Sun has recently published a whitepaper that discusses how the Solaris OS will take advantage of the next generation Intel Xeon processor (codename Nehalem-EX) for expandable servers (4 sockets & greater).  Sun with over 20 years of experience in larger socket, core & threading capabilities is working to have the Solaris OS be ready to take advantage of the features & new capabilities of “Nehalem-EX”.  The three areas of collaboration for Solaris & Nehalem-EX are around  scalable performance; advanced reliability and energy efficiency between the specific features in Solaris and the next generation Intel Xeon processor.  Read this recently published whitepaper

I’m a bit late in relaying my thoughts from Intel’s Developer Forum (IDF), but there was definitely some excitement around virtualization and high performance networking that I wanted to get the word out about!

In the past I’ve shared some details about SR-IOV and the advantages you can gain by being able to present virtual LAN hardware to each Virtual Machine (VM), effectively avoiding the Hypervisor when presenting virtual devices to each VM.  The advantage of being able to do this is clear:  The less interaction in the networking stack there is from the hypervisor, the less processing overhead is required for the system process the data.

That’s all good because if you have a dual 10 Gigabit adapter, you can segregate those two physical pipes into perhaps 16 virtual pipes that get exposed to 16 VMs.  By segregating these LAN pipes at the hardware level with SR-IOV instead of using Hypervisor switching, the performance gains in both CPU utilization as well as maximum total throughput can be very large.  There were several demos at IDF with various configurations, but reductions in CPU utilization of 40% were possible coupled with dramatic improvement in throughput!

But there is unfortunately one minor complication that I didn’t mention in my last post on the topic of SR-IOV.  There is the little fact that when VMs move between physical boxes (a usage that is highly desired and commonplace these days) you run into some problems with this SR-IOV capability.  When the hypervisor owned the network hardware abstraction, the performance was worse, but the functionality was better because you could seamlessly migrate from one box to another and the virtualization application would handle the details.  But with SR-IOV, a new layer needs to be added so that the direct hardware connection between the VM and the LAN hardware can be moved to a new box.

The really exciting part of IDF demos that I saw was the demonstration not just of the SR-IOV functionality on multiple hardware and virtualization configurations, but that these demonstrations also showed updated software from two virtualization vendors allowing mobility of the VMs while supporting SR-IOV! 

There was a demo on Dell systems showing this fully functional SR-IOV implementation with Citrix’s Virtualization suite.  There were two separate demonstrations on Dell systems, with VMWare displaying their new Network Plug-In Architecture (NPIA) solution that allows for the migration of SR-IOV connected VMs seamlessly between servers.

For those hungry for more detail, I’ve included the three SR-IOV demonstration videos here:

The first is the Citrix demonstration on Dell and Intel hardware of SR-IOV with VM mobility:

These next two are two videos are demos on Dell and Intel hardware with VMWare and their NPIA software implementation.

Each virtualization demo shows the massive performance benefits under various workloads when moving from Hypervisor based LAN segregation to SR-IOV implementation.  But most importantly, each demonstration proves out the capability to migrate VMs between physical hardware.  The only system hardware requirement is that the server itself supports VT-d.  If the networking hardware in the newly migrated-to box supports SR-IOV you get better performance, and if not, the solution falls back on the legacy Hypervisor virtualization.  Backwards compatibility is maintained!

I didn’t get firm details on when this full support for SR-IOV and migration will be available in Citrix and VMWare’s releases, but the demos looked pretty clean, and hopefully these suites will be available soon with this new functionality.  The LAN and Server hardware ecosystems are ready today, and it looks like the software vendors are just around the corner.  Virtualization momentum continues!

While virtualization was the big takeaway for me from IDF, there were also several other interesting demos for us networking hounds.  I’ve linked a couple videos of them below for anyone still thirsting for more of the latest networking technology and performance details!

The first video is a demonstration of Intel’s 82599 10 Gigabit Ethernet-based adapter card with Fiber Channel over Ethernet (FCoE) support.  Storage and Ethernet together at last!

The second video is a demonstration of Intel’s NetEffect 10 Gigabit Ethernet card publishing 1 million messages per second in a simulated NYSE floor trading scenario.  Oh yeah, only 35uS of latency.  That is fast.

So although I am two weeks after IDF, I hope some of you got a little taste of the networking excitement that took place.   Industry wide, hardware and software vendors alike are delivering ultra high performance low latency applications for the financial services industry, as well as mainstream performance increases for virtualization.  The performance and technology beat moves forward.  Exciting times!

--

Ben Hacker

Since we started the Ask An Expert discussion thread in the Server Room a couple years ago, I found that the community often asked for guidance between selection of server system type and processor number as IT professionals sought to make the best purchase for them.

As I responded to these threads, I realized there were a lot of the same questions occurring over and over again.  I then thought that having a selection tool to allow the community to guide themselves through a few questions to help narrow the options might be a valuable.

Sometimes the world (ok Intel) moves too slowly for me.  My brainchild on this was something I wanted to have done about a year ago with the first 45nm quad-core processors (Xeon 5400).  However, our server and corporate marketing teams got a little distracted by the Xeon 5500 (Nehalem) processor launch.

However, after much delay I’m proud to introduce this simple, interactive Xeon Server processor selector tool that can help you choose which server system type and processor would be ideal for your application and business goals.  With Three Easy Steps, you can narrow your choices.          

• Step 1: Identify the business environment, application type and primary purchase criteria
• Step 2: Compare and Choose the processor family (7000, 5000, 3000)
• Step 3: Compare and Choose the specific processor within that family

In this 3^rd step you can look at price, performance, power and feature set across multiple CPUs to help you narrow.  Take a short cut and look at the most popular CPUs or expand your options and look at the whole range of offerings.

We also have a Workstation Selection Tool (this tool was what  triggered the idea to create a server one)

Other IT and business value assessment tools from Intel include:

• Intel Xeon Server Refresh Savings Estimator
• Intel RISC Migration TCO Calculator
• PC (Client) Refresh ROI Calculator

Chris

Follow me on twitter

Intel Intelligent Power Node Manager is a new technology that is available with the Xeon 5500 Series Platforms released earlier this year.  Many of you have asked me questions via Twitter (@Toadster) about "How can I use Node Manager?" - so I wanted to present some simple use cases to simplify the explanation of Node Manager and how you can best use the technology in your own enterprise.

First of all, let's explain the growth problem at hand.  As servers shrink in size, the density of each server 'footprint' is growing from a power perspective... a few years ago, a single 42U rack could hold about 21 servers (estimating 2U servers) - and usually hosting one or two apps/servers per physical server, depending on if you had single or dual-socket servers.  In modern datacenters, that same 42U rack can hold 42 servers (1U each) with 2P per server - so you have an immediate density increase of 2X the # of servers, and 2-4X the number of sockets - which can equate to 16X the number  processor threads per rack...  one good thing is that Intel has been developing newer technologies to keep the TDP of each CPU roughly the same over the same time period between processor updates... where you used to have 2 or 4 cores, you now have 8 to 16 cores at the same thermal envelope!

Knowing how much power your platform uses is a key factor in populating racks and rows in your datacenter.  Prior to Node Manager technology, most Datacenter Managers would base rack population on 'nameplate' power - or the (W) rating on your power supply.  That's the 'max' power utilized by the platform, and what the PSU is rated for (worst case).  See the image below...

As you can see - using Intel Intelligent Power Node Manager technology, you can view your system's power utilization in real-time using Intel Datacenter Manager and the administrator can implement the power caps to ensure your server rack stays within your required power limits.  By utilizing the 'actual' power limits instead of nameplate power, you can increase your rack density thereby increasing your ROI, and decrease your TCO!  Lets face it - everyone loves saving money!

Many of us are familiar with this next scenario... it's summertime, and the power company is announcing that the power grid is under strain.  Personal homes start having their A/C cut-off to save the power grid from brown-outs...  now your enterprise can help reduce those risks as well!

Over the next few weeks, I hope to post more blogs/videos:

1. Single Node Power Monitoring & Management
2. Group/Rack Power Monitoring & Management
3. Thermal Monitoring & Management

Please provide some feedback, and post your questions and ideas for upcoming blogs!

You’ve seen it on the front pages of the papers lately.  The program that offers consumers incentives to trade in older used cars for more fuel-efficient new cars is pushing auto sales into overdrive.  The $1B in govt. funding for it was burned through in less than a week. The U.S. House of Representatives rushed through an additional $2B in emergency funds just to keep the program going, but will need Senate approval if it’s going to extend beyond Tuesday August 4^th. My guess is to make a continuation of the program palatable to the U.S. taxpayer, the incentive will need to be cut (from $4500 for a new fuel-efficient car to somewhere in the $1-2k range) but it’s great to seen people buying cars and stimulating part of the economy – while getting older fuel-inefficient cars off the roads.

I saw an interesting article talking about whether a similar program for servers would work…and though I think it’s a creative idea, I’ll argue that Intel and our OEM partners have been offering “Cash for Clunkers” for quite some time now – without any U.S. taxpayer help.  How? Through promoting the benefits of server refresh, a strategy that is proving to be one of the most beneficial investments to IT and business. Using the Xeon ROI Estimator I spent 2-3 minutes modeling potential savings by comparing 4-year old 2P Intel Xeon based servers to new 2P Intel Xeon 5500 based servers – and this is what I found:

An investment in one Intel Xeon 5500 based server (~$8.5k including purchase price, migration cost, and software validation) enables up to 10x performance per server, a 10:1 server consolidation opportunity vs. 10 older servers purchased 4 years ago that as an IT manager I can now get rid of.  So where’s the cash for the clunkers? Well, I would save over $4k a year in energy costs and over $11k a year in server / software maintenance costs by cutting out the old and putting in the new.  The 4-year total savings is about $38k, with a break even period of about 9 months. Not bad…and that doesn’t even take into consideration software licensing costs that I probably can save by cutting down the server count. Try modeling this yourself and check out the new PowerPoint report that you can generate from it – really explains the benefits in a way that the finance and facilities folks will find useful.

I also found this link that explains why Intel IT decided to move ahead with server refresh in 2009 after current economic conditions forced Intel to re-evaluate the strategy. Analysis found that delaying server refresh for a year would increase costs by USD 19 million.

And a refresh strategy also applies to the bigger 4 Socket and above servers as well, as documented in this server refresh brief. 

Server Refresh is a strategic investment for IT – the cash for clunkers program that keeps on giving.

bryce

I wrote a while back about how the Xeon 7400(Dunnington) processor series compared to RISC. Since then I have shared information through other blog posts and sharing content about how Xeon 7400 and Xeon 5500 will compare to both SPARC and POWER.

Xeon 7400 and Xeon 5500 are the current products shipping into the marketplace today. I.M.H.O they offer a pretty compelling alternative from both a performance and TCO perspective Vs SPARC and POWER. But I will not try and repeat all the reasons here

What I wanted to share with you was some thoughts about what the next product to succeed Xeon 7400 will bring to the RISC party. Nehalem-EX is the code-name for our next generation of product designed to serve workloads currently serviced by Xeon 7400 today (i.e. Database, ERP,  BI etc). EX btw is what we all would traditionally call MP or multi processor servers

Don't stop reading now, here is why I'm EXCITED about what Nehalem-EX will bring to the RISC party.

My excitement is actually based on real customer discussions about what Nehalem-EX will do for them and why it delivers some new stuff (my code for features and benefits) which they see as a pre-requisite to make the move from RISC to Xeon. For some customers the TCO and performance of  products have been enough to convince them to move. For some other customers there are still some checkboxes remaining which I believe Nehalem-EX will address

Here is a snapshot of some of the cool new stuff which is actually convincing customers (from some real deals that I have worked)

1. Improved bandwidth. Up to 9 times memory bandwidth of previous generations
2. Introduction of Quickpath Interconnects to the EX systems
3. Add new RAS features previously seen on Itanium products to Xeon products
4. Significant improvement in performance vs previous generations e.g. Database 2.5xe
5. More scalable platforms through 8 OEMs offering >8S. These platforms are key to manage large databases and for large scale consolidation
6. Mainframe class availability in scalable platforms

For more information check out the press briefing from May. See more the details in the presentation

Nehalem-EX goes into production later this year and I am pretty excited about how it will change the game. What do you think?

     Your first server, that is. There’s nothing like a real server to help your business become more competitive. While some small businesses can get away with using a desktop as a server, there’s really no substitute for the real thing. With the ability to more efficiently handle more users, accommodate the latest applications, and deliver greater reliability, having a real server will make all the difference.

     Aren’t you ready for the real thing? A server built on an Intel® Xeon® processor has a lot to offer your business, so be prepared for dramatic performance and productivity improvement. If you want to be able to handle the demands of more customers, more data and more staff, an Intel Xeon processor based server is the way to go. 

Can you afford downtime? Of course not! Maximize business uptime with technology that’s ready to work all day, every day.  And protect your critical digital assets with error correcting memory and support for RAID storage.

     And just to build the IT excitement for your first server, check out this animation to see what a real server can do for your business:

     So, if you’re flirting with transitioning to a real server, I would just advise that you make sure that your first really is the best.  Talk to your IT solutions provider [ http://premierlocator.intel.com/] about implementing an Intel Xeon processor-based server. And remember, once you go Xeon, you’ll never go back!

I have been around the supercomputing market for over 25 years and have had an opportunity to see some interesting ideas come and go.  Let me share two that I experienced firsthand. 

·         CDC’s Cyber205 or a Cray 1S.  The CRAY-IS and the CDC Cyber 205 both offered effective vector processing, however, code conversion between them may have required some significant algorithmic changes. Cray of course won the HPC race at that time.  Note, the Cyber 205 was a tremendous performer, when you could keep their extreme.ly long vector pipeline busy. However, one branch or gap in the vector processing pipeline would cause a flush of the vector unit and what performance advantage you appeared to have vs. a Cray 1S was quickly erased.

·         An early day accelerator was Floating Point Systems.  In particular the FPS 164 was an awesome “off load” system where the needs of a few users were satisfied with better throughput than the Cray X-MP and Y-MP of the day. Convex, had a better idea.  It was better at serving the needs of more than an FPS 164 and was simpler to develop, maintain and scale software to next generation systems.

So what are the lessons from history? Perhaps it is that there it is there is a tight connection between application, architectures and algorithms and that it is extremely important to maintain a level of application flexibility and versatility in order to adopt new architectures as they become available in the market.  The old adage still remains true, software will outlive the useful life of hardware.  So it is important to be able to quickly adapt new shifts.

The same questions probably still apply today as they did when Cray, CDC and FPS were around.

When does an accelerator computing strategy work best?

The easiest answer is if your application is extremely data parallel in nature, then it may be well suited for an accelerator strategy. The word extremely is the critical part. 

If your application only performs some level of data parallelism and includes task, thread and cluster level parallelism or contains a small fraction of branching or is host to irregular data sizes, then perhaps an accelerator may not be the best fit.

How much real performance will an accelerator strategy deliver? 

Often times we hear claims of 10X, 20X or even greater than 30X. 

These are great headlines, but as many have noted, you need to understand an accelerators impact on the total execution time of your application.  What may have been 10X to 30X or more on a kernel of the application may only deliver a mere 2X to 3X or even less in terms of total application performance improvement.

Of course the real question is what are we really comparing performance speed ups to?

I have seen well tuned software on accelerators compared to “baseline” code running on one core of an old processor.  However, when you use available software technology and turn compiler flags on and add in a math kernel library call the performance on multi-core solutions can jump by over 10X and in some cases can exceed 30X multiples for total execution time.  This standards based accelerated software will scale forward as newer microarchitectures are made available from Intel.

Why is the difference between the promise and the actual performance so great? 

Always a good questionJ. 

The promise deals with a small part or a kernel of the software that is data parallel and can potentially scale linearly as more compute resources are added.  Again if the application is extremely data parallel, then an accelerator strategy may be the correct approach.

However, when the actual performance result, or total application performance, is significantly different it is often because of several things. 

·         One common reason is that you may be comparing optimized software on multi-core systems to optimized software on an accelerator.  When I compare similarly optimized software on a multi-core system I see that 20 – 30X difference often fades to less than 2X  and in most cases better than hardware accelerators.  This is because optimized software on a multi-core solution accelerates all components of the application.

·         Another situation is the bandwidth imbalance of the attach points of the accelerators, typically the attach speeds do not match the memory bandwidth or the ALU speed on the accelerators and the theoretical peak flops are tough to achieve.  Sometimes, for larger workloads due to limited amount of memory on the accelerator card, performance deteriorates.

·         Another situation may be that your application depends on different forms of parallelism which include task, thread or cluster level parallelism and even in some cases sequential forms of your software

So back to the differences in performance between the Cray 1 and CDC Cyber 205.

While Cyber 205 was great at edges of science the Cray proved to be the workhorse of high performance computing.  It offered better system balance than the Cyber 205.  Here is an example, if you take great care to optimize your software for a particular architecture you will no doubt see tremendous performance gains.  However, like the Cyber 205, if you break that pipeline you need to pay for the overhead to restart the long vector pipeline.  Often times, even with today’s accelerators, that start up cost reduces what appears to be stellar performance gains of the Cyber 205 to being no better than, or sometimes, even slower than the Cray 1.  There were of course examples with the Cyber 205, as there is today with accelerators that demonstrate where select sciences can see tremendous advantages over traditional computing solutions.

What other considerations may weigh in your decision to adopt an accelerator strategy?

Are you constantly refining your software?
Many researchers would probably answer yes.  They are constantly refining their software to improve the results the performance or both.

As I mentioned at the beginning of the blog, the old adage still remains true, software will outlive the useful life of hardware.  So it is important to be able to quickly adapt to new shifts.  One way to simplify these moves is to use standards based tools which can give you the flexibility to create applications that can use the multiple types of parallelism mentioned above via tools, compilers, and libraries.  You may also want to use standards based tools to acquire the versatility you need in order to scale your software across multiple architectures – e.g. large, many and heterogeneous cores. 

The caveat with respect to using non standard tools is that you become locked into a specific architecture.  If that architecture from the same vendor would happen to change, you may be required to make some significant changes (e.g. tuning to grain sizes).

Do you want to maintain, support and update multiple code bases?
I don’t.  I want to invest n the development of parallel algorithms.  The old adage is that software will far out live any hardware implementation still applies and I need the flexibility and versatility to quickly and as painlessly as possible be able to adopt new architectures as they are made available.  I do not want to invest in maintaining, supporting and updating an ever increasing set of code streams as newer architectures are made available.

Our team goal at Intel is to develop software tools and hardware technology that can help you scale-forward your application performance to future platforms without requiring a massive rebuild – just drop-in a new runtime that is optimized for the new platform to experience the improvement (akin to the printer/display driver model, buy a new printer/display, install the respective driver, and your system enjoys improved benefits).  That is the goal.

If you want to learn more about what we are doing to deliver high performing HPC solutions that are both flexible and versatile please visit www.intel.com/go/hpc