I was measuring the parallelized part of the ACIS-SAT converter of CAD Exchanger (my personal project) to investigate its scalability. (SAT is a file format of ACIS, a well known 3D modeling kernel) The way to do this was to execute the same test case on the same architecture (e.g. Core 2 Duo or Core i7) changing the number of threads. Obviously, the ideal curve is 1/n where n is a number of cores. Below is the data I collected on a few files:

So why is that ? Discussing this issue with my colleagues at Intel, the first guess was that the application consumes all the bus bandwidth trying to load data from memory. That is, though the cores are free to execute the application cannot be fed it with data. We used another experimental Intel tool (Intel Performance Tuning Utility, available for free download at http://whatif.intel.com) to check this. This did not prove to be true ? the experiments showed that though there are a lot of objects in memory they are allocated quite locally and can fit into Level 2 cache and so there are no bottlenecks of working with RAM.

So what else then ? Another guess was HT and that the OS perhaps not optimally distribute the workload among the cores and that a work running on the same core (there are actually 4 physical cores) but in different threads start competing for data. So I took another machine with 2 processors (i.e. 8 physical cores) with disabled HT. The data slightly changed ? running in 5 threads gave speed up over 4 cores but then again, the time remained flat when increasing number of threads from 5 to 8.

To dive deeper into what happens I launched Intel Thread Profiler which (along with VTune Performance Analyzer) is a predecessor of Parallel Amplifier but has a nice feature which is not yet available in Amplifier ? timeline.

I only focused on the parallelized part (highlighted on the image below).

I guess if I were taking a more formalistic viewpoint, I might observe that social media services are available as first class entities in the OS, on a par with applications, panels, etc. This is a powerful basis of a metaphor for collaboration and innovation, since Moblin is an open source project and makes it straightforward to innovate new features and value.

And innovation around people's data is an essential quality of web 2.0. The best early example of this is the Google Maps web service. When people discovered that they could reverse-engineer the AJAX interface and integrate new applications with Maps data, suddenly Google Maps was mashed up with all kinds of new applications and ideas.

For Moblin, there is no reverse engineering required - it's an open source project, so it's it should be clear how to add more services or find new ways to integrate them into the OS.

How does Cloud Computing intersect with Moblin? It's actually something I have not considered much before, and I need to think about it more.

http://www.stevemcconnell.com/

At ICSE 2009 he was a keynote speaker and chose this provocative topic: ?10 Most Powerful Ideas in Software Engineering.? Was it a full house who listened to him? You bet. If you haven?t heard Steve talk about this particular topic, you should (see above URL). All this said, there were 2 of his 10 ideas that really resonated with me, and I want to spend a little time talking about those now:

Idea #1: Software Development Work is Performed by Human Beings, and,
Idea #9: Different Kinds of Software Call For Different Kinds of Software Development

Let?s go in order.

What are the human factors that might influence the performance of programmers? That?s a good question and according to Cocomo II (the COnstructive COst MOdel for software development, created by Barry Boehm) these factors include, from greatest financial impact to lesser:

Product Complexity, Time Constraints, Storage Constraints, Requirements Analysis Capability, General Programmer Capability, Turnover, Platform Volatility, Database size, Multi-site Development, Required Software Reliability, Documentation needs, Architecture and Risk Resolution

Cocomo II further sites human influences on entire projects has a cumulative effect based on experience, technology knowledge, business knowledge, and ones? own personal processes. I haven?t studied the math behind these findings but think about those results for a minute:

Cocomo II states human influences make a 22x difference in total project effort AND COST. Experience factors alone make a 3.0x difference; capability factors alone make a 3.5x difference.

Human Beings -- it turns out -- exhibit exactly the same variations in individual performance that programmers do (duh). What does this mean to my teams or yours? Glad you asked:
o Recruiting top staff is easily cost justified
o Elaborate staff retention programs are also easily cost justified
o Technical successes by companies such as Amazon, Google, and Microsoft are not accidents (!)

Ok so fair enough: Mr. McConnell shows us by means of Cocomo data that it makes a lot of sense to get the right people for the right software job; in fact, it?s one of the 10 most important things you can do. Does this mean that all jobs are created equal, that all software projects can benefit from the use of one, single development approach that is a guaranteed ?one size fits all? success?

Some say yes: McConnell says absolutely not. There can never be one single development approach that works BEST for all projects.

The image that he uses to illustrate this point is a striking one, or rather, a sawing one...

McConnell asks, suppose you need a saw to get some work done? What saw will work. He then present a series of 11 saws asking, would this one work for you? This one? What about that one? Those? Of course some are electrical, many are not; at least one is gasoline driven; several very small for fine and delicate work; others, large, designed for the largest tree trunks.
Each is clearly a saw, designed to do the work of a saw, but for VERY different circumstances. He tells us software development is as difficult as sawing, tongue firmly planted acheek: and yet, the point is extremely well made. The wrong saw (approach) in a given situation would result in catastrophic disaster.

Matching the right approach (saw) to the job is paramount: again in fact, one of the 10 most important things you can decide. He invites us to take this conclusion to the bank (a certainty): ?Which software development approach works best depends on the kind of project.?

You can watch Parallel Programming Talk every week on Intel Software Network Television (ISNTV)

Threading Challenge Update

• Set #1 - Denghui0815
• Set #2 - Haojn
• Set #3 - Denghui0815
• Set #4 - BradleyKuszmaul
• Set #5 ? Mataocilk.com
• Set #6 - BradleyKuszmaul
• Grand Prize Winner For Phase 1:  "akki"

Phase 2 is expected to begin August 24, 2009 - Don't miss your chance to test your threading skills and possibly win a prize in the next phase!

New Technical Content on the Parallel Programming Community

• How to sound like a Parallel Programming Expert Part 2 by Timothy Mattson (Intel)
• Optimizing Parallel Data Transformation by Andrew Binstock
• A trivial and proven method to implement application performance optimizations with Intel® Streaming SIMD Extensions 4.1 instruction by Eli Hernandez, Application Engineer (Software Performance)

Upcoming Events

• Aug 19-21 - UC Berkeley Par Lab Boot Camp

Listener Question Show

Our next listener question show is August 4 ? send in questions or idea for our show to parallelprogrammingtalk@intel.com.

Episode #40

"Is Parallel Programming Hard?" with Prof. Guy Blelloch of Carnegie Mellon University

Aaron and Michael talking with Dr. Blelloch about his views that with the right level of abstraction, Parallel Programming is not and in fact can be as easy as sequential programming. He does note that, however, that "at the right level of abstraction" should be considered with care.

Download a MP3 of of the show.

Download a video of the show.

Episode #41

Software Development for Embedded Multi-core Systems; A Practical Guide Using Embedded Intel Architecture by Max Domeika, Senior Software Engineer, Intel Corp.

Download an MP3 of the show.

Download a video of the show.

The multi-core revolution has reached the deployment stage in embedded systems ranging from small ultramobile devices to large telecommunication servers. The transition from single to multicore processors, motivated by the need to increase performance while conserving power, has placed great responsibility on the shoulders of software engineers. In this new embedded multicore era, the toughest task is the development of code to support more sophisticated systems. This book provides embedded engineers with solid grounding in the skills required to develop software targeting multi-core processors. Within the text, the author undertakes an in-depth exploration of performance analysis, and a close-up look at the tools of the trade. Both general multicore design principles and processor-specific optimization techniques are revealed. Detailed coverage of critical issues for multicore employment within embedded systems is provided, including the Threading Development Cycle, with discussions of analysis, design, development, debugging, and performance tuning of threaded applications. Software development techniques engendering optimal mobility and energy efficiency are highlighted through multiple case studies, which provide practical ?how-to? advice on implementing the latest multicore processors.

Finally, future trends are discussed, including terascale, speculative multithreading, transactional memory, interconnects, and the software-specific implications of these looming architectural developments.

August 4: Aaron and Clay will be talking taking a break, and Intel Software Network TV's Visualize This! show will be coming to you LIVE from SIGGRAPH09 in New Orleans. Tune in and get the latest on graphics, visual computing, and game development.

Dawn: I recently read on Engadget that you developed a Linux driver for the NEC USB 3.0 controller. Can you describe what other Linux work you do within the Open Source Technology Center (OTC) at Intel?

Sarah: The Linux driver is designed to work with all xHCI (USB 3.0) host controllers, not just the NEC host controller.

Within the Open Source Technology Center, I work on Linux USB support. That includes working on USB device power management. We're basically trying to suspend USB devices when they're idle, so that you don't drain your laptop battery when you plug in a USB mouse. It's challenging because some USB devices don't properly support auto-suspend, and we don't want to break any devices that already work under Linux.

Dawn: What is the most interesting story that you have about being a Linux Kernel hacker?

Sarah: I think the best part of being a Linux Kernel hacker is interacting with other developers at conferences. You can solve problems a lot faster in a 5 minute face-to-face chat than a day's worth of email exchanges. Plus you get to hang out after the conference is over. I got to play board games with the Linux graphics guys and Linus Torvalds once, which was a lot of fun.

Dawn: How did you get involved in the Portland State Aerospace Society?

Sarah: I got involved in the Portland State Aerospace Society (PSAS) because my then-boyfriend-now-husband dragged me to a meeting. At first I thought these people who were launching 13 foot rockets were just crazy! Then I decided they were cool, and I wanted to join the group. I wasn't sure where to jump in on a technical project, so I became PSAS' student group president and handled the group's finances and marketing for a year.

Once I knew everyone and was exposed to several projects, I worked on creating USB sensor nodes for my senior capstone project. The USB nodes pass data from GPS, IMUs, and other sensors to the flight computer on board, which then decides when to deploy the parachutes. Check out my Linux Journal article if you want more info on that project.

Dawn: What are the advantages of using open source/open hardware to build rockets?

Sarah: PSAS has everything on our wiki, including board schematics, design documents, software, and analysis of failed launches. There are several advantages of keeping all that information open and using only open source.

An advantage of using open source software and hardware with fully-open specifications and firmware is there's no gotchas. If there's a bug in the firmware for our wireless card, we might be able to fix it. If the open source driver for the card is accessing a register wrong, we can check the open spec and contribute a patch to the maintainer.

One obvious advantage to having all our software and hardware designs openly available on our website is it's easy for new members to find information and ramp up on a project. We also learn from any group that takes a design from our website and uses it in their own project. If another group wanted to take our flight computer software and run it on their own rocket, our group would learn from that experiment and be able to improve our code.

It's also beneficial to the community because it can save other rocket developers time and money. For example, we spent 3 years developing a spread sheet to generate a board layout for a cylindrical patch antenna, based on desired transmission frequencies. A design company quoted us $30,000 to design and build one patch antenna. With our spread sheet, it costs about $500.

Actually it's David Perry talking about the future of video gaming. So please join the show and watch and listen closely to what David is sharing with us. And please mark your calender already as we will videoblog during the whole conference live from Cologne trade fair center. Hopefully with a lot of interesting chats, games and free drinks. So, load your RSS feed and follow us. See ya at GDC'09!

Our topic today is the Kaboom project, and the new whitepaper multi threaded fluid simulation for games. Joining us today are Jeff Freeman and Quentin Froemke . Jeff and Quentin are software engineers in the Visual computing software division at Intel.

Direct download link to the high quality MP4 video file (250 MB)

News :

Mark Randel President and CTO of Terminal Reality started a blog series on the Highlights and Challenges during the development of Ghostbusters. You can read the fascinating journey at http://software.intel.com/en-us/blogs/author/mark-randel/

Thread like wildfire continues : send in your entries by the last day of the month and be entered to win 1 of the 5 game stop gift certificates. More details can be found at http:///www.intel.com/software/thread-like-wildfire

Jeff and Quentin talked to the below questions:
1. This year at GDC San Francisco you spoke about the Kaboom project and recently released the source code and white paper on it. Can you speak to what is the goal behind this project?
2. You mentioned that this project shows how to add real-time 3D smoke, fog and other fluid simulations to your game. How is this different from traditional methods of rendering smoke and fog?
3. Tell us how you were able to use the Navier-Stokes equations in this project?
4. You based this paper and work on Mick West's paper "Practical Fluid Dynamics". How did you use this?
5. What kind of hardware are you using for this project and what kind of frame rates are you seeing?
6. Are there other Intel tools used?
7. The message you have with this demo is proper threading for multi core architectures will help you reach the full potential of your games. What have you included in the source code that helps a developer do this?
8. Where can we go for more information : http://software.intel.com/en-us/articles/multi-threaded-flui.....for-games/

Join me in 2 weeks for another episode of Visualize this!

If you have ever parallelized a serial program, you know how tricky it can be to answer these questions! With the move to multi-core processors, the need to parallelize software is becoming somewhat inevitable. Those who chose to continue to ship serial code are potentially exposing themselves to obsolescence. Parallelization is becoming essential for survival in this tough marketplace.

Moving from an existing serial code base to efficient and correct parallel code is often difficult, error prone and expensive. Using intuition to identify the parallelization opportunities and then ascertaining that the parallelization was done correctly can be hit-or-miss. The whole experience can be improved using a tool like Intel® Parallel Advisor Lite ? which has been built for this exact problem.

Parallel Advisor Lite is a parallel modeler for your serial application ? you would use this tool to explore or model parallel possibilities while keeping your application serial. This is important because a parallel modeler lets you continue to use your existing test system and debugging methods as you modify your application to resolve data access issues.

The Parallel Advisor Lite workflow guides you through the process of

Tell us what you think of Parallel Advisor Lite - we appreciate your feedback! Did it lessen a pain or solve a problem you encountered on the road to effective parallelization? How can it be improved?

p.s. Did you know that Parallel Advisor Lite is not tied to a threading paradigm? Or that you can build with either the Microsoft® VC++ compiler or the Intel® C++ compiler?

This link brings up a site where you can get information on viewing the International Space Station along with the Space Shuttle when it?s flying too.  Currently they are separating and thus will soon be visible as 2 separate objects (they are brighter when together).

http://spaceflight.nasa.gov/realdata/sightings/

Perhaps it?s the nerdy engineer in me but watching that fly by many times now and knowing there are 13 people up there is an amazing thing.  If you can take some time and take a peek, you?ll be surprised at just how large it is, and far brighter than any star in the sky!

I wonder if there?s a way to find out how many Intel processors are used on the Space Shuttle and International Space Station?

My guess is that along with live audio and video feeds, eventually we'll also be able to follow the Astronauts working in the Space Station via social networks!

With Windows Server 2008 R2 installed and configured with the Hyper-V role you'll need to open the Hyper-V Manager which will initially have no virtual machines as shown here.

If you then click the Import button the virtual machine will be copied to the virtual hard drive folder defined for your Hyper-V environment. Within Windows Server 2008 R2 the default location is:

C:\Users\Public\Documents\Hyper-V\Virtual hard disks\

It can take some time to import the virtual machine and so it might be a good time to get some coffee while the virtual hard drive is copied.

Given that the virtual machine was imported from Windows Server 2008 to Windows Server 2008 R2 you'll also have to update the integration services within the virtual machine. You can do this simply by selecting the Insert Integration Services Setup Disk option within the Hyper-V Virtual Machine Connection.

Essentially that is all that is involved in moving virtual machines from Windows Server 2008 to Windows Server 2008 R2.

Setting the Consolas font as the default command prompt font is done within the Font tab of the command prompt properties Window, shown here.

These related markets (smartphones, MIDs and netbooks) have been a sweet spot for open source software. Sam Dean from GigaOM says that "open source operating systems are suddenly flourishing, especially in the growing netbook and smartphone arenas, and it looks like significant competition between them will lead to much innovation."

Juniper Research is currently forecasting an increase from 106 million this year to 223 million by 2014 for smartphones running open source operating systems. Juniper also sees open source operating system-based smartphones as an additional outlet for contributions from Linux and open source developers.

Jay Lyman from The 451 Group sees Moblin as the middle man: "I see all of this headed to a place where Moblin rests below a variety of other software that is more specialized to the particular device, whether it is a smartphone, a netbook, a tablet PC or something else." Dirk Hohndel talked about netbooks and the Moblin project as part of his session on Intel and open source software at OSCON.

Regardless of what you call this segment or where you draw the line between these markets, open source is playing a role. It is impacting innovation, developer communities and more. While these markets have existed for a while, they are now starting to become more mainstream with smart phones and netbooks moving out of the realm of early adopter geeks and into wider usage. Open source isn't the only choice in these markets, but it is becoming an increasingly important part of the solution.

What do you envision as the future of open source in smartphones, Mobile Internet Devices (MIDs) and netbooks?