WASP: Selective Data Prefetching with Monitoring Runtime Warp Progress on GPUs

Yunho Oh; Myung Kuk Yoon; Jong Hyun Park; Yongjun Park; Won Woo Ro

doi:10.1109/TC.2018.2813379

WASP: Selective Data Prefetching with Monitoring Runtime Warp Progress on GPUs

Yunho Oh, Myung Kuk Yoon, Jong Hyun Park, Yongjun Park, Won Woo Ro

Department of Computer Science & Engineering

Research output: Contribution to journal › Article › peer-review

5 Scopus citations

Abstract

This paper proposes a new data prefetching technique for Graphics Processing Units (GPUs) called Warp Aware Selective Prefetching (WASP). The main idea of WASP is to dynamically select warps whose progress is slower than that of the current warp as prefetching target warps. Under the in-order instruction execution model of GPUs, these prefetching target warps will certainly execute the same load as the current warp. Exploiting that, WASP prefetches the data for prefetching target warps, which allows the prefetched data to be accurately accessed. To simply verify the progress of the warps, WASP monitors the counts of the dynamic load executions for all warps. When a warp executes a load, WASP searches the warps with lower load execution counts than the current warp and generates the prefetch requests for them. In our evaluation, WASP achieves a 16.8 percent speedup compared to the baseline GPU.

Original language	English
Article number	8309426
Pages (from-to)	1366-1373
Number of pages	8
Journal	IEEE Transactions on Computers
Volume	67
Issue number	9
DOIs	https://doi.org/10.1109/TC.2018.2813379
State	Published - 1 Sep 2018

Bibliographical note

Funding Information:
This work was partly supported by Institute for Information & communications Technology Promotion (IITP) grant funded by the Korea government (MSIT) (No. R7117-16-0233, Development of Application Program Optimization Tools for High Performance Computing Systems) and Korea Evaluation Institute of Industrial Technology (KEIT) grant funded by the Korea government (MOTIE) (No.10080674, Development of Reconfigurable Artificial Neural Network Accelerator and Instruction Set Architecture).

Publisher Copyright:
© 2012 IEEE.

Keywords

cache performance
data prefetching
GPGPU
warp scheduling

Access to Document

10.1109/TC.2018.2813379

Cite this

@article{65ce778fe082452b8e46e77e8cab18d9,

title = "WASP: Selective Data Prefetching with Monitoring Runtime Warp Progress on GPUs",

abstract = "This paper proposes a new data prefetching technique for Graphics Processing Units (GPUs) called Warp Aware Selective Prefetching (WASP). The main idea of WASP is to dynamically select warps whose progress is slower than that of the current warp as prefetching target warps. Under the in-order instruction execution model of GPUs, these prefetching target warps will certainly execute the same load as the current warp. Exploiting that, WASP prefetches the data for prefetching target warps, which allows the prefetched data to be accurately accessed. To simply verify the progress of the warps, WASP monitors the counts of the dynamic load executions for all warps. When a warp executes a load, WASP searches the warps with lower load execution counts than the current warp and generates the prefetch requests for them. In our evaluation, WASP achieves a 16.8 percent speedup compared to the baseline GPU.",

keywords = "cache performance, data prefetching, GPGPU, warp scheduling",

author = "Yunho Oh and Yoon, {Myung Kuk} and Park, {Jong Hyun} and Yongjun Park and Ro, {Won Woo}",

note = "Funding Information: This work was partly supported by Institute for Information & communications Technology Promotion (IITP) grant funded by the Korea government (MSIT) (No. R7117-16-0233, Development of Application Program Optimization Tools for High Performance Computing Systems) and Korea Evaluation Institute of Industrial Technology (KEIT) grant funded by the Korea government (MOTIE) (No.10080674, Development of Reconfigurable Artificial Neural Network Accelerator and Instruction Set Architecture). Publisher Copyright: {\textcopyright} 2012 IEEE.",

year = "2018",

month = sep,

day = "1",

doi = "10.1109/TC.2018.2813379",

language = "English",

volume = "67",

pages = "1366--1373",

journal = "IEEE Transactions on Computers",

issn = "0018-9340",

publisher = "IEEE Computer Society",

number = "9",

}

TY - JOUR

T1 - WASP

T2 - Selective Data Prefetching with Monitoring Runtime Warp Progress on GPUs

AU - Oh, Yunho

AU - Yoon, Myung Kuk

AU - Park, Jong Hyun

AU - Park, Yongjun

AU - Ro, Won Woo

N1 - Funding Information: This work was partly supported by Institute for Information & communications Technology Promotion (IITP) grant funded by the Korea government (MSIT) (No. R7117-16-0233, Development of Application Program Optimization Tools for High Performance Computing Systems) and Korea Evaluation Institute of Industrial Technology (KEIT) grant funded by the Korea government (MOTIE) (No.10080674, Development of Reconfigurable Artificial Neural Network Accelerator and Instruction Set Architecture). Publisher Copyright: © 2012 IEEE.

PY - 2018/9/1

Y1 - 2018/9/1

N2 - This paper proposes a new data prefetching technique for Graphics Processing Units (GPUs) called Warp Aware Selective Prefetching (WASP). The main idea of WASP is to dynamically select warps whose progress is slower than that of the current warp as prefetching target warps. Under the in-order instruction execution model of GPUs, these prefetching target warps will certainly execute the same load as the current warp. Exploiting that, WASP prefetches the data for prefetching target warps, which allows the prefetched data to be accurately accessed. To simply verify the progress of the warps, WASP monitors the counts of the dynamic load executions for all warps. When a warp executes a load, WASP searches the warps with lower load execution counts than the current warp and generates the prefetch requests for them. In our evaluation, WASP achieves a 16.8 percent speedup compared to the baseline GPU.

AB - This paper proposes a new data prefetching technique for Graphics Processing Units (GPUs) called Warp Aware Selective Prefetching (WASP). The main idea of WASP is to dynamically select warps whose progress is slower than that of the current warp as prefetching target warps. Under the in-order instruction execution model of GPUs, these prefetching target warps will certainly execute the same load as the current warp. Exploiting that, WASP prefetches the data for prefetching target warps, which allows the prefetched data to be accurately accessed. To simply verify the progress of the warps, WASP monitors the counts of the dynamic load executions for all warps. When a warp executes a load, WASP searches the warps with lower load execution counts than the current warp and generates the prefetch requests for them. In our evaluation, WASP achieves a 16.8 percent speedup compared to the baseline GPU.

KW - cache performance

KW - data prefetching

KW - GPGPU

KW - warp scheduling

UR - http://www.scopus.com/inward/record.url?scp=85043367821&partnerID=8YFLogxK

U2 - 10.1109/TC.2018.2813379

DO - 10.1109/TC.2018.2813379

M3 - Article

AN - SCOPUS:85043367821

SN - 0018-9340

VL - 67

SP - 1366

EP - 1373

JO - IEEE Transactions on Computers

JF - IEEE Transactions on Computers

IS - 9

M1 - 8309426

ER -

WASP: Selective Data Prefetching with Monitoring Runtime Warp Progress on GPUs

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this