Driving the Future: Semiconductor Innovation, AI, and the Rise of RISC-V
Talkat RISC-V Summit North America 2024 · on Tuesday, 2024-10-22· at 16:00· Expo Hall - Exhibit Hall A - Demo Theater·
summary
Samsung Foundry maps AI demand to concrete tech: GPT‑4 = 1.8T parameters, model size grows ~10×/year while AI hardware improves ~2–3×, and compute rose ~990,000× versus memory bandwidth ~30×, creating a memory wall. Roadmap lists 7→5→4→3→2 nm with MBCFET/GAA at 3 nm and enhanced GAA at 2 nm, production 4 nm variants (S4/S4P/S4X/S4A), packaging work up to HBM 12‑stack in development and 2.5D/3D I‑Cube/XCube integration, plus custom HBM dies and cited customer platforms from 14 nm onward.
AI-written summary · about
related talks
58mThe Next Decade of Software Development
22mtheCUBE Research Mini Pod
21mAhead of the Plasticity Curve: AI, Biology and Technology
16mKeynote: Paving the Road to Datacenter-Scale RISC-V
34mThe Future of Computing: AI technologies in the exascale era
35mAI innovation begins with trust
01h 07mThe Next Decade of Software Development
57mPervasive and Sustainable AI with Adaptive Computing