Paving the Road Ahead: RISC-V and Chiplet Technologies in Modern Automotive & More
summary
Claims a RISC‑V 16‑core compute cluster with L1/L2/L3 caches and a coherent fabric that matches or beats incumbent single‑socket performance, delivered as a chiplet to cut custom SoC cost/time from ~$200–300M and 3–4 years to about $25M and <1 year. Uses low‑latency die‑to‑die links (UCIe/ODSA style) to make disaggregation transparent, supports modular IO/accelerator chiplets for zonal automotive “multiple servers on wheels” and already ships a first‑gen product with a second generation running.
AI-written summary · about