In-Memory Computing: From General Purpose to Custom Processing
페이지 정보
작성자 최고관리자 댓글 조회 작성일 24-01-17 09:44본문
- 강연자
- Muhammad Imran
BK21 지능형반도체 IT 융합 혁신인재 양성사업단 세미나
제목 : In-Memory Computing: From General Purpose to Custom Processing
연사 : Muhammad Imran
일시 : 2023년 12월 19일 화요일 13시
장소 : 제4공학관 D510호
내용 : Conventional Von Neuman computing architectures have undergone significant optimizations, incorporating various methods such as pipelining, out-of-order execution and various forms of parallel processing, in order to achieve high performance. The major challenge today lies in the efficient retrieval and storage of data from memory. This becomes particularly crucial with the rise of data-intensive and computationally demanding AI workloads. Memory access is costly both in terms of latency as well as energy consumption. In-Memory Computing is an out-of-box solution to address this limitation by performing processing tasks directly within the memory, minimizing the need for data transfer.
제목 : In-Memory Computing: From General Purpose to Custom Processing
연사 : Muhammad Imran
일시 : 2023년 12월 19일 화요일 13시
장소 : 제4공학관 D510호
내용 : Conventional Von Neuman computing architectures have undergone significant optimizations, incorporating various methods such as pipelining, out-of-order execution and various forms of parallel processing, in order to achieve high performance. The major challenge today lies in the efficient retrieval and storage of data from memory. This becomes particularly crucial with the rise of data-intensive and computationally demanding AI workloads. Memory access is costly both in terms of latency as well as energy consumption. In-Memory Computing is an out-of-box solution to address this limitation by performing processing tasks directly within the memory, minimizing the need for data transfer.
