| Mobile Web

Search results for KV cache

AI & Enterprise

Xcena joins OCP Global Summit 2026, showcases memory infrastructure for AI inference

Fabless semiconductor company Xcena, which develops next-generation memory infrastructure, took part in OCP Global Summit 2026 held in San Jose, California from Oct. 12 to 15, it said on Thursday. The company showcased memory infrastructure technology aimed at efficient AI inference. It presented two AI inference demos using its CXL-based memory solution MX1, including sharing KV cache across servers via a CXL memory pool and accelerating vector search using near-data processing.

Industry

Chips&Media unveils data compression IP to cut bandwidth up to 53.5 percent

Chips&Media unveiled a data compression IP designed to ease memory bottlenecks in NPUs and said it will target the AI edge and server markets. The company said the multimedia IP, its first aimed beyond video, compresses and restores intermediate feature-map data in real time without loss, preserving inference accuracy. In tests of major vision networks, it said memory bandwidth savings were 35.2 percent for FP16, 45.7 percent for BF16 and 53.5 percent for INT8.

Industry

Snapdragon Summit 2026: Smartphones can run 30 billion-parameter AI models

On-device AI models, previously constrained by smartphone memory capacity, can grow to 30 billion parameters through data placement design, Qualcomm said. At Snapdragon Summit 2026, Qualcomm executive vice president Chris Patrick said more capable AI agents need access to larger models with stronger reasoning and richer context. Qualcomm unveiled the Snapdragon 8 Elite 6th Gen platform, expanding Hexagon NPU shared memory and adding an element accelerator supporting 32,000 tokens.