EEMBC
Reveals First Performance Benchmarks for Scale-Out Servers and Associated SoCs
ScaleMark-Caching
measures memory caching latency and throughput while demonstrating EEMBC’s
industry-standard methodology
EL
DORADO HILLS, Calif. — November 11, 2015 — The Embedded
Microprocessor Benchmark Consortium (EEMBC, pronounced “embassy”) today unveiled
ScaleMark-Caching
test, which measures a server’s responsiveness to client requests for web data.. This benchmark, the first produced by EEMBC’s Cloud and Big-Data Server
working group, is
based on the popular Memcached application, with the inclusion of a fixed set
of operating parameters and utilization of a large-scale web-server workload. As
Memcached is used in data centers to optimize
performance and energy usage, EEMBC’s ScaleMark-Caching
is not a proxy benchmark, it benchmarks a real-world
application utilizing real-world data.
The Cloud and Big-Data Server working group is
chartered by EEMBC to build a standardized, industry-endorsed suite of speed and
efficiency benchmarks that characterize SoC and
system-level performance for modern cloud and big-data-related workloads,
commonly called hyperscale and “scale-out” computing.
Although ScaleMark-Caching is only one segment of the benchmark suite, it successfully
embodies the comprehensive methodology that EEMBC has established. Like
previous benchmarks, perhaps the most important attribute of the
ScaleMark-Caching benchmark is its clear definition of parameters. This
approach ensures repeatable and verifiable benchmark results. The EEMBC
standard mode implements the “Facebook ETC” profile explained in “Workload
Analysis of a Large-Scale Key-Value Store[1]”.
This profile defines the key size, value size, and inter-arrival distribution.
“ScaleMark-Caching
is an extraordinarily complex application under the hood that accurately
represents both the server under test as well as clients making requests to the
server,” said Markus Levy, EEMBC’s president. “Consistent with our
industry-proven methodology, our Cloud and Big-Data Server working group
utilized various industry-standard tools to allow users to easily configure
parameters to define and test their system’s configuration.”
"EEMBC's Cloud and Big Data Server
Benchmark working group is filling a significant gap for modern web-scale infrastructure
vendors and customers – it is creating an accurate set of processor and system-architecture-neutral
benchmarks for high-value workloads," said Paul Teich,
Principal Analyst, Tirias
Research. “This working group is enabling much more accurate total cost of
ownership metrics for modern data centers, workload by workload. I’m looking
forward to seeing EEMBC ScaleMark-Caching results.”
Following the release of its
ScaleMark-Caching benchmark, the group is developing a media-streaming
benchmark and is defining requirements for a subsequent generation of benchmarks.
Like all EEMBC benchmarks, all Cloud and Big-Data Server working group
benchmarks will be based on real-world applications.
The Cloud and Big-Data Server working group,
led by Shay Gal-On, principal engineer at Cavium, Inc., is comprised of EEMBC
members with strong interest in the server market. These include AMD, ARM, Cavium,
Intel, and others. EEMBC encourages vendors and manufacturers to join the
consortium’s working groups to contribute to the definition and development of its
next-generation benchmark suites. To join the Cloud and Big Data Server, or
other working group and/or try out the EEMBC ScaleMark-Caching benchmark,
contact Markus
Levy for details.
# #
#
About EEMBC
EEMBC was
formed in 1997 to develop performance benchmarks for the hardware and software
used in embedded systems. EEMBC benchmarks help predict the performance and
energy consumption of embedded processors and systems in a range of
applications (i.e. automotive/industrial, digital imaging and entertainment,
networking, office automation, telecommunications, and connected devices) and
disciplines (processor core functionality, floating-point, Java, multicore, and
energy consumption).
EEMBC members
include Ambiq Micro,, AMD, Analog Devices, Andes
Technology, ARM, Atmel, C-Sky Microsystems, Cavium, Cypress Semiconductor,
Dell, Freescale Semiconductor, Green Hills Software, IAR Systems, Imagination
Technologies-MIPS, Infineon Technologies, Intel, Lockheed Martin, Marvell
Semiconductor, MediaTek, Microchip Technology, Nokia Networks, Nordic
Semiconductor, NVIDIA, NXP Semiconductors, Qualcomm, Realtek Semiconductor, Red
Hat, Renesas Electronics, Samsung Electronics, Silicon Labs, Somnium
Technologies, Sony Computer Entertainment, STMicroelectronics, Synopsys, Texas
Instruments, TOPS Systems, and Wind River Systems.
[1] Workload Analysis of
a Large-Scale Key-Value Store, by B. Atikoglu, et al,
June 2012
SIGMETRICS’12, June 11–15, 2012, London, England, UK.
Copyright 2012 ACM 978-1-4503-1097-0/12/06 ...$10.00