In the construction of exascale computing systems energy efficiency and power
consumption are two of the major challenges. Low-power high performance
embedded systems are of increasing interest as building blocks for large scale
high- performance systems. However, extracting maximum performance out of such
systems presents many challenges. Various aspects from the hardware
architecture to the programming models used need to be explored. The Epiphany
architecture integrates low-power RISC cores on a 2D mesh network and promises
up to 70 GFLOPS/Watt of processing efficiency. However, with just 32 KB of
memory per eCore for storing both data and code, and only low level inter-core
communication support, programming the Epiphany system presents several
challenges. In this paper we evaluate the performance of the Epiphany system
for a variety of basic compute and communication operations. Guided by this
data we explore strategies for implementing scientific applications on memory
constrained low-powered devices such as the Epiphany. With future systems
expected to house thousands of cores in a single chip, the merits of such
architectures as a path to exascale is compared to other competing systems.
%0 Generic
%1 Varghese2014Programming
%A Varghese, Anish
%A Edwards, Bob
%A Mitra, Gaurav
%A Rendell, Alistair P.
%D 2014
%K hpc
%T Programming the Adapteva Epiphany 64-core Network-on-chip Coprocessor
%U http://arxiv.org/abs/1410.8772
%X In the construction of exascale computing systems energy efficiency and power
consumption are two of the major challenges. Low-power high performance
embedded systems are of increasing interest as building blocks for large scale
high- performance systems. However, extracting maximum performance out of such
systems presents many challenges. Various aspects from the hardware
architecture to the programming models used need to be explored. The Epiphany
architecture integrates low-power RISC cores on a 2D mesh network and promises
up to 70 GFLOPS/Watt of processing efficiency. However, with just 32 KB of
memory per eCore for storing both data and code, and only low level inter-core
communication support, programming the Epiphany system presents several
challenges. In this paper we evaluate the performance of the Epiphany system
for a variety of basic compute and communication operations. Guided by this
data we explore strategies for implementing scientific applications on memory
constrained low-powered devices such as the Epiphany. With future systems
expected to house thousands of cores in a single chip, the merits of such
architectures as a path to exascale is compared to other competing systems.
@misc{Varghese2014Programming,
abstract = {{In the construction of exascale computing systems energy efficiency and power
consumption are two of the major challenges. Low-power high performance
embedded systems are of increasing interest as building blocks for large scale
high- performance systems. However, extracting maximum performance out of such
systems presents many challenges. Various aspects from the hardware
architecture to the programming models used need to be explored. The Epiphany
architecture integrates low-power RISC cores on a 2D mesh network and promises
up to 70 GFLOPS/Watt of processing efficiency. However, with just 32 KB of
memory per eCore for storing both data and code, and only low level inter-core
communication support, programming the Epiphany system presents several
challenges. In this paper we evaluate the performance of the Epiphany system
for a variety of basic compute and communication operations. Guided by this
data we explore strategies for implementing scientific applications on memory
constrained low-powered devices such as the Epiphany. With future systems
expected to house thousands of cores in a single chip, the merits of such
architectures as a path to exascale is compared to other competing systems.}},
added-at = {2019-02-23T22:09:48.000+0100},
archiveprefix = {arXiv},
author = {Varghese, Anish and Edwards, Bob and Mitra, Gaurav and Rendell, Alistair P.},
biburl = {https://www.bibsonomy.org/bibtex/2551bea6cb10fa3ca60a6392e911383bc/cmcneile},
citeulike-article-id = {13417070},
citeulike-linkout-0 = {http://arxiv.org/abs/1410.8772},
citeulike-linkout-1 = {http://arxiv.org/pdf/1410.8772},
day = 30,
eprint = {1410.8772},
interhash = {34df1bb2f6c5477b19ee98ae716a9875},
intrahash = {551bea6cb10fa3ca60a6392e911383bc},
keywords = {hpc},
month = oct,
posted-at = {2016-10-03 20:27:12},
priority = {2},
timestamp = {2019-02-23T22:15:27.000+0100},
title = {{Programming the Adapteva Epiphany 64-core Network-on-chip Coprocessor}},
url = {http://arxiv.org/abs/1410.8772},
year = 2014
}