♣ Pesquisar ♣

What's in a name? That which we call a contest, by any other name would succeed as gloriously.

We just successfully wrote and ran a student contest for TeraGrid ?09. It was an exhausting satisfying experience. The questions we came up with are at http://wiki.sc-education.org/index.php/Tg09-student-contest. We had nine fascinating diverse teams ranging from an all high school team through an all grad school one, with many interesting permutations and combinations of high school, college and grad school students forming the other teams.

In a previous blog-post, we ended up talking about diversity of problem solving ability. Check out the diversity of the competing students: http://contracosta.edu/cs/tg09studentCompetition.jpg. An even greater diversity is seen in the pictures of all the students attending TG09 at http://contracosta.edu/cs/tg09Students.jpg: age, color of skin, color of shirt, hues of enthusiasm and ideas. Heady stuff, that. My team from Contra Costa College also competed: http://contracosta.edu/cs/tg09CCCstudentTeam.JPG. I am very proud of them and told them so. They asked if that meant they had scored well in the 9 hour-long contest in which they'd competed the previous day. I told them we hadn't started the grading, which due to busy hours, would ultimately not complete till late into the night preceding the awards ceremony. They we're still confusing grades/awards with recognition of their outstanding dedicated effort. We have a lot more time to get that one straight.

The Clemson team was coached by Brian Dean, an enlightened teacher, in many interesting respects, including a clever way of generating a flash animation of narrated class notes. He also leads the USA Computing Olympiad. Coincidentally, I'd been brainstorming with Alejandro Queiruga from CMU, a name not only resonant with that of Alonso Quijano, but also a young look-alike, complete with a wonderfully developed productive madness. I know it was productive because he suggested a number of interesting follow-on competitions for us to consider. The one that caught my attention was embedded systems programming. Boy did this catch my attention. What a clever way to return to the non-bloated (memory/disk/executable) days of computing, where performance counted. A jaunt playing games and watching movie clips with a few other students till 3am, led us Alejandro and I to the refinement of the cell processor being a great basis for the contest. It is a gloriously difficult programming environment including heterogeneous processors using a non-standard OS. Perfect. When I pitched it to Brian the next day he correctly noted that it would make things almost impossible for the competing high school students, which actually what attracted me in the first place. I have a lot of experience assigning problems to students a working professional would correctly identify as difficult to impossible. The trick is to give the students no clue of that, including the unnecessary burden that it can't be done. It frees them to more easily solve the problem.

I am hoping we can work out a compromise for I think helping students gain a working hands-on knowledge of squeezing cycles from reluctant silicon is not only great fun, but worthwhile life skills for a CS professional.

By the way, Alejandro, here are the words as I remember hearing them in a theater in San Francisco when I was roughly your age, words Wasserman put in Cervantes mouth I suggested you memorize: "I have lived for over forty years, and I have seen life as it is: pain, misery, hunger, cruelty beyond belief. I have heard the singing from taverns and the moans from bundles of filth on the street. I have been a soldier and seen my comrades fall in battle, or die more slowly under the lash in Africa. I have held them in my arms at the final moment. These were men who saw life as it is, yet they died despairing; no glory, no gallant last words, only their eyes filled with confusion, whimpering the question: 'Why?'. I do not believe they asked why they were dying, but why they had lived. When life itself seems lunatic, who knows where madness lies? Perhaps to be too practical is madness. To surrender dreams may be madness. To seek treasure where there is only trash. Too much sanity may be madness. And maddest of all to see life as it is, and not as it should be!"

So readers, what do you think would be a good basis for a computing oriented contest? I'll bet you don't come up with Alejandro?s other very clever idea. I'll leave it for his elaboration, should he choose to grace this blog.

May we all heed Cervantes' words, avoiding any unnecessary sanity, while crafting life as it should be. I won't contest that while seeking to build a contest around it.

Tribal Knowledge in Software Engineering

In software engineering we often talk of various ilities such as maintainability, securability, or scalability. All of these ilities are much easier to obtain, and to retain, if the software is effectively designed and implemented. 

The design and implementation of the software must then be effectively documented such that it can also be effectively communicated.

Why then do we continuously encounter software projects that instead prefer to rely upon tribal knowledge, stored only within the collective wetware of the development team.

Software should always be designed and implemented with maintainability foremost in the minds of architects and developers. I would argue that with a reliance upon tribal knowledge it is not possible to attain and retain the required level maintainability.

While you don't have to become fully compliant with CMMI level 5 to have an adequate level of maintainability, you should aim to reduce the reliance upon tribal knowledge,

Related Posts

1. Complexity within Software Engineering
2. Model-Driven Architecture and Development

What is the significance of 11.4 cents??

Did you know that using an electricity rate of 11.4 cents per kWh provides a simple method of calculating annual electricity cost of any device? 

1 watt of power consumption, with an electricity rate of 11.4 cents per kWh, cost $1 per year, assuming power usage remains constant.  Also, as a general rule of thumb, every 1W of device power consumption in a data center requires an additional 1 watt for overhead power (Source: Intel IT). So a device that consumes 1W actually consumes 2W of power at a data center level. 

Here's the math:  1 Watt power * 8760 hours per year / 1000 * $0.114 electricity rate per kWh = $1 per year.  This math holds the same for any currency, Euro, Yuan, etc.  11.4 cents per kWh is the crossover point?and as electricity rates increase over 11.4 cents, 1 watt will cost more than a $1 per year. 

The datacenter overhead power, often referred to as Power Usage Effectiveness (PUE) is a number which has emerged as the leading metric for data center energy efficiency.

You might say that 1W = $2 annually doesn't sound like much, but start doing the math for 1000 servers that consume 200W in a data center with a PUE of 2.0 which works out to annual electricity cost of ~$400,000 per year.  Now, for every 1 watt the server power consumption is reduced, this would translate into $2000 annual savings.  Note, this is a very rudimentary example, but it is useful to illustrate why customers are really starting to focus on power as one of their key purchase decisions.    

If you need energy efficient servers, there are multiple server vendors currently have some exceptional energy efficient products based on Intel(R) Xeon(R) 5500 processors.  And looking forward, we are also actively working on how to reduce power of the processor and at the system level for the upcoming generations of products. 

Here?s some good reference on electricity rates:

For United States, state by state electricity rate comparison. 

For Europe, 1st half of 2008 rate comparison by country.

Remember, power is one purchase decision, but it is not the only one.  A rack of servers that consumes less power that does less work isn't an efficient way of deploying servers either.  Ensure that the performance vector is considered.  Intel® Xeon® 5500 processor based servers provid exceptional performance and perf/watt leadership over the competition.

Quick question for you:  How does electricity rate of 11.4 cents per kWh and a data center PUE of 2.0 compare to your data center? 

Some recommended steps to take migrating from RISC -> IA

We talk a lot about how great the Intel Xeon processor compares vs. competing RISC architectures when it come to price and price/performance on various workloads, but unfortunately for many existing people running on RISC hardware, simply throwing out the old and standardizing on the shiny new Intel-based servers isn't always that simple of a proposition.  Why? Your existing software running on UNIX (i.e. AIX, Solaris) may be custom-coded on your flavor of UNIX, the source code may be lost, the guy who wrote retired 5 years ago, etc.  So, how do you account for this when 'running the numbers' to see if it makes sense to rid yourself of the power and money-sucking old RISC server collecting dust in the back of the data center?  These five steps may help:

1. (and compare vs buying new RISC hardware)This is the simple analysis that looks at performance of your existing system, compares it to new hardware and then factor in other significant cost items like power consumption, software licensing, software/hardware maintenance costs, etc.2.Meaning, look at all the software that actually runs on these servers, the packaged applications and the custom code.3.You may chose to do this on your own if it doesn't look too intimidating, but for more complicated migrations, likely you will need some external help.4. TestYou may only be able to test the 'easy stuff' initially, but verify the performance deltas between the new and old systems calculated in step 1 to correctly size how much hardware you will need in actual deployment.5. DeployIf you have your migration plan in place from step 3, now you execute according your plan, ensuring your migrate data in the right order to ensure minimal downtime. These steps can be very intimidating, many people in IT find it hard to justify the migration costs (particularly if you need to pay for some services), but taking a systematic approach to it and carefully calculating your ROI including these extra costs will often make it worth the effort.

My work made a difference!

The new iPhone sold a million units over its first weekend.  Facebook is so popular that it is now being used by businesses and adults, not just college students. Recently, Twitter has been one of the only methods of communication for getting out the news in certain world hot spots.  When I hear these things, I think what a difference the software developers who created these solutions or innovations made.  They must feel great satisfaction with their work.  How about you?  Have you ever felt that your work has made a difference with people, business or the world?  If yes, how about sharing with us?

The Future of Autonomic Computing Innovation

The computer industry is filled with pundits, speculators, visionaries, salesman, brilliant architects and professors. Each provides invaluable insight into their experience, their intelligence, their alma mater, their ticker symbol, their ego and what?s next. Some win the ?what?s next lottery?, others work for years of brilliance in relative obscurity.

Seemingly, a world that has deployed over 1 Billion devices a year for the last 3 years , is incapable of understanding the gravity of a new programming models, a new hardware architecture, a sleek new design that delivers on a vision that Gene Rodenberry thought of in the 1960?s or Da Vinci in the 15th Century. What is old is new?..and let me tell you why? It will revolutionize the industry (not evolutionize?a term reserved for slower growing industry?s that require government assistant every decade or so?), transform your environment and provide freedoms you had only hoped to enjoy?.and we invented it 40 years ago. Does any of this sound familiar?

It should. These are the paraphrased slogans of an industry in transition. Real products matters, product differentiation matters, standards matter, interoperability matters?.and shareholders pay for future expectations.

The future of computing?is NOW. The future of the computer industry is NOW. The next generation of computer programming, software architectures and transformational technologies is NOW. As an industry we have finally begun to embrace interface, architectural and software programming standards to usher in a new era of interoperability and scalability. Behind us are the days of ?proprietary interfaces? (What does that actually mean other than I am going to sell you some extra accessories that will be worthless in 2 years?), which do not provide a differentiated performance/cost advantage. Gone are the days of developing programming languages that lock-in customers to individual companies, whether vendors innovate or not. These rules of the past are slowly melting away, allowing the entire industry to embrace interoperability and standards at the highest level in history. Industry diversity is healthy and insures that the most innovative and technologically relevant companies will ?win? most of the time. Allowing the 1 Billion and the Next Billion customers of the world to enjoy the best interface technology yet developed?.each other.  It also provides us with a unique ability to move to the next phase in our dynamic industry?s growth, autonomic instrumentation.

At Intel, we are constantly working to develop the next great performance architecture, filled with new innovative ?goodies?, as our Chief Virtualization Architect Rich Uhlig calls them. These ?goodies? (a technical term that Rich borrowed from his nephew, I believe) come in the form of virtualization technologies (Intel VT-x, Intel VT-d and Intel VT-c), security technologies (Intel LT-SX), performance technologies (Hyper-Threading, Turbo Boost) and energy efficiency instrumentation (Node Manager and Data Center Manager). Soon they will also include differentiated services in the cloud which facilitate ease of use and growth for a host of vertical industries in need of innovation. The resulting architectures that emerge will be instrument rich, feature capable and as scalable as users are willing to pay for.

Why is this important? Instrumentation matters. As we apply business and personal rules to our growing compute environments it has become increasingly clear that the more tools we make available to users the better informed we are in making decisions. The more disclosure we provide to investors through the use of autonomic programming architectures the more informed they will be of their investing decisions.

How can you day trade $1B in 35 different stocks without clear autonomic controls in your data center, your database, your application and your client devices?

How can you move 450 Million people efficiecntly throughout a country for 2 weeks without autonomic controls on transportation: plains, trains, boats and automobiles, as they do during the Spring Festival in China?

How can you process 1 Billion text messages a day without clear business rules? What happens when these messages are also coming from machines to other machines, modifying databases, applications and clients?

As humans, we must apply guidelines, much like laws,  for our machines to take action when we are asleep, when we are tired, when we are not present, when we are just simply being human?.to slow to react to a rapidly changing environment.

The innovators of the computer industry today understand this NOW. We do not need to discuss a vision of 40 years ago without a plan to act NOW. Claiming ideas without action is dishonorable at best, criminal at worst. The innovators of today must build products and services that help solve the problems of today. We do not need to look to 2050 without a plan to act NOW. The visionaries of tomorrow are?..not born. The visionaries of today?can call me in 10 years.

Autonomic controls are in place today, machine to machine computer architectures are here today, scalable compute engines are here today. Are they perfect, no. Are they effective, yes. The design architects, product engineers and systems designers of today need to address these concerns. Autonomic Instrumentation delivers control to the administrator, the user and the developer. Rules engines can be modified to maximize efficiency, minimize consumption and increase productivity. All of these will lead to increase shareholder (read: No just people who buy shares of stock) value across your enterprise, your school, our hospitals, our governments, and your home.

When executed properly, Autonomic controls should be able to deliver 20-25% performance and efficiency increases with each new generation of Moore?s law. In some cases, as in the Intel Xeon® 5500 Series these increases have been over 150% in virtualization performance, these increases will be a combination of software architecture enhancement and silicon optimization. In other cases, it will be through the dedicated hard work of increase instrumentation capability of a processor platform at the same price of the previous generation through energy efficiency and memory controls.

Autonomic controls will also allow end users to avert disasters in our data centers, our homes and in our hands. Autonomic instrumentation design frameworks, allow users to set parameters on data migrations, data backup, security, memory access, power consumption and virtual machine architectures.

For Intel and our new Xeon® 5500 Series processor family, and our recently announced

Intel® Nehalem-EX platform provide the new generation of platform instrumentation. As product developers, designers and architects we should all find a way to increase the tools available to our customers to take advantage of these instrumentation capabilities. I look forward to being able to announce more of these new features as we announce them and help to provide development frameworks for developers, engineers and architects to build new products and services, ushering in the future of autonomic computing innovation?today.

Science and Math rock stars on The Tonight Show with Conan O'Brien

Science and Math rock stars hit the mainstream last night. The Tonight Show with Conan O’Brien featured finalists from the Intel International Science and Engineering Fair, held last month in Reno. Best thing about the segment, which was part of an Intel marketing relationship with the Tonight Show, was that the finalists were shown to be just like most young people: smart and fun. They described their projects to Conan, who appeared, for the most part, to not understand the sometimes long and complex titles. There was witty give-and-take and what appeared to be an orchestrated argument and group hug between participants. Generally, the sentiment that came across was that this was a great event celebrating what was an opportunity for ~1500 youth - out of an estimated 6 million+ that attempt to get to the world’s premiere pre-college science competition - to share their science and math, and maybe, just maybe, excite more youth to experiment with math and science and become the next innovators.

Conan also talked with Intel’s Chairman of the Board (at the time), Craig Barrett.

Comments (0)

Free shipping to US on boxed Intel® Parallel Studio and other software development products

Intel® Parallel Studio and many other software development products from Intel are now available in physical boxed format in addition to downloadable format from the Intel® Business Exchange software store.

FedEx 2Day shipping is free to 50 U.S. states. Customers outside the U.S. can purchase downloadable products.

HP Tech Forum '09: "Amazing ROI with Intel & HP"

I was out at HP Tech Forum last week and had a chance to catch up on all the latest technology advancements with HP and Intel. What I saw was staggering, over 17 new HP-Intel designs, the HP Performance Optimized Datacenter (POD), and lot's more that I will be sharing with you in coming days as I add more video from the event and help to tell the story if you couldn't be there. First off, I caught up with John McAtee from Intel's HP account team. He was showing a cool demonstration on why now is the right time to invest in XEON 5500 processor series technology. Check out this video and find out how you can start saving in your datacenter today !

If you want more information on how the XEON 5500 processor series can starting saving in the datacenter, check out this ROI Calculator tool. Also, if you are looking for detailed information or are just looking to gain more knowledge, you can always "Ask The Professor" in our Server Learning Center.

Did You See It? Intel Featured on the Tonight Show!

Did you see it?! I stayed up way past my bedtime last night to watch The Tonight Show with Conan O'Brien. Why? Because as part of a new sponsorship deal, Intel and the "Sponsors of Tomorrow" campaign will be woven into on-air and online components of Conan's new show! 

To kick off the sponsorship, Conan showed up in Reno NV to film at the Intel International Science and Engineering Fair. I was lucky enough to be in Reno and check out the filming live! Conan spent hours wading through the enormous show floor interviewing students and learning more about their projects.There were moments where Conan seemed truly awe-struck by the genius of the teens he was interviewing, and it's very apparent in last nights clip :) Conan also took some time to interview Craig Barrett:

 
The footage from Intel ISEF debuted during last night's show and I must say it was a very funny clip! 
As some of you may know, Conan is no stranger to Intel. Back in 2007 a similar sponsorship brought Conan to Santa Clara headquarters resulting in a handful of hilarious clips highlighting Intel's one of a kind personality! 
Both the clips from last night show and the clips from Conan's Intel visit can be seen on the new Intel section of The Tonight Show website: http://www.tonightshowwithconanobrien.com/intel/. 
Even though The Tonight Show starts at 11:30pm (which I already mentioned is way past my bedtime) I'm really looking forward to seeing more Intel and Sponsors of Tomorrow on the show! Thank goodness for DVR.

Twilight - Sun is setting for Solaris and AIX?

Sunsets can last a while, but in the end the sun will go down.  I talk to a lot of companies and listen to a lot of data center managers.  Customers trust their AIX-Power and Solaris-Sparc platforms.  These are solid platforms and deliver good features and reliability, but, if these managers could get the sense of security, performance, and reliability with Linux / Intel Xeon platforms, they would move tomorrow.  It is simple economics.

The reality is that customers are making this move, and being successful.  The hardware reliability on Intel platforms today is amazing.  Intel recently announced that their next generation of Xeon(Nehalem) EX based servers will support Machine Check Architecture.  This brings high end Xeon X86 servers into the RAS family previously reserved to proprietary RISC & mainframe platforms.  Intel Xeon already eclipses the performance of proprietary RISC processors both on a per processor basis and a per dollar basis.  It is reasonable to say Xeon can deliver better performance, better value, and equal or better reliability.  The only hurdle left is the software.

Linux has come a long ways.  It is no longer a university OS, run by geeky dudes in black T-shirts emblazoned with the quadratic formula.  It is mainstream and solidly supported.  Linux is the primary development and delivery platform for Oracle.  Other OS environments are ports, delaying support and innovation.  Linux is used by major financial companies.  Linux is available in solid and well supported distributions with a 20 year history of enterprise business.  Linux experts are broad community, worldwide, and growing in number.  Linux is economical vs proprietary RISC.

In an era of big budgets and conservative (donfailing to examine the opportunities for better performance and lower cost with Linux on Intel Xeon platforms is business negligence.Performance, Price, Infrastructure, Ecosystem, TCO, RAS

Want the Windows 7 RTM Build? Order Windows 7 Today...

Are you one of the millions of people around the world who have downloaded the Windows 7 beta and release candidate builds? Would you like a copy of the final RTM code?

Here is your chance now that Microsoft has announced that Windows 7 can be ordered beginning today and you can order your copy right now!!!

I have personally been running various builds of Windows 7 since October last year and have had an awesome experience with the operating system.

With Windows 7 supporting up to 256 logical processors all I need now is an HP SuperDome and I'll be in multi-core heaven.

Future Manageability Will Drive Virtualization Architectures

Based on Intel?s current processor core count and extrapolating from their ?Tick Tock? model for scheduled new CPU designs, by 2017 Intel could very well be designing a 20-Core CPU!  Do we need that functionality with so many single threaded applications in the market today?  Maybe not today, but in 8-10 years computing usage models are going to be a lot different than they are now.

Whether virtualization environments will be running on 20-Core Intel processors by then or not, one thing is very clear high end virtual environments will require much more powerful management environments that what we have today.  Management tools that now only look at high level performance metrics will need to look at detailed server component and CPU level power consumption, detailed Core level performance metrics, and managed thermal output.  The more finitely we can manage our virtual elements the greater control we will have at the server, rack, and data center level for optimizing data center virtual server density to the physical limitations of our data center environment.

With VMware?s latest server virtualization version, renamed vSphere, new manageability capabilities that increase the usability and decrease the cost of managing a virtual environment are included.  Although they don?t go to the level I just described, they do provide some nice improvements from version 3.5.  Storage and network optimizations have been added that allow hosts to power down when not needed using their power management (DPM) tool (which is now a fully supported feature and not just ?experimental?). VM Monitoring now uses VMware Tools to evaluate individual VM?s and check to see if they are running and Fault Tolerance now ensures continuous availability for virtual machines against hardware failures.

Intel is driving the creation of not just multi-core CPU?s but the tools that will drive virtualization architecture adoption in the future.  Providing powerful tools that manage power, thermal, and performance to help make our lives as data center operations personnel easier and make the value proposition of virtualization that much greater.  Management definitely will continue to be a key component to determining TCO in the future.  See the following around what Intel is doing around management. (http://www.intel.com/design/servers/ism/sms.htm)

What are the key management tools, in your opinion, that drive virtualization adoption, those ?can?t do without? management apps?

Mark

Setting up a Network File System (NFS) Share:

One of the reasons I setup an NFS share is so that I can have an inexpensive shared storage place for my VMs on the ESX Server 4.0. I can also use the features on vCenter like VMotion and Fault Tolerance using the shared storage. Creating the NFS is only available for the Windows 2003 R2 SP2. For ESX Server?s vCenter, you will also need to install the 32-bit version of the OS. I have a desktop client which I have installed the OS and also where I will store the VMs at. I also have vCenter & vSphere Client installed on the system as well. The actual server doesn?t have any VMs locally stored on it, but ESX knows to use all available resources on the SERVER to run any VMs applications instead of using the client?s resources.

If you have Windows 2003 Sp2 32-bit installed, you can add the NFS portion. Make sure you have the 2nd CD in.
1. Go to the Add windows components --> Other network files & print services --> Microsoft services for NFS, and check everything there to install

2. On your hard drive on the client system, create a etc folder and a Share folder (You can choose whatever name you want)
3. Go to your ESX server and grab the group and password files in the /etc folder. You can use WinSCP (Google & download) to grab files from a Linux environment to a Windows OS. Place the 2 files in the /etc folder you created on the client HD.
4. Under Start --> Administrator Tools --> Microsoft services for Network File system, right-click on user name mapping.
a. Point to the path of where you have the group and password files from ESX server

b. Open up the use name mapping tree and right-click on user maps to create a new user map
i. Choose the Administrator for windows and root for Unix and Add the mapping

5. Go to the Share folder you created w/ the etc folder
a. Go to the NFS sharing tab ? Permissions, check Allow root access and change to read-write & check the Allow root access like the picture below

6. On the vSphere GUI, click on your server IP/name and go to Configuration tab ? Networking, and properties. Add a VMkernel and type in an IP for this kernel. This is required to use an NFS share. You can also enable this kernel for the VMotion or Fault Tolerance features in vSphere.
7. On your vSphere client GUI, go to the configuration tab of your server and add the NFS storage under the Storage section. This is case-sensitive. Point to the IP on your client and the folder name (/Share).
Note: The NFSShare folder should have a green arrow showing the NFS sharing.

The NFSshare is good to store iso images, or just to get started on VMotion/FT or other features in ESX Server that requires a shared storage. However if you?re going to run heavy duty benchmarks on the VM, the performance may suffer in the NFS storage.

Virtualization and Performance: Understanding VM Exits

VM exits in response to certain instructions and events (e.g., page fault) are a key source of performance degredation in a virtualized systems. But have you ever wondered why? What exactly happens during a VM exit anyway?

A VM exit marks the point at which a transition is made between the VM currently running and the VMM (hypervisor) which must exercise system control for a particular reason. In general, the processor must save a snapshot of the VM's state as it was running at the time of the exit. For Intel architectures, here is an approximation of the steps:

1. Record information about the cause of the VM exit in the VM-exit information fields (exit reason, exit qualification, guest address), and update VM-entry control fields.

2. Save processor state in the guest state area. This includes control registers, debug registers, MSRs (see next item), segment registers, descriptor-table registers, RIP, RSP, and RFLAGS, as well as non-register state like pending debug exceptions.

3. Save MSRs in the VM-exit MSR-store area. MSR stands for Machine Specific Registers, in case you are not familar. They are used to control and report on processor performance.

4. Load processor state based on the host-state area and some VM-exit controls. This includes host control registers, debug registers, MSRs, host table and descriptor-table registers, RIP, RSP, and RFLAGS, page-directory pointer table entries, as well as non-register state.

5. Load MSRs from the VM-exit MSR-load area.

Oh, and don't forget that after the VMM has performed its system management function, then a corresponding VM entry will be performed that transitions processor control from the VMM to the VM!

Now you can see why VM exits generate considerable overhead, to the tune of hundreds or thousands of cycles for a single transition. To mitigate this problem, BTW, considerable effort has gone in to both reducing the number of cycles required by a single trasition, and creating features that obviate the need for system management (and hence VM exits) in the first place.

For more information, see the "Intel 64 and IA-32 Architectures Software Developer's Manual" at http://download.intel.com/design/processor/manuals/. In particular, take a look at chapter 23 in Volume 3B: System Programming Guide, Part 2.

David Ott