Interactive dashboard tracking 40 years of GPU evolution (1986-2025) across 2,824 GPUs.
- 17 interactive charts across Technology, Performance, and Financial tabs
- Color By toggle - switch all scatter plots between Vendor and Foundry coloring
- Die Size filter - zoom into specific die size ranges (>800mm2, 400-600mm2, etc.)
- Vendor/Foundry filter - isolate specific vendors or foundries
- Plotly.js hover tooltips with full GPU specs on every data point
| Source | Data |
|---|---|
| RightNow GPU Database | 2,824 GPUs - specs, foundry, process node, die size, transistors |
| FRED | US Semiconductor Production Index |
| yfinance | NVDA, AMD, INTC stock performance |
| SemiAnalysis | H100 GPU Price Index |
- Moore's Law - Transistor Count (log scale, with 2x/2yr reference line)
- Process Node Migration (stacked area)
- Process Node Distribution by Vendor and by Foundry
- Foundry Market Share over time
- Die Size Trends (scatter, dual color)
- Die Size Distribution by Vendor and by Foundry (box plots)
- Memory Type Evolution (GDDR3 to GDDR7, HBM adoption)
- FP32 Compute Scaling (TFLOPS, log scale)
- Compute Efficiency (TFLOPS per Watt)
- GPU Launch Cadence by Vendor
- Clock Speed Evolution
- Memory Bandwidth Scaling (HBM marked with diamonds)
- GPU Stock Performance (NVDA/AMD/INTC indexed to 100, with launch event markers)
- H100 GPU Price Index
- US Semiconductor Production Index (with recession shading)
pip install pandas numpy plotly requests python-dotenv yfinance
python build_dashboard.pyGenerates index.html (single self-contained file, ~2MB). Open in any browser or deploy to GitHub Pages.
- Python 3.13 build script
- Plotly.js 2.35.2 (CDN, client-side rendering)
- Static HTML - no server required