课题基金 / 基金详情

Experimental Partnership - FlexRAM: An Advanced Intelligent Memory System

Experimental Partnership - FlexRAM: An Advanced Intelligent Memory System
实验性合作伙伴关系 - FlexRAM:先进的智能内存系统
批准号:
0072102
负责人:
Josep Torrellas
金额:
$0.0万
依托单位国家:
美国
项目类别:
Standard Grant
财政年份:
2000
资助国家:
美国
项目状态:
已结题
起止时间:
2000-09-15 至 2006-09-30

项目摘要

项目成果

Josep Torrellas的其他基金

相似基金

相关文献

中文摘要
翻译
eia - 0072102约瑟TorrellasUniv。题目:实验合作伙伴关系:FlexRAM:一种先进的智能存储系统可以集成在单个芯片上的晶体管数量的急剧增加使得微处理器性能和存储芯片容量都有了惊人的提高。然而,它们也导致处理器和存储系统之间的数据传输瓶颈日益受限。认识到需要新的架构方法来缓解这一瓶颈,研究人员提出将处理器和DRAM集成到单个芯片中。这种架构通常被称为智能内存(IRAM)或内存中处理器(PIM)。不幸的是,将当前的微处理器和DRAM简单地集成在一个芯片上,往往只能带来有限的性能改进——它在芯片上移动了芯片外的瓶颈。此外,生成的系统通常很难编程。最后,很少投入精力来尝试确定能够有效利用此体系结构的广泛应用程序。
英文摘要
EIA-0072102Josep TorrellasUniv. of Illinois-ChampaignTITLE: Experimental Partnership: FlexRAM: An Advanced Intelligent Memory SystemDramatic increases in the number of transistors that can be integrated on a single chip have enabled both microprocessor performance and memory chip capacity to rise spectacularly. However, they have also led to an increasingly constraining data transfer bottleneck between processor and memory system. Recognizing the need for new architectural approaches that alleviate this bottleneck, researchers have proposed the integration of processor and DRAM in a single chip. This architecture is popularly known as intelligent memory (IRAM) or processor-in-memory (PIM).Unfortunately, simple integration of current microprocessors and DRAM on a single chip often delivers only modest performance improvements-it moves an off-chip bottleneck on chip. Furthermore, the resulting system is often very hard to program. Finally, Little effort has been invested trying to identify a wide range of applications that can exploit this architecture effectively.
期刊论文(0)
专著(0)
科研奖励(0)
会议论文
Collaborative Research: PPoSS: LARGE: General-Purpose Scalable Technologies for Fundamental Graph Problems
SHF: Medium: Cross-Cutting Effort to Make Non-Volatile Memories Truly Usable
PPoSS: Planning: A Cross-Layer Approach to Accelerate Large-Scale Graph Computations on Distributed Platforms
CNS Core: Medium: Rethinking Architecture and Operating Systems for Modern Virtualization Technologies
海外基金