Intel Communities : Blog List - The Server Room

Nehalem-EX for High Performance Computing

webadmin@intel.com — Mon, 23 Nov 2009 20:58:17 GMT

Nehalem-EX: Big Memory for Big Science

I was at SuperComputing’09 last week in Portland, Oregon. I talked with some brilliant people, and saw some fantastic stuff.

It was good timing on my part because last week Intel also announced that it would offer a 6-core, frequency-optimized version of its Nehalem-EX product due out next year. This part is intended for use in tackling some of the types of high performance computing (HPC) workloads prominently displayed at SC’09.

Most people know that the majority of HPC workloads today are based on clusters of relatively small-memory, 2-socket systems. That is because most HPC workloads may be broken into smaller, discrete units of work that can be efficiently processed using such clusters. For these workloads the primary hardware capability selection criterion is typically a balance of both memory bandwidth and compute FLOPs (floating point operations per second).

But there are other types of HPC workloads. Specifically, those that deal with very large datasets (some as large as a terabyte) or those that have to deal with non-sequential memory access. This means the workloads simply aren’t easily divisible--or it is inefficient to do so-- into the relatively small memory footprints used in traditional clustered 2-socket HPC solutions. Examples of these types of bigger memory applications can be found in a variety of fields such as weather prediction, manufacturing structure analysis, and financial services.

The high-speed processing requirements and size of these workloads put a greater premium on system memory capacity/bandwidth than on compute FLOPs.

If the larger dataset won’t fit into available memory, and dividing up the dataset to spread across multiple nodes cannot easily be done, then data has to be moved in and out memory to hard disk. But using hard disk drives (which are many times slower than RAM memory) can drastically impair performance.

There are now two better alternatives to the use of hard drivers. One is SSDs and the other is having a larger memory footprint. Solid State Drives have fairly high data density vs RAM, but much faster access than hard-disk drives--albeit still markedly slower than RAM. Another solution is to simply have more capacity of the faster RAM. This last one is what the Nehalem-EX HPC part is aimed at.

Nehalem-EX is the Expandable Class of Nehalem. The Expandable Class brings all the goodness of the Nehalem architecture (Xeon 5500 product line) to the HPC market, but in the form of a “super node” that has greater: a) core/thread count, b) socket scaling (up to 256), c) I/O and memory capacity (up to 1 terabyte in a 4 socket system) and bandwidth at capacity, d) reliability features, e) and other features.

The 6-core frequency-optimized Nehalem-EX part has also been tuned to offer the highest core frequency possible for this chip. In creating this part, Intel is meeting the needs of the HPC community that want higher scalar performance along with the benefit of large memory capacity and bandwidth per core.

Of course the 8-core version of NHM-EX is still an option for those HPC workloads that scale well with more cores while still looking for the high memory capacity of the expandable class.

By having both 8-core and frequency optimized 6-core versions of the NHM-EX class of processors means HPC researchers have greater choice in selecting the processor best suited for their specific workloads.

After talking with some of the researchers at SC’09 last week I’m really excited to see how the Nehalem-EX “super node” will deliver the necessary compute and memory capabilities to help those researchers solve some of their biggest challenges.

'You are not Alone' - When it comes to moving from RISC

eoin.j.mcconnell@intel.com — Mon, 23 Nov 2009 17:18:41 GMT

Change is hard, but it can be done and the benefits of change usually outweigh the concerns which were on our minds before we made the change.

When making the change from running your solution on a RISC architecture to running that solution on a Xeon architecture, the biggest concern usually relates to whether that solution will run at the same level as on the previous architecture. I'm not talking about performance specifically, but usually the question is around whether operating systems like Linux, Windows, and Solaris on Xeon will meet your business needs for yourmission critical solutions.

Like the underlying improvements in the microprocessor, I believe that there have also been major fundamental improvements in the operating systems that run on both today's and the soon to come next generation microprocessors (sorry, my obligatory Nehalem-EX advertisement... coming soon in 2010). A decision made many years ago to run your solution on Unix/RISC was made based on comparing all the different variables at that time to pick what was right for your business. At that time you likely decided that your solution would not run on these operating systems, these operating systems were not suitable for your mission critical workloads etc. Probably right decision at that point, but like everything else decisions get revisited based upon the here and now and what may have been the right solution in the past (and right decision) may not be the right solution for your needs now.

I wanted to share some thoughts specifically on Redhat Linux today. Lets take a little look at Redhat Enterprise Linux. Current versions of Redhat can deliver what is required for your critical solutions. RHEL is ready and here are some of the reasons cited by Redhat in recent webinars on this topic and my interpretation of their comments

Hosts real-time global mission-critical infrastructures and operations 24 X 7 - its tried and tested by other Enterprises
Enables 5x9s availability in highly secure environments - pretty important to most critical solutions
Contributes measurable reductions to TCO and enables, agile, standardized, and virtualized infrastructures - TCO benefits through standardization
Has major ISVs on-board with the majority of 3rd party Unix applications have Linux and/or Windows versions available - the ISVs that traditionally delivered applications to you based on Unix, also have versions supported on Linux/Windows
Many customer unique applications are developed with programming languages such as C, C++, JAVA, or J2EE and can be migrated to Linux and / or Windows - your applications can be moved
Hosts most major database systems standard for your infrastructure - all the major databases run and run well on Linux

One of the other things we encounter a lot is around whether the technical considerations to move from one operating system environment are too high to overcome and outweigh the benefits of moving. There are always technical considerations and things that you need to know to move from one environment to another. However you are not alone in trying to understand these technical considerations. Redhat have done a phenomenal job of documenting the challenges of moving from say Solaris to Linux and have developed a great Strategic Migration Planning Guide. This is available on request. In recent webinars Redhat outline some of the things that you need to consider for the following technical categories

- Development Environment; Kernel tuning; Security; Filesystems; Debugging, tracing, Profiling; Command Differences; Deployment methods; Software Management; Virtualization; Application considerations

In addition to the current versions of Redhat running on Intel architecture, we are also working very closely on future versions that will take advantage of the 20+ new RAS features that are planned for Nehalem-EX - more on that in a future blog

You are not alone, resources, tools and expertize exist to help you make that move and reap the business benefits while still delivering to the requirements of your business. Check out Redhat online tools for more information that dives deeper into all the areas for consideration http://www.redhat.com/migrate/solaris_to_linux/

We think Redhat Linux and Xeon are ready to run your mission critical workloads and solutions...What do you think?

Education at Super Computing 2009

webadmin@intel.com — Thu, 19 Nov 2009 22:29:34 GMT

I have to admit, I have never had the opportunity to be involved in HPC, Super Computing, or the communities that have evolved around such things. My first real experience with it was yesterday at SC09 in Portland, Oregon. A conference like any other was my thinking. But, when I started walking around the exhibition area (booths), I was amazed at the number of Universities and education based solutions that were represented.

Here is a quick montage of images that I put together of the educational facilities I saw and took a picture of...

I am sure there are many, many more, but I only captured the few that are represented here.

The one on the bottom left is not really an educational institution, but rather a company that I stopped and talked with for a few minutes. They essentially offer up their datacenter and supercomputer infrastructure for all the education facilities in the state of Alabama... K-12 and Universities too.

Here is another picture I took of a scout troop that was visiting the event. What a great opportunity for them to see and hear all about the most powerful computers in the world.

Here they are again listening (attentively) to a speaker from NASA talkiing about the expansion of the universe and how we study it.

Education is and should remain a priority for all of us. This is hopefully a good reminder of that. It certainly was for me.

-Greg

10 Gigabit iWARP @ SC’09!

webadmin@intel.com — Thu, 19 Nov 2009 18:48:09 GMT

What is iWARP? (Click here to find out)

Ethernet Tom here. I’ve recently come into the Intel Ethernet group – marketing Ethernet products in the HPC, Financial, and Cloud verticals.

And very HPC relevant - I’ve been very busy getting things lined up for Super Computing ’09 and wanted to share what we have happening:

· 3 demos:

o Intel Booth: Two 6-node clusters (96 total cores!) running NYSE’s Data Fabric* middleware – showing iWARP vs. non-iWARP

o Supermicro Booth: 4-node cluster running Fluent

o EA Booth: 4-node cluster running Linpack in a converged Ethernet environment

· 5 presentations:

o Data Transportation at iWARP Speed – Feargal O’Sullivan – NYSE Technologies

o Memory Virtualization over iWARP for Radical Application Acceleration – Tom Matson – RNA Networks:

o A 10Gigabit Ethernet iWARP Storage Appliance – Paul Grun – System Fabric Works

o CD-adapco Benchmarking Performance using Intel iWARP – William Meigs – Intel Corporation

o iWARP – What & Why? – Tom Stachura – Intel Corporation

If you are here today (final day of SC’09!), stop by and check it out. If not, I’ll come back later with some video links.

Green Storage

webadmin@intel.com — Wed, 18 Nov 2009 20:52:02 GMT

It’s not just about energy-sipping systems—it’s also about your storage footprint

Most of us are familiar with the concept of green IT: increasing energy efficiency across the enterprise to trim costs and optimize resources. While you hear a lot about servers helping to reduce energy usage, not as much is said about storage. Intel and the storage industry are working together to provide green storage solutions, too.

For the storage community, every system has to be cost-effective as well as performance-driven, which means energy efficiency is a key consideration. It starts at the processor level, where the Intel® Xeon® processor 5500 series is extending the boundaries of energy efficient performance.

Many storage system providers have picked up on the Intel Xeon processor 5500 series since it was introduced last March. For example, the HP StorageWorks XP10000* Disk Array and 3000 Enterprise* Virtual Array are based on the new processors. Schooner Information Technology appliances leverage quad-core Intel Xeon 5500 processors and half a terabyte of Intel® X25-E flash memory. The bottom line for the Schooner appliances is an 80 percent decrease in power and cooling requirements versus ordinary servers.

But green storage isn’t just about power consumption at the processor or system level. An equally important green strategy is to reduce the overall storage footprint, and a number of technologies are available to help IT organizations implement this strategy.

Virtualization is driving huge data center energy savings by greatly reducing the number of physical machines in the data center. As Bob Fine, director of product marketing at Compellent, pointed out at the 2009 Storage Networking World conference last spring, many large enterprises realize that they’re approaching a cap. “They can only get a certain amount of power in their data centers and see virtualization as a way to reduce their power requirements,” says Fine. “Instead of building new data centers, they can stay in the ones they have, saving millions of dollars in the process.”

Many IT managers tell Intel that storage can be a big gating factor when it comes to scaling virtual environments. The Intel Xeon processor 5500 series uses Intel® HT Technology within each processor core, doubling the number of threads that can be processed at the same time. This option permits more efficient workloads and enables storage servers to virtualize more applications. Intel HT Technology is also more energy efficient than traditional threaded processing.

Compellent and Hitachi Data Systems (HDS), both users of the Intel Xeon processors, recommend reducing the storage footprint in other ways as well. “Limit the amount of content you need to store by using technologies like data deduplication,” advises Asim Zaheer, vice president of product and competitive marketing at HDS. “Also, don’t have wasted capacity or wasted systems—that’s where tiered storage and virtualization come into play.”

Compellent’s Fine sees tiered storage as especially important when using expensive disk resources like solid-state drives (SSD). By limiting SSD to the top tier, a company could save on drive costs and increase storage efficiency. “Only the active data would sit on SSD, and all the inactive data would go onto a tier-three SATA drive,” says Fine. “Since SSD drives are about 10 times the cost of Fibre Channel, it’s very important to gain those kinds of efficiencies.”

Isilon Systems, another user of Intel processors, has a pay-as-you-grow model for its clustered storage products that makes it easier to avoid over-provisioning and wasting power. If a customer needs to add more performance, Isilon can provide nodes with Intel processors and memory, but no storage. If the customer requires capacity only, Isilon sells nodes with just disks. In addition, Isilon uses ColdWatt power supplies, which it says are about 30 percent more efficient than traditional power supplies.

As Intel works with the storage industry to deliver more energy-efficient and high-performance storage solutions, we’d like to know what IT organizations are doing to implement green storage technologies in the data center. If you work in IT and have fresh perspectives to make your organization more efficient, you’re invited to share your ideas here.

"Live From Super Computing 2009"

webadmin@intel.com — Wed, 18 Nov 2009 20:12:27 GMT

This week I'm in Portland, Oregon, where I call home. Its interesting for me since this is my first Super Computing conference, and soo far, I'm really impressed, not only by the intense knowledge and the plethera of scientific discovery all around, but also by the fact this conference is so well attended. There -s a huge trade show floor, filled to capacity where you can see everything from genome research to oil and gas exploration, to bio-computing. . It's very cool to see NASA, Oak Ridge Labratory, and many top universities all showing off the lastest in High Performance Computing, some very cool stuff indeed. From the point of view of higher learning and how super computers are changing the world, this is the place to be. Here are a few shots of the Intel booth in case you get a chance to come by and see us.

I'll be capturing some cool videos from the conference and you should keep a look out for these on Channel Intel at YouTube. Thanks for stopping by The Server Room.

Jasper Forest: One sexy beast?

sarah.m.jacobson@intel.com — Tue, 17 Nov 2009 22:56:12 GMT

I have never heard our storage processors called sexy. Ever. We just recently announced the details of our next gen processor, Jasper Forest, which will officially start shipping in early 2010. We integrated a lot of key storage features into the Intel(R) Xeon(R) processor 5500 series and the storage industry seems to really like it. Our animation is posted on YouTube and one viewer called it a sexy beast. We also had an executive from HP, Dave Roberson, Senior Vice President and General Manager for the Storage Works division talk about how HP is designing it into future storage products. Check out the video, we are pretty excited about it!

Isilon and Intel, delivering high performance scalable storage

sarah.m.jacobson@intel.com — Tue, 17 Nov 2009 18:01:08 GMT

I attended VMWorld in San Francisco and captured some video on Isilon, a great Intel-based scale-out solution. The first link is John Gallagher, Director of Product Marketing at Isilon, giving an overview of their products and how Intel adds to their solution. He also talks about some of their more successful markets.

The second link is a chalktalk provided by Nick Kirsch, Senior Product Manager at Isilon, in which he discusses how Isilon storage delivers scale out storage for large scale server virtualization. I am also looking for any great Isilon success stories, so let me know!

Why Did Intel IT Build Two SuperComputers?

webadmin@intel.com — Mon, 16 Nov 2009 18:09:27 GMT

In order to deliver to the continued promise of Moore’s Law, Intel’s Information Technology team needs to enable Intel’s Silicon designers with the tools, capabilities and streamlined processes to bring higher performing processors to market every year. The latest generation of 45nm products (ie the Intel microarchitecture, codenamed Nehalem) was an especially challenging project for us.

With Intel design computing demand growing an average of 45% year over year coupled with the rich technology capabilities in the 45nm based Nehalem micro-architecture, the computational requirements of silicon tape-out (the last stage of design before manufacturing) represented an approximate 13x increase in increase in demand from prior 65nm processors. Staring at this demand (1.2 million hours of compute demand per day) plus a need to bring products to market faster and more efficiently, our IT team realized we needed to do something different - our standard grid computing solution that was sufficient for earlier stage design work was insufficient for tape-out.

Solution: Intel IT built a High Performance Computing (HPC) solution that currently rank in the Top 500 list of supercomputers (#261, #308, Nov 09) and feature a new parallel storage environment to support our 45nm Silicon tape-out process. The details of this effort are captured in this whitepaper.

In summary, the Intel IT HPC solution employs two of the world’s fastest supercomputers to create the fastest microprocessors helping Intel achieve the following results.

· Completed 45nm tape-out in 10day, less than HALF the time of prior products

· Delivered an estimated incremental value of $44M to Intel

I can’t wait for what tomorrow will bring as Intel IT is already upgrading and evolving this HPC solution to support our future generations of micro-processor designs. Tune in tomorrow at SuperComputing 2009 in Portland where Shesha Krishnapura from Intel IT will present more details on our HPC environment or join us December 8th, 2009 from 10-12am PST for a live chat with Intel IT experts in the Server Room

Chris (twitter)

Why Are You Still Running Your CAD Application As 32 Bit Application?

webadmin@intel.com — Fri, 13 Nov 2009 19:33:48 GMT

Ok it may be that your IT department or enterprise applications are limiting your opportunity to adopt 64 bit version of your favorite CAD application, but your inability to adopt a 64 bit CAD application can be very limiting to your productivity.

Here as an example from a recent discussion with some end users who are involved in a workstation pilot with Intel. When they moved to a 64 bit version of their favorite CAD application the time to open a 2.5GB file dropped from 20 minutes to less than 1.

Question - How many files does your engineering team open a day? What is the cost of the 20 minutes?

Customers operating in a 32 bit world are forced to work with smaller models. You knew that. And of course smaller file sizes will open faster.

But….

Rather than working with the chassis, engine and transmission in a single view, you will need to work each one independently. The results is you may miss a design interference, a misalignment or another obvious design issue, because you only had a partial view of the entire design. More rework and more delays.

Yes but... many of the enterprise applications you use are 32 bit and you need to have a 32 bit workstation environment in order to access these tools. That may have been true once, but with technology like Intel® Virtualization Technology for Directed I/O and Parallels™ Workstation Extreme software you now have the opportunity for an uncompromised workstation experience. You get all the benefits of a 64 bit CAD application and you can still work within a 32 bit environment when you need to. You can even pass the data between workstation environments.

Do not be too slow to adopt a 64 bit version of your favorite CAD application, just opening files faster and working with a complete design can make eth cost of a new workstation irrelevant.

To learn more about Intel® Xeon® based workstations visit www.intel.com\go\workstation