Logo

회원가입로그인 ENGLISH naver youtube  
search 

IP명 An Energy-Efficient and High-Throughput Language AI Processor
Category Digital Application Language Processing
실설계면적 4㎛ X 4㎛ 공급 전압 1.8V
IP유형 Hard IP 동작속도 250 MHz
검증단계 Silicon 참여공정 SF28-2402
IP개요 This chip is designed for a language model based-on non-attention algorithm with homogeneous-core processor. Each core communicates with an adjacent core through 2D mesh Network-on-Chip and contains the fetch unit to control the various dataflow pattern.
- 레이아웃 사진 -