Stories from September 28th, 2010

GTC2010 In Review

Now that GTC has come to a close, I wanted to look back and report on how the event went. Overall, it was probably one of the best technical conferences I’ve ever been to. Several things contributed to this, so let me break it down.

The exhibition hall

SuperComputing, SIGGRAPH, VisWeek, and pretty much every other conference I’ve ever been to that contained a vendor display area or exhibition hall suffers from the same problem: maintaining a useful concentration of people in both the exhibits and the technical tracks. At every show it’s the same story: the one interesting talk or keynote begins, and the exhibition hall is deserted save the vendors who must man their booths while waiting for the one latecomer to arrive and troll the floor, then the exhibition floor is flooded with so many people that presenters are talking to empty rooms. It devalues the entire event and frustrates vendors and presenters.

GTC solves this problem in an unusual way. I’ve seen other smaller conferences tackle this problem similarly, but never as successfully as NVidia did. The exhibition hall was only open from noon-2 and again from 6-8, and during both times food and drink were served. The rest of the time it was locked. Having the food there prevents newcomers having to wander out in search of meals, and convinces everyone to at least look around at the vendors. Vendors have the opportunity to venture out and check out some talks, and attendees don’t feel any pressure to choose between food, vendors, or technical papers.

Although it did have a downside. Since the hall was only open for 2 hours at a time, some of the more popular booths were swarmed with people, all hoping to get some hands-on time with the latest gadget. I’m sure vendors loved all the attention, but it made some booths a bit crowded. Also, the few technical talks on the last day fell into the trap of “If you want to know more, feel free to stop by our booth and…oh wait” as they realized the exhibition hall was closed and wouldn’t be opening again.

Content

I’ve been to several VisWeeks and SIGGRAPHs and frequently found the technical talks disappointing. They always seem to fall into one of 4 categories:

  • Type 1: presenting some proprietary technology that, while really kewl, you’ll never see.  This frequently comes from the ILMs and Pixars of the world.
  • Type 2: presenting some piece of code or algorithm that while fascinating, it’s not publicly available in any form that a normal person could use.  It’s available as an obscure library or MATLAB scripts, completely lacking any documentation.
  • Type 3: presentations so confoundingly technical that you need a PhD even to stand in the room.
  • Type 4: presentations of readily available and useful technology.

While some of the first three is to be expected, even desirable, all too often the number of “Type 4″ talks are so tiny as to be counted on one hand.

GTC pleasantly flips the usual, and is almost entirely these real-world uses and products. The  few algorithmic and pie-in-the-sky talks were easy to find, and in just the right balance to show you what’s possible today and what’s on he horizon.

The big difference: the focus

Perhaps the biggest difference between GTC and a conference like VisWeek is that GTC is not a profit machine, and NVidia shows no signs of making it such. You’re not hounded with coupons, discounts, and sales pitches. The registration packet had a coffee mug, the proceedings, and a few pages on after-events like the charity dinner Thursday night. A sharp contrast to the 20 pound bags you get from other conferences that fill you up with free magazines and postcards.

The conference is not run to push sales or increase enrollment, but rather to push CUDA anf GPGPU technology. The fact that NVidia is sub an engineering-heavy company means they know exactly what’s important in a conference and stop there. They know the big 3 of every geek: free food, cool gadgets, and minimal effort. The RFID badges made sure every technical talk had plenty of room (no standing in the back, juggling your laptop and proceedings), and nVidia brought some serious projection and display technology to the event, making sure every single room had a massive screen that was crystal clear from anywhere in the room.

NVidia knows that the value of the conference is the information , not the sales. Pushing the edge of GPU technology isn’t just good for the industry, it’s good for nVidia too, so they don’t need to gouge you for wifi (free), food (free), drinks (free) or registration (not free, but have you compared it to anything else? Cheap).

Conclusion

This was my first GTC , but only the most recent in a long series of technical conferences I’ve attended. I have to admit it wad the best organized, most entertaining, and most educational conference I’ve attended in almost a decade. With over 250 hours of technical talks, and roughly 2000 attendees (my own informal estimate), it had the kind of small informal atmosphere that makes these conferences great.

GTC2011 can’t get here soon enough.

Note: Some of these photos come from NVidia’s Flickr Photostream.

Graphics, Hardware, Science ,

GPU-Assisted Malware

Just what we need … viruses that evade detection by using the GPU.

Malware writers constantly seek new methods to obfuscate their code so as to evade detection by virus scanners. Two code-armoring techniques that pose significant challenges to existing malicious-code detection and analysis
systems are unpacking and run-time polymorphism. In this paper, we demonstrate how malware can increase its robustness
against detection by taking advantage of the ubiquitous Graphics Processing Unit. We have designed and implemented unpacking and run-time polymorphism for a GPU, and tested them using existing graphics hardware. We also discuss how upcoming GPU features can be utilized to build even more robust, evasive, and functional malware.

via : GPU-Assisted Malware (Link to PDF)

Hardware ,

Dual Camera Rig Live Stereoscopic 3D Preview and Alignment Method

Anton over at the 3D Vision Blog takes two Sony HDR-SR11E high definition camcorders and builds himself an inexpensive 3-D camera rig. It is a nice setup. Head on over there to learn more about how to do it yourself, and to learn some of the shortcomings of it.

If you are making a custom dual camera rig for shooting video in stereoscopic 3D format, then you’ve probably ran yourself with some issues in actually getting a live 3D preview in order to setup the cameras correctly for the subject you are going to shoot and even if you just want to properly align the two cameras. Of course the two cameras/camcorders must be the same make and model, but even that does not guarantee you that they will be perfectly aligned with each other, so you might have to do a little bit of additional tweaking in order to get the proper vertical alignment of the two lenses. Here I’m using two Sony HDR-SR11E HD camcorders (on the photo above) and as you can see I’ve connected the video out cable to each other, something that is not so convenient on this model. The closest interocular distance I could get with the SR11Es was about 8 centimeters in a parallel configuration, but with the video out cables connected the minimum distance goes to about 10-11 centimeters which is not very useful for filming close objects in 3D. However it has worked very nicely helping me properly align vertically the two cameras, because they’ve had a bit of an offset as you’ll see on the photos below…

via Dual Camera Rig Live Stereoscopic 3D Preview and Alignment Method – 3D Vision Blog

Hardware

Jupiter’s Great Red High-Resolution Spot

Björn Jónsson took 12 high-resolution images of Jupiter’s great spot and compiled them into the great mosiac you see above.

This 12-footprint color mosaic of Jupiter’s Great Red Spot was composed of 24 images captured through orange and violet filters by Voyager 1 on March 4, 1979 at a distance of about 1.85 million km. The resolution is roughly 24 km/pixel in the “click to enlarge” version; for a full-resolution version at 18 km/pixel (2.5 MB), click here. Credit: NASA / JPL / color mosaic by Björn Jónsson

via Possibly the best view of the Great Red Spot ever – The Planetary Society Blog | The Planetary Society.

Science , ,

Panasonic HDC-SDT750, the First Consumer 3D Camera

We have told you about the Panasonic AG-3DA1 3-D professional camera, and have even told you about the Panasonic HDC-SDT750 3-D consumer camera. Now Akihabara News has posted a review of the Panasonic HDC-SDT750.

Based on pretty much the same design introduced by Panasonic for the first time in July 2007 with its SD5, the TM750/SDT750 does not strike us at first really radically different from other models. To make things even worse, the TM750/SDT750 comes with the same 3CCD Technology introduce by Panasonic on its very first full HD consumer camcorder the SD1 launched in April 2007…

But as usual, looks are deceiving; our new TM750/SDT750 features Panasonic’s latest state-of-the-art improvements including a new 3MOS sensor offering a total of 9.15Mpix with 7.59 effective pixels just dedicated to movies, offering not only better video quality but also drastically improving Noise Reduction as well as offering better, more vivid colors. Add to this list a new HYBRID O.I.S image stabilizer working on 4 different axes, SDXC support, 96MB of internal memory, an impressive 12x optical Zoom expandable to 18x in iA Mode, 1080/60p, a manual focus ring and you have here Panasonic’s ultimate pocket camcorder.

If this is not enough for you yet, we should remind you that we are talking here about an ultra-compact Camcorder that just weighs 440g for 66x69x138mm without its 3D lens converter and 635g for 78x69x235mm with its 3D Lens Converter.

via [Review] Panasonic HDC-TM750/SDT750, the First Consumer 3D Camera. « Akihabara News

Hardware ,

Honda CR-Z Times Square 3-D Takeover

Honda held the first-ever takeover of Times Square in 3-D technology to support the launch of the all-new CR-Z sport hybrid coupe on Thursday, September 23. Spanning five blocks of Times Square, the event marks the first time 3-D footage will be broadcast to the public space.

via YouTube – Honda CR-Z Times Square 3-D Takeover | DCH Honda of Temecula.

Hardware

Converging Flight Paths

A surprisingly good graphic from the New York Times today shows what’s happened to the airline industry since deregulation in 78, with each bar representing the total percentage of passengers in the US.  Currently, Southwest seems to have the slight lead with 17%, while classics like American Airlines have dropped down to 12.2%.

The deregulation of the airline industry in 1978 led to a wave of mergers that continues to this day. But even as the legacy carriers have been consolidating and growing, they have been losing market share to low-cost carriers. Two of them, SouthWest and AirTran, have just agreed to merge and carried the most passengers in 2009 combined.

It’s a great representation, the “growing bars” do a great job of not only representing the information, but symbolizing the growth of the companies in a quick overview as well.

via Converging Flight Paths – Graphic – NYTimes.com.

Graphics , ,

Daily Viz from Visual Loop – 28/09/2010

Online education has been a growing educational option around the world, and in America it’s already a huge business. Two infographics that show that are the first picks of today’s selection: Homeschooling by the numbers, from Degree Search, and the tuition costs for Online Colleges, made by Degree Central. We then move on to the obscure side of gift cards, by Mint, the top ten ways the Auto Industry rips you off, according to Auto Insurance, and, finally, a look at the Online Retailers’ $44 Billion Customer Experience problem, from Fast Company.

Read more…

Graphics, Science , , , , , ,

 
Stories from September 27th, 2010

Infographics Summary for 2010-09-28

youtube-thumb

20 Interesting Facts About Youtube

Graphics, Science , ,

First Details on SGI’s Project Mojo supers

DARPA’s “Exascale Challenge” asks vendors to design a supercomputer than can cram a petaflop of power into a single rack.  Several vendors have begun on the effort, and SGI’s entry was codenamed ‘Project Mojo’.  Folks at The Register got some details on the system, and… Well, first read the brief description:

As it turns out, the stick of the Project Mojo system is a computing element that is nearly as long as the rack is deep – three feet – with the width and a little more than the height of a double-wide PCI-Express peripheral card. Mannel wouldn’t say what processor is implemented on the stick, but it is possible that SGI has variants with both Intel Xeon and AMD Opteron processors. Considering that Project Mojo is an experimental system with limited sales on the front end, it is reasonable to conjecture that SGI will start with Xeons and expand into Opterons if there is customer demand.

So an element is 3-feet deep, and roughly the size of a double-wide PCIe card, and (if you read a bit further) you can cram 80 of them into a full-size rack.

This sounds mighty familiar.  In fact, over at my “other job”, I have one of the BOXX “RenderFarm on Wheels” systems containing their renderBOXX module.  The renderBOXX module is 32-inches deep (4 inches shy of 3 feet), 7 inches tall and  under 4 inches wide, but each module actually contains 2 nodes.  In a full-size rack, you can pack in 80 nodes.

So basically, it sounds like SGI is taking BOXX’s RenderBOXX system and splitting the nodes in half, and found a way to pack in a passively-cooled GPU.  Nothing too earth shattering there.

I hope they have some bigger plans coming down the pipe.

via SGI readies first Project Mojo supers • The Register.

Hardware ,

VizWorld.com is a production of VizWorld, LLC © 2009