🧮 Development Performance Optimization: 2D Multicore Scaling Map
Interactive 2D die map of a 64-core processor showing where parallel speedup is lost: lock contention, cache-coherency traffic, DRAM latency and memory bandwidth, using Gunther's Universal Scalability Law, an average-memory-access-time CPI model and a roofline bandwidth cap, with a live speedup-versus-cores graph.
⚙ Under the hood
Interactive 2D die map of a 64-core processor showing where parallel speedup is lost: lock contention, cache-coherency traffic, DRAM latency and memory bandwidth, using Gunther's Universal Scalability Law, an average-memory-access-time CPI model and a roofline bandwidth cap, with a live speedup-versus-cores graph.
parallel computingAmdahl's lawuniversal scalability lawcache hierarchyroofline modelperformance engineering
2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install