Adaptive correlated prefetch with large-scale hybrid memory system for stream processing

Sung Min Lee, Su Kyung Yoon, Jeong Geun Kim, Shin-Dug Kim

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Owing to the exponential growth of real-time data generation, the importance of stream processing is ever increasing. However, the data processing paradigm of stream processing is quite different, so it is difficult to expect high performance from memory systems applied to existing data centers. To solve this problem, two main solutions are suggested in this paper. First, a hybrid main memory and small buffer architecture are designed to reflect the execution characteristics of stream processing. Second, a hardware-based prefetch module supports correlation prefetching. Stream processing tends to accept incoming data in the main memory, so the prefetch module is used to divert data from the main memory layer to the buffer layer based on an intelligent clustering algorithm. This clustering algorithm affects the rapidly changing data access pattern of stream processing applications. By using heterogeneous main memories, not only can one enjoy the fast access latency of DRAM but also its nonvolatility, scalability, and low power consumption. The proposed hybrid memory architecture with our prefetch buffer structure can improve the buffer hit rate by 9–14% over other prefetch methods, reduce energy consumption by 26% over the conventional DRAM-only model, and achieve similar execution time over the 1/8-size DRAM space of the DRAM-only model.

Original languageEnglish
Pages (from-to)4746-4770
Number of pages25
JournalJournal of Supercomputing
Volume74
Issue number9
DOIs
Publication statusPublished - 2018 Sep 1

Fingerprint

Stream Processing
Dynamic random access storage
Data storage equipment
Buffer
Processing
Clustering algorithms
Clustering Algorithm
Memory architecture
Buffer layers
Module
Prefetching
Data Center
Exponential Growth
Scalability
Hits
Electric power utilization
Energy utilization
Execution Time
Power Consumption
Energy Consumption

All Science Journal Classification (ASJC) codes

  • Software
  • Theoretical Computer Science
  • Information Systems
  • Hardware and Architecture

Cite this

Lee, Sung Min ; Yoon, Su Kyung ; Kim, Jeong Geun ; Kim, Shin-Dug. / Adaptive correlated prefetch with large-scale hybrid memory system for stream processing. In: Journal of Supercomputing. 2018 ; Vol. 74, No. 9. pp. 4746-4770.
@article{959d3037acd1476c9dabe9a1b0a70a3a,
title = "Adaptive correlated prefetch with large-scale hybrid memory system for stream processing",
abstract = "Owing to the exponential growth of real-time data generation, the importance of stream processing is ever increasing. However, the data processing paradigm of stream processing is quite different, so it is difficult to expect high performance from memory systems applied to existing data centers. To solve this problem, two main solutions are suggested in this paper. First, a hybrid main memory and small buffer architecture are designed to reflect the execution characteristics of stream processing. Second, a hardware-based prefetch module supports correlation prefetching. Stream processing tends to accept incoming data in the main memory, so the prefetch module is used to divert data from the main memory layer to the buffer layer based on an intelligent clustering algorithm. This clustering algorithm affects the rapidly changing data access pattern of stream processing applications. By using heterogeneous main memories, not only can one enjoy the fast access latency of DRAM but also its nonvolatility, scalability, and low power consumption. The proposed hybrid memory architecture with our prefetch buffer structure can improve the buffer hit rate by 9–14{\%} over other prefetch methods, reduce energy consumption by 26{\%} over the conventional DRAM-only model, and achieve similar execution time over the 1/8-size DRAM space of the DRAM-only model.",
author = "Lee, {Sung Min} and Yoon, {Su Kyung} and Kim, {Jeong Geun} and Shin-Dug Kim",
year = "2018",
month = "9",
day = "1",
doi = "10.1007/s11227-018-2466-7",
language = "English",
volume = "74",
pages = "4746--4770",
journal = "Journal of Supercomputing",
issn = "0920-8542",
publisher = "Springer Netherlands",
number = "9",

}

Adaptive correlated prefetch with large-scale hybrid memory system for stream processing. / Lee, Sung Min; Yoon, Su Kyung; Kim, Jeong Geun; Kim, Shin-Dug.

In: Journal of Supercomputing, Vol. 74, No. 9, 01.09.2018, p. 4746-4770.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Adaptive correlated prefetch with large-scale hybrid memory system for stream processing

AU - Lee, Sung Min

AU - Yoon, Su Kyung

AU - Kim, Jeong Geun

AU - Kim, Shin-Dug

PY - 2018/9/1

Y1 - 2018/9/1

N2 - Owing to the exponential growth of real-time data generation, the importance of stream processing is ever increasing. However, the data processing paradigm of stream processing is quite different, so it is difficult to expect high performance from memory systems applied to existing data centers. To solve this problem, two main solutions are suggested in this paper. First, a hybrid main memory and small buffer architecture are designed to reflect the execution characteristics of stream processing. Second, a hardware-based prefetch module supports correlation prefetching. Stream processing tends to accept incoming data in the main memory, so the prefetch module is used to divert data from the main memory layer to the buffer layer based on an intelligent clustering algorithm. This clustering algorithm affects the rapidly changing data access pattern of stream processing applications. By using heterogeneous main memories, not only can one enjoy the fast access latency of DRAM but also its nonvolatility, scalability, and low power consumption. The proposed hybrid memory architecture with our prefetch buffer structure can improve the buffer hit rate by 9–14% over other prefetch methods, reduce energy consumption by 26% over the conventional DRAM-only model, and achieve similar execution time over the 1/8-size DRAM space of the DRAM-only model.

AB - Owing to the exponential growth of real-time data generation, the importance of stream processing is ever increasing. However, the data processing paradigm of stream processing is quite different, so it is difficult to expect high performance from memory systems applied to existing data centers. To solve this problem, two main solutions are suggested in this paper. First, a hybrid main memory and small buffer architecture are designed to reflect the execution characteristics of stream processing. Second, a hardware-based prefetch module supports correlation prefetching. Stream processing tends to accept incoming data in the main memory, so the prefetch module is used to divert data from the main memory layer to the buffer layer based on an intelligent clustering algorithm. This clustering algorithm affects the rapidly changing data access pattern of stream processing applications. By using heterogeneous main memories, not only can one enjoy the fast access latency of DRAM but also its nonvolatility, scalability, and low power consumption. The proposed hybrid memory architecture with our prefetch buffer structure can improve the buffer hit rate by 9–14% over other prefetch methods, reduce energy consumption by 26% over the conventional DRAM-only model, and achieve similar execution time over the 1/8-size DRAM space of the DRAM-only model.

UR - http://www.scopus.com/inward/record.url?scp=85049007792&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049007792&partnerID=8YFLogxK

U2 - 10.1007/s11227-018-2466-7

DO - 10.1007/s11227-018-2466-7

M3 - Article

AN - SCOPUS:85049007792

VL - 74

SP - 4746

EP - 4770

JO - Journal of Supercomputing

JF - Journal of Supercomputing

SN - 0920-8542

IS - 9

ER -