations Per Second) performance metrics. The display shows:

• Current FLOPS/s: Instantaneous performance rate
• Cumulative FLOPS: Total operations performed
• Color-coded metrics: Performance scaled automatically (FLOPS → KFLOPS → MFLOPS → GFLOPS)

GPU Mode

The GPU stress tester performs large matrix multiplications (4096x4096) using CUDA:

• Uses PyTorch or CuPy for GPU acceleration
• Each iteration performs 2*N³ floating-point operations
• Reports performance in TFLOPS (Teraflops)
• Fully utilizes GPU compute capabilities
• Monitors GPU temperature and utilization (with verbose mode)

Stopping the Test

Press Ctrl+C to interrupt the test at any time.

Use Cases

• Thermal testing: Check CPU/GPU cooling performance under load
• Stability testing: Verify system stability after overclocking
• Performance benchmarking: Compare CPU/GPU performance across systems
• Burn-in testing: Stress test new hardware
• Power consumption testing: Monitor power draw under full load

Warning

This tool will push your CPU/GPU to 100% utilization. Monitor your system temperatures and ensure adequate cooling. Not recommended for extended periods on laptops or systems with insufficient cooling.

Notes

This project uses some Vibe Coded attributes (mainly math as I suck at math) and I cannot test the GPU option as I have integrated graphics.