Automatically accelerating non-numerical programs by architecture-compiler co-design

Simone Campanoni; Kevin Brownell; Svilen Kanev; Gu Yeon Wei; David Brooks; Timothy M. Jones

doi:10.1145/3139461

Automatically accelerating non-numerical programs by architecture-compiler co-design

Simone Campanoni^*, Kevin Brownell, Svilen Kanev, Gu Yeon Wei, David Brooks, Timothy M. Jones

^*Corresponding author for this work

Computer Science

Research output: Contribution to journal › Review article › peer-review

1 Scopus citations

Abstract

Because of the high cost of communication between processors, compilers that parallelize loops automatically have been forced to skip a large class of loops that are both critical to performance and rich in latent parallelism. HELIX-RC is a compiler/microprocessor co-design that opens those loops to parallelization by decoupling communication from thread execution in conventional multicore architecures. Simulations of HELIX-RC, applied to a processor with 16 Intel Atom-like cores, show an average of 6.85× performance speedup for six SPEC CINT2000 benchmarks.

Original language	English (US)
Pages (from-to)	88-97
Number of pages	10
Journal	Communications of the ACM
Volume	60
Issue number	12
DOIs	https://doi.org/10.1145/3139461
State	Published - Dec 2017

ASJC Scopus subject areas

Computer Science(all)

Access to Document

10.1145/3139461

Cite this

@article{af6412d2e428447bb3b2107065a56b43,

title = "Automatically accelerating non-numerical programs by architecture-compiler co-design",

abstract = "Because of the high cost of communication between processors, compilers that parallelize loops automatically have been forced to skip a large class of loops that are both critical to performance and rich in latent parallelism. HELIX-RC is a compiler/microprocessor co-design that opens those loops to parallelization by decoupling communication from thread execution in conventional multicore architecures. Simulations of HELIX-RC, applied to a processor with 16 Intel Atom-like cores, show an average of 6.85× performance speedup for six SPEC CINT2000 benchmarks.",

author = "Simone Campanoni and Kevin Brownell and Svilen Kanev and Wei, {Gu Yeon} and David Brooks and Jones, {Timothy M.}",

year = "2017",

month = dec,

doi = "10.1145/3139461",

language = "English (US)",

volume = "60",

pages = "88--97",

journal = "Communications of the ACM",

issn = "0001-0782",

publisher = "Association for Computing Machinery (ACM)",

number = "12",

}

TY - JOUR

T1 - Automatically accelerating non-numerical programs by architecture-compiler co-design

AU - Campanoni, Simone

AU - Brownell, Kevin

AU - Kanev, Svilen

AU - Wei, Gu Yeon

AU - Brooks, David

AU - Jones, Timothy M.

PY - 2017/12

Y1 - 2017/12

N2 - Because of the high cost of communication between processors, compilers that parallelize loops automatically have been forced to skip a large class of loops that are both critical to performance and rich in latent parallelism. HELIX-RC is a compiler/microprocessor co-design that opens those loops to parallelization by decoupling communication from thread execution in conventional multicore architecures. Simulations of HELIX-RC, applied to a processor with 16 Intel Atom-like cores, show an average of 6.85× performance speedup for six SPEC CINT2000 benchmarks.

AB - Because of the high cost of communication between processors, compilers that parallelize loops automatically have been forced to skip a large class of loops that are both critical to performance and rich in latent parallelism. HELIX-RC is a compiler/microprocessor co-design that opens those loops to parallelization by decoupling communication from thread execution in conventional multicore architecures. Simulations of HELIX-RC, applied to a processor with 16 Intel Atom-like cores, show an average of 6.85× performance speedup for six SPEC CINT2000 benchmarks.

UR - http://www.scopus.com/inward/record.url?scp=85037027025&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85037027025&partnerID=8YFLogxK

U2 - 10.1145/3139461

DO - 10.1145/3139461

M3 - Review article

AN - SCOPUS:85037027025

SN - 0001-0782

VL - 60

SP - 88

EP - 97

JO - Communications of the ACM

JF - Communications of the ACM

IS - 12

ER -

Automatically accelerating non-numerical programs by architecture-compiler co-design

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this