Tags: high‑bandwidth memory

DeepSeek, in partnership with Peking University, unveiled Engram, a new training method that separates static memory from computation in large language models. By using hashed N‑gram lookups and a context‑aware gating mechanism, Engram reduces reliance on high‑bandwidth memory (HBM), allowing models to operate efficiently on standard GPU memory while scaling parameter counts. Tests on a 27‑billion‑parameter model showed measurable gains across industry benchmarks, and the approach integrates with existing hardware solutions such as Phison’s SSD‑based accelerators and emerging CXL standards. Engram could ease pressure on costly memory hardware and stabilize DRAM price volatility. Ler mais

OpenAI has signed letters of intent with Samsung Electronics and SK Hynix to supply high‑bandwidth memory DRAM wafers for its Stargate AI infrastructure. The agreements, reached after a meeting in Seoul with South Korean leadership, call for scaling production to up to 900,000 chips per month, more than doubling current industry capacity. Both chipmakers will also integrate OpenAI’s APIs and ChatGPT Enterprise into their operations, supporting the broader push to expand AI compute capacity through new data centers in South Korea and beyond. Ler mais