TAILIEUCHUNG - FlexRAM: Toward an Advanced Intelligent Memory System A Retrospective Paper

FlexRAM: Toward an Advanced Intelligent Memory System A Retrospective Paper lead to our ICCD-1999 FlexRAM paper started in 1996. At that time, there was great interest in potential of integrating compute capabilities in large DRAM memories — an architecture called processing in memory (PIM) or intelligent memory. | FlexRAM Toward an Advanced Intelligent Memory System A Retrospective Paper Josep Torrellas Department of Computer Science University of Illinois torrellas@ http I. Context of the Work The work that lead to our ICCD-1999 FlexRAM paper 4 started in 1996. At that time there was great interest in the potential of integrating compute capabilities in large DRAM memories an architecture called Processing-In-Memory PIM or Intelligent Memory. Prof. Kogge at the University of Notre Dame had been an early and persistent proponent of the technology since his EXECUBE work 6 . Prof. Patterson at UC Berkeley had been leading the Berkeley IRAM project 11 and co-organized a workshop on these architectures in June 1997 12 . In addition Dr. Lucas from DARPA was outlining plans for an effort in this area. Finally some chip manufacturers were investing in a DRAM technology that could be compatible with high-speed logic . IBM s CMOS 7LD and Mitsubishi s eRaM. At Illinois my research group was trying to understand what a PIM computer architecture might entail. Our initial emphasis was on understanding the needs of the driving applications for this technology. In May 1998 I started a sabbatical at IBM . Watson Research Center which ended up being very useful for this project. My host Dr. Pattnaik also a co-author of the ICCD-1999 paper joined the project and put me in touch with IBM experts on DRAM technology and potential applications for PIM systems. II. DEVELOPING the Ideas We proposed a design that places many tiny cores in the memory system of an otherwise commodity machine. The main processor s of the machine is left unmodified since it works well for legacy applications. The memory cores are small to minimize losses in memory integration and numerous to extract high bandwidth. Moreover they are general-purpose and not designed to pattern-match any particular algorithm. We also detailed a possible layout of the intelligent memory chips for 180nm.

TAILIEUCHUNG - Chia sẻ tài liệu không giới hạn
Địa chỉ : 444 Hoang Hoa Tham, Hanoi, Viet Nam
Website : tailieuchung.com
Email : tailieuchung20@gmail.com
Tailieuchung.com là thư viện tài liệu trực tuyến, nơi chia sẽ trao đổi hàng triệu tài liệu như luận văn đồ án, sách, giáo trình, đề thi.
Chúng tôi không chịu trách nhiệm liên quan đến các vấn đề bản quyền nội dung tài liệu được thành viên tự nguyện đăng tải lên, nếu phát hiện thấy tài liệu xấu hoặc tài liệu có bản quyền xin hãy email cho chúng tôi.
Đã phát hiện trình chặn quảng cáo AdBlock
Trang web này phụ thuộc vào doanh thu từ số lần hiển thị quảng cáo để tồn tại. Vui lòng tắt trình chặn quảng cáo của bạn hoặc tạm dừng tính năng chặn quảng cáo cho trang web này.