Garrett McKibben (Intel) describes some of the Intel Platforms that are tailored for HPC and server clusters.

Part 2 of 2:

Sun Sets Records with New x64 Server


Sun Microsystems, Inc. today announced its new Sun Fire X4640 server, based on the Six-Core AMD Opteron processor, delivered outstanding performance on industry-standard benchmarks. These results highlight the enterprise-class performance and significant infrastructure and management cost savings Sun's integrated infrastructure with the Solaris Operating System (OS) deliver through the convergence of compute, networking and storage.

The new Sun Fire X4640 server uses up to eight Six-Core AMD Opteron processors in 4RU, making it the most compact 24- to 48-core system available from tier one vendors. With up to 65 percent better performance than the previous generation Sun Fire X4600M2 server(1) and up to half a terabyte of memory in 64 memory slots, the Sun Fire X4640 server helps customers increase data center efficiency and is an exemplary platform for enterprise databases and HPC applications as well as virtualization and consolidation.


More . . .

SC09: "Intel's Next Generation 2-Socket HPC Platform-Visualize Faster with Westmere-EP"


In this video, Brian Rea (Intel) and Philip Lunn (Bunkspeed) show a pre-release Westmere-EP Platform running 24 threads with Bunkspeed's Hypershot application. This is truly amazing performance, check it out!

The top data center 2010 trends


Most companies are riding the global economy rollercoaster, and that means data center managers and their plans and budgets need to be flexible in 2010. At the 2009 Gartner Data Center conference in early December, analysts recommended IT managers develop multiple budgets to be better prepared for various macroeconomic scenarios.

Looking to 2010, data center managers are facing more uncertainty than at any time in the past decade. As trusted vendors collapse, application delivery and data center design models evolve, data center pros need to try to forecast change to the best of their ability. But planning further than 12-months out will be a challenge.


More . . .

It was a showdown in Las Vegas as the data center container solutions from three leading server vendors - the HP POD, IBM's Portable Modular Data Center and the ICE Cube from SGI/Rackable - were all on display at the Gartner Data Center Conference at Caesar's Palace. Here's a brief look at the three 40-foot containers showcased at the event.

Server vendor Rackable Systems recently introduced its CloudRack C2, a rack system optimized for low energy use that can operate safely at temperatures up to 100 degrees in the data center hot aisle. In this video from Uptime Symposium 2009, Rackable's George Skaff and Lior Pastor explain the innovations in the CloudRack and server trays.

For more data center news and video, visit our web site: http://www.DataCenterKnowledge.com

On June 1, 2009, NVIDIA announced the Teslaâ„¢ M1060 Processor. An integral part of the worlds fastest 1U computing servers engineered to meet the requirements of enterprise-class data centers, the Tesla M1060 processors deliver supercomputing performance while requiring less power and space.

Featuring the revolutionary NVIDIA CUDAâ„¢ parallel computing architecture and powered by 240 parallel processing cores, the Tesla M1060 shatters your performance per watt expectations to help you to solve the toughest computing problems faster.

Maloney [Sean Maloney, executive vice president and general manager of the Intel Architecture Group] also provided attendees with a preview of Intel's next-generation intelligent server processor, codenamed Westmere-EP, and Intel's commitment to the high-end of the server market with its Xeon and Itanium processor families. Maloney discussed the unprecedented generational improvement that the forthcoming Nehalem-EX server processor will deliver, with performance improvements even greater than what the current Intel® Xeon® 5500 Series provided versus Intel's previous chip generation.

Maloney described the convergence of compute, networking and storage in the data center, sharing the company's vision of a converged datacenter IO fabric led by Intel 10GbE solutions. Intel also has a number of joint efforts with other industry leaders to deliver optimized platforms, systems, technologies and solutions to address the hyper-scale data center environments of the Internet and cloud services trend.

Maloney disclosed a new ultra-low-voltage Intel® Xeon® 3000 series processor featuring a TDP (Thermal Design Power) of only 30 watts. To complement the broad range of dense and power-optimized platform offerings, Intel also demonstrated publicly for the first time a single-socket "micro server" reference system which will help enable micro server innovation and future specification.



More . . .

IBM Turns Up the Heat in New Data Center to Cut Back Energy Bill


IBM isn't exactly playing with fire, but it is playing with higher temps in its new North Carolina data center. At a glance, it might seem counterproductive to raise temperatures, but IBM is doing so in order to reduce its energy usage.

To make sure things don't get too far out of hand, IBM has equipped its 60,000 square-foot data center with thousands of sensors that dynamically keeps tabs on temps, humidity, air flow, and circuits. And to help with cooling, the company will rely largely on outside air.

Temps could rise as high as 80.6 degrees, which is exactly the latest environmental recommendation by the American Society of Heating, Refrigerating, and Air-Conditioning Engineers, who recently raised the threshold from 77 degrees because of improvements in equipment design.


More . . .

Intel ships first quad-core Itanium: Quad-core Intel Itanium 9300 CPUs announced


Intel on Monday introduced its latest processor, the Itanium 9300-series, which is said to more than double the performance of the chip it replaces. Meant for use in enterprise servers, the 9300-series chips have two billion transistors and four cores, doubling that of the previous chips. This and other changes mean the new chip has eight times the interconnect bandwidth, five times the memory bandwidth and seven times the memory capacity with DDR3 components than the outgoing processors.

There are also now eight threads per processor thanks to Intel's enhanced Hyperthreading Technology. Also improved are the reliability, availability and serviceability (RAS) features. Recovery from potentially fatal errors is ensured by the use of new, advanced machine-check architecture for the hardware, firmware and OS. Performance is otherwise improved with a newer generation of in-hardware virtualization.

OEM systems using the Itanium 9300 CPUs will ship within the next three months. Prices in 1,000-chip batches range from $946 to $3,838.



More . . .

IBM Announces New POWER7 Servers


IBM has announced its latest generation of its Power-based servers, the new POWER7 system, which is designed to manage demanding emerging applications such as smart electrical grids and realtime analytics for financial markets, which rely on processing an enormous number of concurrent transactions and data while analyzing that information in real time. The new systems enable clients to manage applications and services at lower cost through technology breakthroughs in virtualization, energy savings, more cost-efficient use of memory, and price performance.

The new POWER7 processor has eight cores with four threads each, which is 4x the maximum number of cores in POWER6 systems and 8x the number of threads per chip. With the TurboCore mode, four cores are deemed active and most of the resources backing all eight cores are put behind the four active cores thus increasing the cache and memory bandwidth, and allowing the clock speed to be increased, which can result in significant per-core performance gains. From a purely computational and transaction processing perspective, this achievement of POWER7 is admirable. However, this is only part of the story.

Through POWER7’s Intelligent Energy technology, organizations can power up or shut down various sections of the server as well as dynamically adjust processor clock speeds based on thermal conditions and system utilization, on a single server or across a pool of multiple servers. The integration of energy management spanning the processor, firmware, PowerVM virtualization, OS, up through IBM Active Energy Manager Software (included in Systems Director Standard and Enterprise Editions) allows organizations to tune not only their systems performance and overall energy usage but also the specific price/performance yield of each processor and, by extension, applications supported by each processor. This degree of energy management flexibility illustrates the energy efficiency on a performance per-watt basis afforded by the POWER7 over competitive platforms such as x86, SPARC, and Itanium-based solutions.


More . . .

I've seen articles about that. I look at my G5 Mac and think "I waaant!!!".

Sun also released information on the SPARC Niagara. I can't find the article now that compared Niagra and Power7. Basically, Niagara has more cores, very good data transfer, short pipe-line, but rather low cache. Basically it's designed to do lots and lots of parallel, low processing-power tasks. It is targeted towards very large web servers, database servers, etc. Intense number crunching is actually a weak point for Niagara, though I imagine some work could be divided into highly parallel tasks and distributed across the many threads of many cores.

Power7 is the opposite, and was probably guided by DoD and Department of Energy funding for massive number crunching. The P7 actually has 32Mb of level 3 cache, which is a relatively huge amount. To put it in perspective, you could easily and very comfortably run Windows For Workgroups 3.11 (not a modern but still quite useful) operating system in that much processor cache. The P7 is designed for fast crunching of large amounts of data.

Personally, while the P7 is obviously designed for supercomputers, I think the SPARC Niagara may also fit. In heterogeneous computing environments, it seems like a simple-task workhorse like Niagara would be good for quickly moving data around, and basically managing everything. (As I understand it, some already use AMD Opteron mixed with CUDA for crunching in a similar situation.)

Mike

AMD aims for GPUs in mainstream servers starting 2012


Advanced Micro Devices will put more focus on tightly integrating graphics processor cores into mainstream servers starting 2012 as it tries to increase system performance, a company executive said.

Over the next two years, computing using a GPU will remain a relatively niche area, so AMD's focus will be on CPUs and adding cores to it, Longoria said. Over time there could be tighter integration, and AMD could de-emphasize CPU cores if heterogeneous computing using GPUs takes off, Longoria [Gina Longoria, director of the product management and workstation division at AMD] said. AMD currently has a 16-core server CPU code-named Interlagos planned for release in 2011, and it will be based on a new chip architecture.

A lot of the high-performance computing community is doing their work based on the OpenCL platform, which could eventually trickle down into general-purpose computing, Longoria said. OpenCL competes with Microsoft, which is promoting its proprietary DirectX parallel programming tools, and Nvidia, which offers the CUDA framework.

AMD is focusing OpenCL platform applications to be written for the FireStream processor, a family of high-performance GPUs. AMD acquired graphics technology when it bought ATI Technologies in 2006.


More . . .

Intel sneak peeks Westmere EP server silicon: Four and six cores, Turbo Boost, AES


Specifically, the Westmere-EP is using the extra transistor budget that the slide from 45 to 32 nanometer processes allows to add two more cores to the processor and to boost the on-chip L3 cache by 50 per cent to 12 MB per chip. Nasser Kurd, senior principle engineer at Intel's Architecture Group confirmed to El Reg that Intel will deliver four-core variants of these chips. Nasser also confirmed that the Westmere-EP chips will support Turbo Boost, which allows for the clock speed of the processor cores to be jacked up a bit as other elements of the chip are quiesced.

Generally speaking, the Westmere-EPs will have the same clock speed range and the same thermal envelopes as the existing Xeon 5500s, but Intel is has not yet announced specific SKUs and won't until the middle of March or so when these new chips are formally launched. The Westmere-EPs will plug into the same sockets and use the same chipsets and DDR3 main memory as the Xeon 5500s and have three memory channels per socket like them as well.



The six-core Westmere-EP chip has 1.17 billion transistors and is 240 square millimeters in size. As you can see from the pretty picture above, it is implemented in two halves of three cores each. The core regions have their own clock speed and power supply, and with the tweaks to the Westmere design the L3 cache and memory controller regions - what Intel calls the "uncore" areas - get their own, separate power gating.


More . . .

Intel's New Xeon 5600 Chips Outperform Predecessors


Intel introduced a line of Xeon server chips that operate up to 60 percent faster than previous server processors, the company said on Tuesday.

The 15 new processors in the Xeon 5600 line include low-power chips and come in quad-core and six-core variants, the company said. The processors are faster than the older Xeon 5500 series chips, which were launched about a year ago.

The clock speeds of the new chips range from 1.86GHz for the quad-core Xeon L5609 processor, which consumes only 40 watts of power, to 3.46GHz for the fastest Xeon X5677 six-core chip, which draws up to 130 watts. All the processor chips include 12MB of cache.

Intel already offers a six-core Xeon processor code-named Dunnington that was announced in 2008. The company is also expected to announce an eight-core Nehalem-EX server processor later this month as it tries to improve the performance of chips by adding more cores.



More . . .

Watch as Harris Sussman (Cisco) shows how the Intel Xeon Processor 7500 completes the portfolio for a Unified Computing System. This segment was shot during the NYC Launch Event. Thanks for watching!

Learn how Elemental Server, a enterprise-class transcoding solution, uses GPU-acceleration to provide unmatched workflow integration, uncompromised video quality and superior price/performance within broadcasters' and content delivery networks' server farms.

IBM Puts Nvidia Tesla GPU in iDataPlex HPC Server


Nvidia's efforts to push graphics chip technology into mainstream computing received a significant boost May 18 when IBM announced that the newest version of its iDataPlex server will run on both traditional computing chips and GPUs from Nvidia.

The announcement marks the first time a top-tier systems vendor has made the move to offer a hybrid CPU-GPU server in the HPC (high-performance computing) space.

Nvidia officials said having IBM adopt Tesla 20-series GPUs in its iDataPlex dx360 M3 system is a validation of both Nvidia and of the vision the company is promoting.

Nvidia officials are saying in order to keep computing capabilities improving at the same pace they have for the past few decades, there has to be shift away from serial CPUs and toward parallel processing found in GPUs.

Intel and AMD are looking to keep making significant gains in performance through small chip sizes and more processing cores. However, in a column on Forbes.com in April, Bill Dally, vice president and chief technologist at Nvidia, said the power needs of compute chips from Intel and Advanced Micro Devices are making it difficult to sustain Moore's Law.


More . . .

Managing cooling across multiple racks in a data center can be a challenge, particularly if each rack of gear is running at different power densities.

At last week's Uptime Symposium, Electrorack demonstrated a new rack design which allows two or even three racks of equipment to be housed in a single enclosure and controlled by a single rack-top chimney system. Todd Schneider, the director of marketing for Electrorack, provides an overview of the new design, which employs a rear cooling plenum and a pressure-sensing management system from Opengate Data Systems.

New, high performance 1U rack system features the Intel® Server Board S5500WB, the Intel® Xeon® processor 5600 series, and supports redundant AC or DC power options. Ideal for use in internet portal data centers, cloud computing, and high-performance computing (HPC).

Learn more: http://www.intel.com/Products/Server/...

IBM unveils hybrid mainframe; 'System of system' eyes data center consolidation


IBM on Thursday will unveil a new hybrid mainframe design that aims to cut data center sprawl and be a bridge to other systems.

IBM’s move is likely to raise the debate about the mainframe, which has had obituaries written about it for years. Some companies, which are on the scale out with standard servers bandwagon, wouldn’t be caught dead with mainframes. Enterprises that already have mainframes to process a massive amount of transactions—think financial services companies like Citi—will be receptive. Add it up and every big tech company has some spin on converging infrastructure to create efficient, automated computing systems. IBM has the same message with a mainframe twist.

Separately, BMC Software, a leading mainframe software management company, said it will support IBM’s hybrid mainframe with its Business Service Management tools. BMC sees the hybrid approach as a logical way to bring systems together. “IBM’s mainframe approach brings the cloud opportunity to the forefront,” said Bill Miller, president of BMC’s Mainframe Service Management business unit. “This could be a lot more than a traditional server and with its split personality could offer scalability like nothing else out there.”

At the very least, IBM has another data center stack to evaluate.



More . . .

IBM on Thursday announced the zEnterprise mainframe server and a new systems design that allow workloads on mainframe, POWER7 and System x servers to share resources and be managed as a single, virtualized system.

Scott Denham of IBMs Deep Computing Technical team speaks about why they selected NVIDIA Tesla GPUs for their new iDataPlex Servers.