Sampo: Designing a 16-bit RISC CPU from Scratch - Part 1: Theory and Architecture
An in-depth exploration of Sampo, a custom 16-bit RISC processor that bridges the gap between classic Z80 workloads and modern RISC design principles.
TinyComputers.io
Sponsors
More Articles
Building a Browser-Based Z80 Emulator for the RetroShield
How I built a WebAssembly Z80 emulator to run RetroShield firmware in the browser, complete with CRT effects and multiple language interpreters.
The Machine Age, the Avant-Garde, and the New Cognitive Frontier
An exploration of the striking parallels between early twentieth century scientific management, the Vorticist avant-garde movement, and today's AI revolution, examining how the cultural logic of fragmentation, control, and mechanized progress echoes across a century of technological transformation.
Persistent Conversation Context Over Stateless Messaging APIs
A deep dive into building persistent multi-turn AI conversations over stateless messaging platforms like Signal—covering database schema design, conversation management, command parsing, and a novel inline image reference system for multi-modal context.
A Bespoke LLM Code Scanner
How I built a nightly AI code scanner using vLLM and Qwen2.5-Coder on AMD ROCm hardware that automatically reviews 50+ source files and creates JIRA tickets for bugs, security issues, and code quality problems—all running locally for about $15/month in electricity.
Building Cross-Platform Rust Binaries: A Multi-Architecture Build Orchestration System
A comprehensive guide to building a production-grade multi-platform build orchestration system that leverages diverse hardware—from single-board computers to powerful x86_64 servers—to build native Rust binaries for macOS, Linux, FreeBSD, NetBSD, and OpenBSD across ARM64 and x86_64 architectures. Includes architecture diagrams, performance comparisons, cost analysis, and the surprising discovery that emulated ARM64 on x86_64 outperforms native ARM64 hardware by 6x.
Rockchip RK3588 NPU Deep Dive: Real-World AI Performance Across Multiple Platforms
Comprehensive benchmarks and analysis of the Rockchip RK3588 NPU across multiple single-board computers. TinyLlama 1.1B achieves 10-15 tokens/second while ResNet18 hits 244 FPS at 4.09ms latency. Includes detailed conversion workflows, power efficiency analysis, platform comparisons, and real-world application recommendations for edge AI deployment on RK3588-powered devices.