An interactive visualization of computer latency at different scales, updated for 2025 hardware. This project provides an intuitive understanding of the relative speed of various computer operations.
Visit the live site: https://martinkalema.github.io/latency-numbers
- Interactive Visualization: Hover over metrics to see relative comparisons
- Multiple Scale Views:
- Linear scale for absolute comparisons
- Logarithmic scale for better visibility across orders of magnitude
- Relative scale comparing everything to L1 cache speed
- Modern Design: Clean, responsive interface with smooth animations
- Data Sources: Compiled from trusted sources including Google SRE, ByteByteGo, Thundergolfer, and Napkin Math
The visualization includes key latency metrics from nanoseconds to seconds:
- Cache Operations (1-7 ns): L1 and L2 cache references
- Memory Operations (100-250,000 ns): RAM access and sequential reads
- Storage Operations (100 μs - 20 ms): SSD and HDD operations
- Network Operations (1.6 μs - 150 ms): Datacenter and WAN round trips
The latency numbers are compiled from multiple authoritative sources:
- D3.js v7: For data visualization
- Pure JavaScript: No framework dependencies
- GitHub Actions: Automated deployment to GitHub Pages
- Responsive CSS: Mobile-friendly design with CSS Grid and Flexbox
Inspired by Colin Scott's original latency calculator and Jeff Dean's "Numbers Everyone Should Know" talk.
MIT License - feel free to use this visualization in your own projects!