Wednesday, April 10, 2013
Intel Brings Haswell To Servers, New SoCs For Datacenters
|You are sending an email that contains the article
and a private message for your recipient(s).
(At the moment, only Text is allowed...)
Intel announced new server chips on Tuesday, including the latest Xeon E3, which is the first server processor based on the company's latest Haswell microarchitecture. The company also introduced a rack reference architecture for data centers as well as the next-generation "Bay Trail" tablet processor.
During her keynote at IDF in China, Diane Bryant, Intel senior vice president and general manager of the Datacenter and Connected Systems Group, outlined plans to
accelerate the expansion of Intel's offerings across the data center
processor product lines based on its 22nm manufacturing
technology before the end of the year, thereby enabling a more
cost-effective and efficient data center infrastructure.
In December 2012, Intel launched the Intel Atom processor S1200 product
family, the world's first 64-bit SoC for servers that ranges from 1.6 to
2.0 GHz, and with a thermal design power (TDP) from 6.1 to 8.5 watts.
Today, Intel revealed details of three new low-power SoCs for the data
center, all coming in 2013.
Intel announced the availability of the low-power Intel Atom processor
S12x9 family. Customized for storage deployments, this SoC shares several
features with the Intel Atom S1200 processor product family, but contains
technologies specifically geared for storage devices.
- With up to 40 lanes of integrated PCIe 2.0, or physical paths between
I/O and the processor, the capacity demands of multiple devices can be
handled more efficiently. Of the 40 lanes of PCIe 2.0, there are 24 Root
Port lanes and 16 Non Transparent Bridge lanes, for failover support.
- Intel provides hardware RAID storage acceleration so that the
computationally intensive RAID function is offloaded in hardware, thus
freeing up the SoC to execute other software applications.
- PCIe Non Transparent Bridge (NTB): NTB provides failover support
- With Asynchronous DRAM Self-Refresh (ADR), the Intel Atom S12x9 family
can protect critical DRAM data in the event of a power interruption.
- Native Dual-Casting can allow data to be read from a source and
delivered to two memory locations simultaneously, which can increase
RAID-5/6 bandwidth by as much as 20 percent on a 16+2 RAID 6 system, as
compared to system bandwidth without the Dual Cast feature
Companies including MacroSAN, Accusys, QSAN and QNAP are currently
supporting the Intel Atom S12x9 family.
In the second half of 2013, Intel will push the envelope in datacenter
efficiency, and deliver the second generation of 64-bit Intel Atom
processor for microservers, codenamed "Avoton." Built on Intel's 22nm
process technology and new microarchitecture "Silvermont," Avoton will
feature an integrated Ethernet controller and expected to deliver
significant improvements in performance-per-watt. Avoton is now being
sampled to Intel's customers and the first systems are expected to be
available in the second half of 2013.
Intel will also expand its presence in the network and communications
infrastructure market by delivering an Intel Atom processor based SoC
codenamed "Rangeley," also built on the 22nm process technology. Rangeley
aims to provide an energy-efficient mechanism for processing
communication workloads and is targeted for entry level to mid-range
routers, switches and security appliances. Rangeley is targeted to be
available in second half of 2013.
Intel Xeon processor E3 Family
This year, Intel will introduce the new Intel Xeon processor E3 1200 v3
product family, based on Haswell architecture. To continue to improve
performance for video analytic workloads, the Intel Xeon E3 1200 v3
product family will support improved transcode performance. The new
Linux-based media SDK provides developers with a standard interface for
video processing, simplifies development and reduces the complexities of
accessing hardware acceleration. The SDK also maximizes simultaneous use
of CPU and Intel HD graphics capabilities for server-based video
streaming, which delivers more concurrent HD transcodes at a lower total
cost of ownership.
Intel also continues to lower the power levels on the Intel Xeon
processor E3 family; the lowest TDP will be 13 watts, approximately up to
25 percent lower than the prior generation. The E3 chip, which will
support up to 32GB of memory, is targeted at single-socket servers. It
will have an integrated graphics processor and other video encoding and
decoding features, which will be good for cloud-based video workloads.
The improvement from eight transcode to 10 transcode with Haswell's
graphics capabilities also results in up to 25 percent improvement in
transcode performance per watt for hardware accelerated media
Intel Xeon processor E5 Family
Intel's next-generation Intel Xeon processor E5 family will be based on
the 22nm manufacturing process, and will be available in the third
quarter of this year. These processors will also continue to deliver
exceptional energy efficiency by supporting Intel Node Manager and Intel
Data Center Manager Software.
The Xeon E5 is for mid-range servers with up to four sockets. The chip
succeeds the older Xeon E5 chip code-named Romley. It will draw between
60 and 130 watts, and servers with E5 can carry up to 768GB of memory.
Security will also be improved with Intel Secure Key and Intel OS Guard
which provide additional hardware-enhanced security and enhanced Intel
AES New Instructions (Intel AES-NI.) Intel OS Guard, the next generation
of Intel Execute Disable Bit, protects against privilege attacks by
preventing malicious code from executing out of application memory space,
in addition to data memory.
The Xeon E5 chip will become available in the third quarter.
Intel Xeon Processor E7 Family
Beyond software, robust hardware technology is required to generate the
computing power to analyze massive data sets. To support in-memory
analytics and rapidly respond to scaling data sets, Intel is on-track for
production availability of the next-generation Intel Xeon processor E7
family in the fourth quarter of 2013. Featuring triple the memory
capacity -- up to 12 Terabytes (TB) in an eight-socket node -- this
processor is ideal for data-demanding, transaction-intensive workloads
such as in-memory databases and real-time business analytics.
Due in the fourth quarter, the Xeon E7 chip, is designed for servers
with up to eight sockets and 4TB of memory. It includes 30MB of L3 cache
and draws up to 130 watts of power.
With the Intel Xeon processor E7 family, Intel is also announcing Intel
Run Sure Technology which will deliver greater system reliability and
increased data integrity while minimizing the downtime for businesses
running mission-critical workloads. These RAS features will be available
with the next-generation Intel Xeon processor E7 family, and will be
comprised of Resilient System Technologies, Resilient Memory
Resilient System Technologies includes standardized technologies
integrating processor, firmware and software layers, including the OS,
hypervisors and databases to allow the system to recover from previously
Resilient Memory Technologies includes features to help ensure data
integrity and enable systems to keep running reliably over a longer
period of time, reducing the need for immediate service calls.
Reference design for rack scale architecture
Bryant also revealed details on Intel's plans to develop a reference
design for rack scale architecture that uses a suite of Intel
technologies optimized for deployment as a full rack.
Traditionally, "balanced systems" such as rack servers, blade servers or
micro servers have to be refreshed to maximize the performance of select
subsystems such as CPU performance, memory, storage or the network.
Today, with the help of Intel, hyperscale customers are leading a
transformation of these "balanced" platforms toward rack scale
architecture solutions that separate, and group server, storage and
network systems, making them more modular and efficient. This
transformation starts with sharing power and thermal components, and
improving rack management to reduce operating costs, and will evolve to
include high-bandwidth fabric interconnects such as Intel Silicon
Photonics technology to enable complete disaggregation of racks to drive
optimal flexibility for large scale data center deployments. Intel sees
the evolution of rack design happening in three phases:
- Physical Aggregation. All non-critical sheet metal removed and key
components such as power supplies and fans taken out of individual
servers and consolidated at the rack level. Savings are expected due to
higher levels of efficiency and lower costs by reducing the number of
fans and power supplies.
- Fabric Integration and Storage Virtualization. Disaggregate and
separate out the storage from compute systems with direct attached
storage, and achieve higher utilization through storage virtualization.
The compute and network fabric is the key technology that is enabling
disaggregation of storage without impact to performance. Intel Silicon
Photonics interconnects will enable higher speed connections between
various computing resources within the rack, thus enabling the eventual
disaggregation of server, memory, network and storage within the rack.
- Future. Ultimately, the industry will move to subsystem disaggregation
where processing, memory and I/O will be completely separated into
modular subsystems, making it possible to easily upgrade these subsystems
rather than doing a complete system upgrade.
Benefits of rack scale architecture include increased flexibility, higher
density and higher utilization leading to a lower total cost of
ownership. Based on the needs of cloud service providers and large
hyper-scale data centers, Intel is developing a reference design that
utilizes Intel technologies and allows a range of solutions for OEM
providers to develop and deliver racks. Intel rack scale architecture
will include a suite of technologies such as: Intel's Xeon and Intel Atom
SoCs for servers, storage, and networking; Intel Ethernet switch silicon
for distributed input/output; and Intel's new photonic architecture,
based on high-bandwidth, Intel Silicon Photonics Technology. Silicon
Photonics Technology will enable fewer cables, increased bandwidth,
farther reach and extreme power efficiency compared to todays
copper-based interconnects. Intel demonstrated a mechanical prototype of
this new rack architecture earlier this year and will publish a complete
reference architecture optimized for deployment as a full rack to make it
easy for system builders and customers to adopt.
One implementation of Intel's rack scale reference architecture is
already taking place in China, as Alibaba, Baidu, Tencent and China
Telecom are collaborating with Intel on Project Scorpio, an initiative
that will deliver a physical aggregated rack consisting of the fan and
power supplies in six zones within the rack, with the goal of
demonstrating TCO savings.
Server makers Dell and IBM already offer the option to decouple storage
and memory with the help of specialized chips. However Intel believes
there is a need for better throughput and traffic management in data
centers so the components could be truly decoupled.
Advanced Micro Devices offers its own proprietary fabric that virtualizes
I/O and storage, reducing the need for components in a server. The
company has said it is looking to expand the use of that fabric, called
Freedom Fabric, beyond servers.
New chips for Ultrabooks and tablets
During his keynote, Kirk Skaugen, Intel senior vice president and general
manager of the PC Client Group, provided a deeper look at the forthcoming
4th generation Intel Core processor family, which he said is now shipping
to OEM customers and will launch later this quarter.
"Ultrabooks based on the 4th generation Intel Core processor family will
enable exciting,new computing experiences and all-day battery life
delivering the most significant battery life capability improvement in
Intel's history," said Skaugen. "It will also bring to consumers a new
wave of 'two-for-one' convertible and detachable systems that combine the
best of a full PC experience with the best of a tablet in amazing new
The new Intel Core microarchitecture will allow the company to deliver up
to double the graphics performance over the previous generation. In
addition, the new graphics solution will have high levels of integration
to enable new form factors and designs with excellent visual quality
built in. Skaugen demonstrated these graphics improvements on the 4th
generation Intel Core processor-based Ultrabook reference design called
"Harris Beach." The demo featured Dirt 3 gaming title, showing
the same visual experience and game play as a discrete graphics card that
users would otherwise have to add separately. He also showed the 4th
generation Intel Core processor-based concept, codenamed "Niagara," a
premium notebook with the ability to play the unreleased enthusiast title
Grid 2 from CodeMasters without the aid of a discrete graphics card.
Along with touch capability, Intel Wireless Display (Intel WiDi) will be
enabled on all 4th generation Intel Core processor-based Ultrabook
devices to allow people to securely stream content and apps from devices
to the big screen, free from the burden of cables. Skaugen said the China
ecosystem is taking the lead on integrating Intel WiDi into systems, and
announced that the leading television manufacturer in China, TCL, has a
new model with the Intel WiDi technology built in. He also announced new
receivers certified for Intel WiDi from QVOD and Lenovo and a set-top box
Illustrating the low-power advances in Ultrabook devices, Skaugen showed
off the new Toshiba Portege Ultrabook detachable, based on the new
low-power line of the 3rd generation Intel Core processors.
Furthermore, Skaugen revealed that voice interaction in Mandarin is now
available on Ultrabook devices from Intel through Nuance.
Augmenting the company's offerings for computing at a variety of price
points, Skaugen announced plans for new market variants of its "Bay
Trail" 22nm SoC with PC feature sets specifically designed for value
convertibles, clamshell laptops, desktops and value all-in-one computers
to ship later this year.
Tan Weng Kuan, vice president and general manager of the Mobile
Communications Group, Intel China, highlighted how the company is working
with ecosystem partners to deliver the best smartphone and tablet
experiences with Intel inside. Tan discussed the company's progress with
the new Intel Atom processor Z2580 ("Clover Trail+") for smartphones and
the Intel Atom Processor Z2760 ("Clover Trail") for tablets.
Taking full advantage of the capabilities enabled by Intel architecture,
processor technology, manufacturing and multi OS support across Windows 8
and Android, Tan discussed the company's forthcoming smartphone and
tablet products based on Intel's 22nm process and an entirely new Atom
microarchitecture.Intel's quad-core Atom SoC ("Bay Trail") will be the
most powerful Atom processor to-date, doubling the computing performance
of Intel's current-generation tablet offering. Scheduled for holiday 2013
tablets, "Bay Trail" will help enable new experiences and designs as thin
as 8mm that have all-day battery life and weeks of standby.
Tan also highlighted Intel's Atom SoC, codenamed "Merrifield," which is
scheduled to ship to Intel's customers by the end of this year. The
product will deliver increased smartphone performance, power efficiency
and battery life over the current-generation offering.
Tan closed his remarks by calling upon China developers to help to
accelerate and grow the mobile market together. He announced the creation
of a China-specific expansion of the company's platform and ecosystem
enabling efforts, focused initially on Atom processor-based tablets
running Android, and designed to speed time-to-market of mobile devices
based on Intel technology. He added that China developers are
instrumental to this effort and will bring speed, scale and ingenuity
that will drive new innovation globally.