Memory Intensive Computing

نویسندگان

  • Shahar Kvatinsky
  • Avinoam Kolodny
  • Uri C. Weiser
  • Eby G. Friedman
چکیده

Over the past years, new memory technologies such as RRAM, STT-MRAM, and PCM have emerged. These technologies employ devices located within the metal layers of the chip, which are relatively fast, dense, and power efficient and can be considered as 'memristors'. To date, research in these devices has primarily focused on memristors as flash, DRAM, and SRAM replacement. In this presentation, we present these emerging memory technologies as enablers to the era of memoryintensive computing, which brings interesting opportunities for novel architectural applications. As an example, we present the multistate pipeline register (MPR), a structure that stores the microarchitectural state of multiple threads. We show that MPR can eliminate the need to flush the pipeline upon a thread switch in Switch-on-Event (SoE) multi-threading machines. We call the new microarchitectural scheme, Continuous Flow MultiThreading (CFMT), and compare the performance and power consumption against traditional SoE machines. Memristor-based CFMT exhibits an average performance improvement of 40%, while reducing power consumption by 6.5%, significantly increasing the performance to energy ratio.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Memory-Intensive Benchmarks: IRAM vs. Cache-Based Machines

The increasing gap between processor and memory performance has led to new architectural models for memory-intensive applications. In this paper, we use a set of memory-intensive benchmarks to evaluate a mixed logic and DRAM processor called VIRAM as a building block for scientific computing. For each benchmark, we explore the fundamental hardware requirements of the problem as well as alternat...

متن کامل

HiDISC: A Decoupled Architecture for Applications in Data Intensive Computing

The ever growing speed gap between processor and main memory has been a major performance bottleneck of modern computer systems. As a result, today’s data intensive applications suffer from frequent cache misses and lose many CPU cycles due to pipeline stalling. Although traditional prefetching methods reduce cache misses considerably, most of them strongly depend on the access pattern being pr...

متن کامل

Enabling NVM for Data-Intensive Scientific Services

Specialized, transient data services are playing an increasingly prominent role in data-intensive scientific computing. These services offer flexible, on-demand pairing of applications with storage hardware using semantics that are optimized for the problem domain. Concurrent with this trend, upcoming scientific computing and big data systems will be deployed with emerging non-volatile memory (...

متن کامل

Data Replication-Based Scheduling in Cloud Computing Environment

Abstract— High-performance computing and vast storage are two key factors required for executing data-intensive applications. In comparison with traditional distributed systems like data grid, cloud computing provides these factors in a more affordable, scalable and elastic platform. Furthermore, accessing data files is critical for performing such applications. Sometimes accessing data becomes...

متن کامل

A parallel computing approach to viewshed analysis of large terrain data using graphics processing units

Viewshed analysis, often supported by Geographic Information Systems (GIS), is widely used in many application domains. However, as terrain data continue to become increasingly large and available at high resolutions, data-intensive viewshed analysis poses significant computational challenges. General-Purpose computation on Graphics Processing Units (GPGPU) provides a promising means to address...

متن کامل

The Characterization of Data Intensive Memory Workloads on Distributed PIM Systems

Processing-In-Memory (PIM) circumvents the von Neumann bottleneck by combining logic and memory (typically DRAM) on a single die. This work examines the memory system parameters for constructing PIM based parallel computers which are capable of meeting the memory access demands of complex programs that exhibit low reuse and non uniform stride accesses. The analysis uses the Data Intensive Syste...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2014