Skip to main content

Technical ReportMarch, 2016Published

The Memory Sluice Gate Theory: Have We Found a Solution for Memory Wall?

Authors
X.-H. Sun, Y.-H. Liu
Venue
Illinois Institute of Technology
Date
March, 2016
Type
Technical Report

Abstract

The memory-wall problem is a long standing issue facing the computing community. Many believe the memory-wall problem can only be solved with new memory technologies that improve memory device hardware performance. In this research, we propose an architectural solution, the memory Sluice Gate Theory, for solving the memory-wall problem. The focus of the Sluice Gate Theory is not on hardware peak performance, but the achieved memory stall time. In other words, the focus is not on removing memory wall, but on mitigating the impact of memory wall. In this theory, data access concurrency in addition to data access locality plays a vital role. Based on Sluice Gate Theory, a memory system is built to transfer data and to mask the performance gap between CPU and memory devices during the data transfer process. Sluice gates are designed to control data transfer at each memory layer (sluice stage) dynamically, and a global control algorithm, named layered performance matching, is developed to match the data transfer request/supply at each memory layer (sluice stage) thus matching the overall performance between the CPU and memory system. Formal theoretical analyses are given to show, with sufficient data access concurrency and dynamic hardware support, the memory wall impact can be fully removed. Experimental testing is conducted which confirm the theoretical findings. Sluice Gate Theory calls to fully investigate and utilize FPGA and memory concurrency technologies to eliminate the memory wall impact. Source note: The hosted file contains the report cover and opening page. The cover states: LIMITED DISTRIBUTION NOTICE: This report has been submitted for publication outside of IIT-SCS and will probably be copyrighted if accepted for publication. It has been issued as a Technical Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IIT-SCS prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g. payment of royalties).