Publications

A look at the future—maybe

2010-03-02T14:22:15Z

We’ve spent some intense and serious time with the AMD folks recently, in Las Vegas, Texas, Santa Clara, here in our labs, on the phone and by email, and we’ve gotten a pretty good understanding of the company’s plans and people—or so we’d like to think (or maybe they’d like us to think…). Putting aside Machiavellian schemes and ideas for novels, being able to understand a company and its people keeps you from making toady comments, misstating facts, and hopefully making bad forecasts. It also helps you better understand other companies in the market.

AMD is into sweets-SSSS

In June of 2008, ATI introduced their Sweet Spot Strategy (SSS) in graphics, basically changing the rules of engagement on how GPUs would be built and employed in AIBs and elsewhere. It seemed a brilliant concept at the time and we all waited to see how it would play out. Turns out it’s played out pretty well and, except for some supply constraint issues, would have propelled ATI to the forefront in market share.

In February of 2010, ATI introduced their Server Sweet Spot Strategy (SSSS.) Basically changing the rules of engagement in how Servers would be built and employed in systems and elsewhere. It seems like a brilliant plan and there’s some discussion about it in this issue.

In August 2008 AMD introduced their Sweet Spot Strategy for singularity (SSSS) in chipsets. Basically, the company began to show signs of being a true platform provider.

This week, the company introduced their latest, best, and last IGP, the 890GX, and there are, I think, some tales in it. IGPs, as you know, have to be small and inexpensive and ATI’s are. They also have to provide as much performance as the silicon and price budget will allow in order to justify their existence—and for the most part the suppliers have been doing that with “good enough” offerings, and sometimes better (i.e., Ion, and the 790GX).

The 890GX is remarkable for a variety of reasons, and one might ask why AMD would put out such a powerful chip, with 40 unified shader processors in a low cost, limited lifetime product?

First, one more data point. You will recall that ATI invented the unified shader before Microsoft implemented it with DirectX 10. ATI invented the unified shader for Microsoft for the Xbox360, and used it again in the Nintendo Wii. ATI/AMD now has several years of making very small unified shaders.

The next sweet thing

Later this year, AMD will introduce their first Fusion product, Llano—the monolithically integrated multicore CPU and graphics processor.

This will be another shot at a sweet spot—namely the value and midrange of desktops and laptops and will be AMD’s Sweet Spot Strategy for Synergy (SSSS.)

Llano will be built in 32 nm, at Global Foundries—AMD has shared that much—in fact they are being built now in preparation for engineering sample deliveries to interested OD and OEMs. And from what we’ve heard those OD and OEMs are very interested.

Specifications have been more difficult to come by and so we’re left to our speculations. I believe Llano will be based on a shrunken Phenom II with a much larger and coherent cache for communicating with the GPU portion of the device. And I think the 55nm built 890GTX shows us that Llano will have well north of 40 unified shaders in it, so I’m guessing 64 is the right number.

Also, I’m guessing Llano will have Direct Connect 2.0 and the ability to attach a sea of memory to it for those folks who want to do heterogeneous computing.

Let’s C if we can compute

As of now, AMD is relying on OpenCL as the programming environment for its heterogeneous offerings. And, as I’m sure you noticed, AMD recently announced they have ported OpenCL to their CPUs, as well as their GPUs. Of course, it makes common sense for AMD to do that. Intel will do it too, but the most interesting part is that it is a perfect prelude to an integrated heterogeneous product isn’t it? Sweet.

And now the bad news. Whereas AMD is doing just about everything right on the hardware side, they are way behind on the software side. Relying on OpenCL as their programming environment will put them in good stead with the programming and OEM community, but it’s like watching paint dry before you have anything useful. Say what you will about Nvidia’s CUDA being proprietary, and how much everyone hates such things—it is nonetheless the current industry standard and more than that it’s the industry’s darling if the enthusiasm at SC09 was any measure.

What if you threw a party and no one showed up? That could be AMD’s fortune if they don’t get their programming act together. They might show up this fall with the amazing and sweet Llano and no tools to use it with—that would not be very sweet.

Look at me when I talk to you

2010-02-16T19:47:40Z

About ten years ago, I wrote an editorial suggesting our PCs could, and should, look at and listen to us. And when they did they would be able to see if we were happy, angry, sad, impatient, or bored. They would know when we were dictating a memo, story, or email, and when we were giving them commands to open a file or get a program. I even went so far as to suggest eye-tracking as a replacement for the mouse. I never suggested hand waving as an alternative input mechanism because of my experience many years ago with a light-pen. For those of you too young to know what that is, the early CAD and computer drawing systems used a pen-like device that was pressed against the display’s screen and would allow the drawing of lines. Inside the light-pen was a photo sensor and the computer would (very quickly and usually unperceptively) scan the screen looking for it. Although you could draw very accurately with that technique, your arm got tried pretty quickly. Try it—hold your arm out toward your display and see how long you can. So for that reason, moving your hand off the desktop (from the mouse, touch pad, trackball or joystick), and suspending it in air does not seem very “natural” to me. But, you can wink at your screen, and look at various parts of it.

Obviously as we get surplus amounts of MIPS, various non-physical contact user interfaces will be tried. And like most things one size or technique won’t fit all (just as some people use a track ball and others prefer the nimble in the keyboard.)

But the computer will look at and listen to us. My computer listens to me almost every day. In fact this editorial was “written” using speech recognition software (Dragon Naturally Speaking v.10) as were several of the other stories in this issue. You can’t tell by reading it, because any mistakes that the software might have made have been corrected as is the case, whether you’re typing it or saying, there’s always an editor.

Although we have the computing power to do the image processing necessary to allow a PC’s camera to watch us, the resolution of most cameras in use today is just VGA. HD cameras are dropping in price and readily available and will be needed if we are to get truly useful and reliable facial and gestural recognition.

Imagine now that we do have such capability, and that we also have adequate bandwidth for video conferencing. Think about the possibility of having your computer’s image processing software evaluate the stress levels in the person you’re communicating with via videoconferencing and reporting to you if your computer thinks your correspondent is lying? Such communications could revive person-to-person communications and lead us back from relying so much on e-mail and text messaging. It certainly would be a useful capability when negotiating a contract with someone. It would also be helpful for people doing online dating. Imagine the savings you would realize by not having to drive to a coffee shop or bar to meet someone for the first time.

Of course, the gaming aspects are well understood, and in fact it was the Wii that ignited the imagination of so many with regard to the potential of gestural and visual vacations with a computer. Human interface studies will be needed to determine the best movements to use to communicate with, and control, the computer. You can envision a common set of operations similar to what we have developed over the years for the operation of an automobile. However, there’s also the possibility of making your computer tamperproof by having a unique set of gestural communications that would prevent anyone else from operating your computer if they didn’t know those moves.

The downside of speech recognition is the noise level colleagues have to tolerate while you are speaking to your computer. It’s not dissimilar from hearing one side of the conversation when someone is using a telephone or mobile phone. The difference is the continuity of speech is more coherent and therefore more distracting.

These will be evolutionary changes, not revolutionary, as PC manufacturers slowly adopt and incorporate high-resolution cameras in the displays. Eventually it will evolve to the point where there are two cameras in the display and will be able to use stereo vision. Two cameras also give you better distance capability so that the user’s position can better track and be used for determining the mood and intonation of communications. Single lens 3D (depth) sensors like Canesta promise a great potential for a low-cost solution. We expect other 3D sensors to appear in the market very soon.

We’re experimenting here at JPR with ambient sensing sound systems that recognize limited instructions such as “wake-up computer,” “lights out,” “start coffee pot.” The bits and pieces to set up these types of systems already exist from a hardware point of view, and mostly exist from a software aspect. They lack universal standards, and so therefore each system, today, is custom. Therefore, I encourage you to build such systems, experiment with them and learn from them. Share your knowledge and help us move forward in more natural communications with our omnipresent companion.

The middle ground—is there one or many?

2010-02-03T18:30:25Z

When Steve Jobs presented the iPad, he set up the premise that we have smart phones, and we have laptops, and there is a gap between them that should be filled. As he made his presentation, I thought, I just saw this presentation a few weeks ago in Las Vegas. Jen-Hsun Huang did it at CES. Huang said the exact same things, and showed tablets that would fit in the middle.

And then I thought about Intel, who also wanted to fill the perceived middle. Intel called it a mobile Internet device—MID. Intel wasn’t successful in creating that category, but a lot of interesting products were created that were neither mobile phones nor PCs. We called them gadgets, and now Intel has also adopted that term. And, some of those gadgets are tablets.

Tablets, of course, aren’t new. We first started talking about them in 1970, and the first commercial product was shown in 1989, when GRID computer introduced the pen-based GRIDpad, 12 years before the Windows XP tablet, and four years before the Newton. A whole community was built up around the notion of a touch sensitive, large-screen mobile device, but no general-purpose usage model could be found for them. They were a solution in search of a problem. They did get applied in some special applications like insurance adjusters and a few doctors tried to use them.

HP introduced a novel version with a two-axis rotating screen that would lie flat for touch operation, or swivel up for convention clamshell laptop operation. Kathleen Maher has one and still uses it. But those early tablets were really just PCs with a big touch screen. And what Jen-Hsun Huang, Steve Jobs, and Paul Otellini are searching for is something that’s neither man nor beast—a narmachine. Have they found it in the tablet?

Probably not. The tablet as conceptualized by Apple is a unique device in that it doesn’t use a PC’s OS. It uses a processor found in most mobile phones with the same graphics engine as the iPhone, but some similarities are to be expected. The iPad is YAG—yet another gadget, like the Kindle, the TomTom, the PSP, the digital picture frame, etc.

I like the iPad, I hope to get one. And I won’t toss my notebook, PSP, or Kindle out. Those devices serve a purpose very effectively and I doubt that as clever and pretty as the iPad is it will be able to match my existing device—but that could just be me being stuck in habit—I’ll try to fight that and give the iPad every opportunity to prove itself.

We’ll know more this time next year as we visit coffee shops, trade shows and attend meetings. Will we see lots of people using a tablet (from the dozen or so providers who will emerge this year)? And what will we see the year after that? Are we on the cusp of a paradigm shift in usage?

Stereo—can we get real?

2010-01-19T12:45:30Z

The movie industry commonly called “Hollywood’ is famous for stereotypes, mislabeling, and misrepresenting. In some sense you could say that’s their business model—to create an illusion, to misdirect. Just the term Hollywood is misdirection as the Canadians, French, Indians, and Nigerians will attest. Those countries produce an amazing amount of movies and some amazing content and yet they are grouped under the banner of Hollywood.

The same thing is true with regard to the trick of creating the impression of depth on a flat two-dimensional surface known as a movie screen. The movie industry and specifically Hollywood in this case wants to call that stereopsis effect, “3D.” They (the Hollywood moguls and their marketing departments) have invested in that nomenclature and we won’t be able to change them. The term “stereo” can’t be used, it’s too similar to dual channel sound. And the whole word stereovision can’t be used because it’s too long—3D is quick, snappy, different, and terribly wrong. But it will be used, and it may leak into usage for stereopsis TV viewing as well.

I must protest. Stereovision is not 3D. 3D is a Cartesian three-dimensional geometric representation of data—a model. It is genuine, specific, and original. Stereovision is a manipulation of 2D images. Stereovision has no real dimensional value (although some of the new sensors can provide a bit).

I like the name Neil Schneider came up with (www.mtbs3d.com) of stereo3D or “S3D,” to distinguish and differentiate stereovision. And so that is the nomenclature we at JPR are adopting when referring to stereovision. We may (probably will) use 3DTV when referring to that platform because it carries with it a definition of information delivery.

So, computer games being viewed with Nvidia’s “3D Vision,” or with IZ3D’s monitor will be referred to as S3D games, and associated equipment such as glasses and screens will be S3D devices.

Is this being pedantic or nitpicking? Maybe a bit, the problem is when someone says 3D to me, if the context is not obvious I always have to ask, do you mean stereo or are you talking about a 3D model.

Stereo glasses

Have you ever driven a car for more than three or four hours while wearing sunglasses?

Have you ever heard some pundit or webzine writer say, “People won’t wear glasses?”

Do you see any incongruity in those two positions?

I fail to understand the notion that people don’t want to wear stereo glasses. And when I ask people who express such a position, why not? They tell me—100% of the time, “they look goofy.” Then I ask, are you watching the movie, game, or TV show, or are you watching the people watching the movie, game or TV show? Look goofy? Have you looked at how you’re dressed?

After I win that argument, they say, some people can’t see stereo and so they won’t wear the glasses. “Some people.” That’s like, some folks say … sure, there are some people who can differentiate stereovision, just like there are some people who are color blind. But they represent less than five percent of the population. Hardly a market killer.

When I win that argument, they make it personal and say, “I get a headache after fifteen or twenty minutes.” I’m thinking, you’re giving me one right now. So then I ask, did you see Polar Express or Avatar, or any stereo movie in the last year? And they always say yes. And then I ask, did you get a headache? And they say, grudgingly, well no, but…

So it’s not a logical, or scientific argument, but rather one of hearsay and emotion, which of course you can never win, so I say, too bad for you Mr. monovision, there’ll be no depth in your entertainment—by the way—do you ever wear earphones … ?

Launch

And while we discussing definitions, I’d also like to get a definition for the term, “launch.” As I see it there are several types:

PowerPoint launch—no product, no schedule, limited specs, no price, lots of TAM data.
Soft launch—first general specs released, suggested delivery date (SR) SR=subject to revision.
Product launch—the product is fully spec’ed, has a model number, has a suggested price (SR), and a scheduled shipment date (SR).
Development product launch—beta version sent to software developers.
Product sampled launch—first limited production run parts sent to partners and OEMs for testing.
Hard launch—product announced and actually available in the channel with published specifications and prices (SR).
Product re-launched—bugs fixed, parts or software updated, shipping soon. Sometimes called re-branding.

If it was possible to remember all this think how much more efficient our communications would be. “How’s it going Sam?” “Pretty good—we’re at four.” “Hey did you hear Jamit Tech just hit six.”

Predictions

2010-01-05T14:36:14Z

They’re pretty predicable—wrong. This is the time of year when everyone and their dog make predictions about what’s going to happen next year. I predict that this year, that will happen again (see, my accuracy count is already ahead). I was thinking about all the predictions we’ve heard and how many of them came true —very few.

Worst prediction? One of my favorite DA (dumb ass) predictions was how the world would come to a screeching halt on 00:001 01.01.2000 because computers could only go to 1999. I haven’t seen a decimal-based computer in quite a while, and 1999 is not a base binary number. So when it didn’t happen, (some folks sold their home and stood naked on their lawns waiting for the rapture so they were pretty pissed off about it). The experts said the problems were solved because they sounded the alarm and new computers were bought and software patched and things got taken care of. Yeah, like industry or mankind ever heeds a warning—did I tell you a hurricane will hit New Orleans?

Biggest missed prediction? No one (except maybe Nigel Dessau at AMD) predicted Larrabee would be canceled.

Second biggest missed prediction? Intel will settle with AMD make payment to AMD, who saw that coming?

Some classic wrong predictions (not by me, thankfully):

Microsoft will buy Yahoo. (need the definition of “buy” worked out).
Stereo (“3D”) will revolutionize PC games. (When will the revolution start?)
Sony will announce the PSP2 with Nvidia graphics.
Intel will acquire Nvidia.
Apple will introduce a tablet (timing, it’s everything).

Some predictions that weren’t made:

Oracle will buy Sun.
Disney will buy Marvel comics.
HP will buy 3Com.

So who the hell knows anything, really, except maybe 30 days ahead and then only with insider information. Best any of us can do is guess. We say we use our experience, our wisdom, our cynicism, or what—our mystical abilities? Horsefeathers—you’re guessing and hoping it comes true so folks will come back and ask for another guess.

The supersmart stockmarket traders, the same folks who are making the buy, sell, put decisions on your 401k, are doing the same thing—guessing. Reading tea leaves, dried up bones, and cloud shapes to predict what AMD’s, Broadcom’s Intel’s, Nvidia’s, or Qualcomm’s stock will do based on rumors in the Inquirer, Digitimes, and “the talk” on Wall Street. These are the same folks who think Micron makes potato chips, Nvidia is pronounced naVIDha, and VIA is a passage way.

And every time you snicker at some DA comment made by some DA web blogger or poster, just remember, your DA stock broker is reading that and playing with your money.

In the land of the blind the one-eyed man is king. Why’s that? Well, dummy, because he can see the road ahead, that’s why. So, are the web posters the one-eyed man and you’re just blind? Blind with ignorance or fear? Who dresses you in the morning and packs your lunch? Are you going to commit your future to someone else’s predictions? Why are his or hers better than your own? For crying out loud, you’re actually working in the industry—what are they doing; sitting in a coffee shop or their mom’s cellar pecking at a keyboard?

Here’s a prediction—fools accept other’s predictions—read about it on the ten o’clock new.

Cautions on predicting. The Mayans never predicted anything. 21-12-2012 isn’t the next 01.01.2000—keep your clothes on, don’t sell the house.

My prediction? Lots of predictions will be made in 2010—most will be wrong, even one or two of mine.

Welcome to the Cloud - It may be scarier than you thought - Cloud Computing

2009-12-10T12:46:48Z

The following is purely fictional. No small children, animals, or data were hurt in the making of this story. Reader’s discretion advised. Content contains suggestive and potentially rude ideas, loud music, and sex scenes that may not be appropriate for readers with an IQ of less than 110.

Winter in Seattle, and the sun had set on a cold and rainy Friday night. Jaymes couldn’t stand the thought of one more cup of coffee to try and fight the fatigue of a long day and even longer week. He just had three more paragraphs to write, make the finals sums in the spreadsheet and then paste all that into the PowerPoint slide for Monday’s presentation in Chicago. A red-eye flight for him but if this project was approved he could get some help and maybe even take a day or two off—once things got rolling.

The proposal was a joint effort between Jaymes and Margret in the Dallas office, and they had been trying to find a better way to exchange information that would be fast and efficient. Email was OK, but lacked the immediacy needed. IM worked fine but was file limited and sometime unpredictable, so they were using a collaborative set of online programs that would instantly make the latest version available to both of them. Editing in real time with a colleague seemed like a dream come true, it was as if they were working in the same office on one machine with two keyboards. They used IM for chat and even though they were two time zones apart the distance wasn’t noticeable.

Margret told Jaymes she was starting to burn out, it was 11:00pm in Dallas and she had been at it since seven that morning. Give me what you’ve got Jaymes told her, and I’ll pull it all together, I’m almost finished here. He turned to check the now cold cup of coffee just as the lights went out and the room went dark and deadly silent. It took a couple of seconds to register in his fatigued brain—it was quiet—the PCs were dead, the UPS didn’t kick in. A screaming “Oh no!” came out of him as he jumped up out the chair. Disoriented in the dark, he tumbled over it, hitting his head on the edge of the desk and slumping to the floor.

He didn’t know how long he had been unconscious, his face was sticky and he was afraid to touch the pounding pain in his head for fear of what he’d find. He climbed up on his hands and knees and stayed like that hoping he wouldn’t feel blood trailing down his cheek. Felling somewhat stable, he crawled to a wall and used it to stand up. He had never been in this room with lights off and since all the lights in whole building must be off as well as outside it was a dark and featureless as being in a cave.

He felt his way around the wall hoping to find a reference point, and get to his desk where he thought he remembered leaving his mobile phone, or maybe if the cell towers were out he could find a desk phone, but that wouldn’t work if the PBX was out, maybe it had a working UPS.

He reached what he was sure was the edge of his desk and began feeling his way along its top when the lights were restored. Blinding at first since his pupils had been so dilated from the pitch blackness of the windowless and closed office, he almost fell over. But it was still deadly quiet. He had to get the computer booted and get in touch with Margaret. He called her on her mobile. She was in the car almost home. “Where did you go—emergency biobreak? You gotta lay off that coffee Jaymes.”

He explained the situation and then, fearful of what her answer might be, asked what she had done last. “Posted the file, shut down and left, what’d you think I’d do? I’m almost home and beat to hell. At this point I don’t care about the project, the weather, and barely about you.”

OK, he said, told her to take a couple of days off and he’d wrap it up. The computer was up now, he logged in and waited for the browser to load; his head hurt like hell and he decided he better go to the men’s room take a look at it and see if he could find some aspirin. He was shocked at the amount of dried blood in his hair and on the side of his face and could see the welt the fall had made. But after a little gentle washing he was relieved to see it didn’t start bleeding again—meaning he wasn‘t going to need stitches. There wasn’t any aspirin, rules about dispensing medicine and insurance rates; he’d get some at 7-11 on the way home.

Back in the office, the browser was blank. He clicked on a few icons and looked for the docs—but they weren’t there. The folders were empty. He began four different searches, looked in the trash can, nothing was found. “That’s freekin impossible,” he shouted at the room, “you can’t just lose six months of work because of a power failure.” He opened the app, started a new file, and saved it. Then he rebooted the computer. Opened the browser, opened the folder and the new file was there, just as it should be. He froze in his chair. It was gone, the project was completely gone. And now he realized he had no backup, it was all being saved online. Surely the web provider had a backup. But who should he call, and on a late Friday night? He didn’t dare call Margret—she’d charter a Lear jet and come and kill him with her bare hands.

“The cloud,” he yelled, “The freeking god-be-all cloud!” He was too tired, too shocked to even cry. He kicked the chair backwards, grabbed his backpack and walked out, leaving the office door open, lights and computer on, it no longer mattered—he’d start looking for a job on Monday.

Your locus of control and what it means to be super

2009-11-25T16:00:17Z

In psychology, there is a term to describe how one sees himself or herself with regard to the rest of the world. If they feel they are in control of their lives and themselves they are said to have an internal locus of control. If they feel they are controlled by the environment, by bosses, family, government, etc., then they are said to have an external locus of control. And yes, naturally it varies from time to time, but in general one’s locus of control is steady.

People whose lives are ruled by their phone, Blackberry, or some email or social network client are demonstrating with their behavior that their locus of control is elsewhere. An external locus of control is typical of anyone who suffers from a compulsive behavior whether it’s game playing, obsessive email or Facebook checking, smoking, compulsive drinking, eating, and gambling.

Governments can have an external locus of control too. Countries that are obsessive about the actions of another country have in effect turned over their own destiny to the leadership of the country that drives them to distraction. Such countries misappropriate resources for inappropriate ambitions, such as a poor country with limited electricity and hungry citizens building a nuclear bomb and missiles instead of power generators and farm equipment.

Those people and countries with an external locus of control want to be super. They want to be superstars recognized for having processed or sent more emails in a day than anyone else, or super nations because they have WMD. They don’t have what my grandmother used to call the gowithits—the whole package. It’s like the guy who buys a fancy car but can’t afford insurance, fuel, or a covered garage to keep it in.

So what does it mean to be super? What, for example, is a supercomputer?

Well super is a fleeting thing in many cases and isn’t easily remembered. Last year’s Olympic gold medal winner is just today’s job seeker. Last year’s Oscar winner is scrambling for parts against Brad Pitt, George Clooney and an animated fur ball. And, last year’s supercomputer is now number three or five —and they were all so super.

Japan has effectively dropped out of the super computer race. Its amazing Earth Simulator has fallen to 31st place and it looks like the country may cancel its pending new supercomputer project. One could say Japan is being adult and not building something with limited immediate payoff with taxpayer’s money in favor of solving more immediate problems. It’s a hard pill to swallow but one that has to be done.

With supercomputers following Moore’s law and surpassing it with GPUs, the title of “Super” is even more fleeting than before. Prior to GPUs, supercomputers went up in cost faster then they went up in performance, despite Moore’s law. The GPU even influenced Blue Waters Petascale Computing System that is being built by the NCSA at the University of Illinois at Urbana-Champaign (slated to be completed by 2011), which is going to cost over $200 million (funded by the NSF.) The canceled Japanese Riken supercomputer was budgeted at $300 million.

Clearly, in these troubled economic times with every country in the world except China running huge deficits, supercomputers look like a questionable luxury. And yet, we can’t stop progress, or stop research. Too many people are counting on it—if not to make our lives better then to save us from our bad eating habits and global warming.

GPUs may save the supercomputer industry, if the supercomputer industry can recognize in time that the government horn of plenty is running dry and that the government leaders’ locus of control is going to have to be focused internally now and until we get out of this economic crisis. These are extremely difficult decisions. Adventures in foreign lands, mass-colliders, space exploration, and supercomputers will only aggravate the debt, not reduce it—and sadly these are the totems we love so much. They’re just so, well, so super.

The world at our fingertips

2009-11-09T23:34:34Z

We are so enamored with our multifunction mobile phone, and our multifunction PC—devices that promise us the world at our fingertips—that we may have forgotten, those of us that ever even knew, about the machines that got us here. The Jacquard loom comes to mind as an early single-function machine, and I’ve commented on that before. But the idea of having all the world’s knowledge available to you, your own library of Alexander, was first attempted by Ephraim Chambers in London in 1728, when he published his two-volume Cyclopaedia, which was the forerunner to the modern encyclopedia developed by Diderot in France (which actually started as a translation of Chamber’s work). H.G. Wells, the author of such famous works as The Time Machine, and War of the Worlds, as well as The Outline of History, envisioned a huge worldwide networked library he called the “World Brain.” The networking was the weak link in the concept and it never got funded, but the idea was fascinating and persisted.

Single-function analog electronic computers were developed in the 1900’s, one of the first being Arthur Pollen’s electrically driven mechanical analog computer for fire-control systems, based on James Thompson’s differential analyzer of 1876. But in the mid to late thirties, before Konrad Zuse had developed his digital computer, the vaunted Vannevar Bush, who designed several single-function analog computers, first publicly proposed the idea of a device that could hold all your information. It was to be a desk-like device that could hold the contents of a university library.

The war and the need for analog fire control systems distracted Bush from the idea, but afterwards, in 1945, he proposed it again in an article in the Atlantic Monthly unceremoniously titled, As We May Think. Here, Bush exposed the ideas of compression and indexing in a device he called the Memex—and it was the precursor single-function device that ultimately led to the ubiquitous PDA that we carry today. A memex, Bush said, “is a device in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility.”

So the big ideas for a networked world library, and a personal information machine were developed before most of us reading this were born. In 1960 J.C.R. Licklinder proposed a geographically distributed network which led to the famous ARPANET. And, about the same time, Ted Nelson started project Xanadu and coined the term “hypertext” and “hypermedia” in 1963, which ultimately led to Tim Berners-Lee building the first web site at CERN in August 1991.

Now that we had all the ideas, the network was under way, and we just needed one more thing, which Gordon Moore observed in 1965. The number of transistors would double approximately every two years. It was easy then to project when we could build a Vannevar Bush’s Memex that would fit in our pocket.

The first attempts were single-function devices called PIMs—personal information management devices. They were like a pocket rolodex and had little LCD screens and an alphanumeric keyboard. The first ones appeared in the late 1980s, and were also known as digital diaries. I remember how proud I was of my Casio 4000. And these were truly single-function standalone devices—no I/O to a computer or phone line. The first PDA is considered to be the Casio PF-15115-36 released in May 1983, although Psion’s model introduced in 1985 is mentioned, and Apple often gets credited with having introduced the first PDA call the Newton MessagePad in 1993, with a touch sensitive screen to detect handwriting (the precursor to the iPhone, 14 years later.)

There’s a similar trail of firsts of single-function music players beginning with the Saehan’s MPMan, sold in Asia starting in the late spring of 1998, and followed by the more popular Diamond Multimedia Rio PMP300 that same year. Of course, most people still believe the Apple iPod introduced in 2001 was first.

So here we are today, 64 years since the concept of Memex, and 26 years since the first Casio unit and, lo and behold, there are these novel new things called single-function devices for tweeting and Wikipedia’ing. And aren’t they novel ideas? Imagine; the idea of having the entire world’s knowledge available to you, your own library of Alexander; a single function, in your pocket. Imagine.

The democratization of the PC—the ubiquitous USB port

2009-10-28T15:46:30Z

I remember back to 1996 when the first USB standard was proposed. I had done a little consulting with Intel prior to that and tried to convince them to use Philips’ I2C which had been developed in the early 80s as a CE inter-box communications system It had a 400 kb capability, and in 1998 got updated to 3.4 Mbit. In 1986 Apple and National Semiconductor introduced Firewire and it a 400 Mbit half duplex capability. Firewire ultimately became IEEE 1394 in 1995. Intel was actually interested in I2C but Philips at the time thought they had the golden goose and thought they had the golden goose and they were asking too much for licensing. —Intel had other ideas about expanding the connectivity of the PC and making the whole market grow. Apple was never interested in being associated with the (ugh) PC, so a deal with Firewire wasn’t going to happen. And since they didn’t know it couldn’t be done, Intel developed the USB.

They showed it to anyone who would listen and with not too much effort formed a consortium composed of Intel, Compaq, Microsoft, Digital Equipment Corporation, IBM, and Northern Telecom, with Philips conspicuously absent. And although Ajay Bhatt, the CO-inventor of USB may be the rock star and gets a lot of the lime light (http://www.tonightshowwithconanobrien.com/video/clips/intel-rockstar-100909/1165473/) my recollection is it was big Jim Pappas who was the driving force behind getting the spec adopted. (I still remember how excited he was in the late 1990s when he proudly showed me 100 devices all hooked up together.)

Bringing out an I/O standard—or a proposed standard—is like jumping off the high board, you hope you got everything right and that there will be enough water below you. Everyone knew RS232 was maxed out, didn’t have any duplex signaling capability (although some tricky stuff was made to work), and other alternatives like game ports weren’t going to scale, something was clearly needed, and if the politics could be managed, maybe, just maybe a new interface might get into the PC.

It did and when it did the face of PC land changed more than any other development. It didn’t happen overnight, these kind of things never do, but once it got rolling it was an unstoppable force. The list of peripherals that can be attached to a PC via USB would fill this entire issue, and more are coming. TV tuners to displays, disk drives to cameras, keyboards, mice, stereo glasses, audio amps and speakers, and the list goes on and on.

But it was a pink slip for geeks. No longer did my dear aunt have to call me if she wanted to plug in a new printer. The users were liberated, the things available exploded and man- and woman-kind began to feel like they actually owned their machines—and that was the moment—the liberation of the user, the democratization when any user could at any time plug in anything and expect it to work. USB became like the power plug in the wall—if the connector fit, the thing attached to the cord worked—period. And the culmination of this can be expressed in the form of the power plug to USB adaptor—how much more universal can you get than that?

But USB does have a few limitations. The current version is limited to 5 m cable length (~16 ft.), and can only carry 500mz max at 5vdc. A special version called Powered USB can carry 6A at 5V, 12V, or 24V, and so that suggests a monitor with just one cable—a USB cable. The next generation, USB 3.0, will take us up to 4.8 Gb, and 900ma standard. Announced in 2007,we expect that to start rolling next year—or maybe not since Intel doesn’t seem to be supporting it strongly, maybe we’ll just skip it and move on to USB4 LightPeak.

Next?

Remember that famous line in The Graduate—“Plastics, son, get into plastics.” Plastics are next, in the form of optical cables. At IDC, Intel publicly showed LightPeak, the new super high bandwidth interface Intel has been working on that uses optics instead of copper wires. Optics offer so many advantages—no EMF signature so they are secure, no IR loss so they are kool, and almost no length limitation. The initial data rates are 10 Gigabits and it has a potential scalability to 100 Gigabits and beyond.

And there’s Intel again up on the high board. Could LightPeak be USB 4, or maybe 5? The first thing needed is a connector. Candidates in discussion are a BNC-like thingie, or maybe a micro USB. Should it carry some copper for power? Still a few questions to workout. And, because of the bandwidth, this could be the universal connector—no more VGA, no more PS2, HDMI, ESATA, just one connector, how cool would that be?

And, think of what will happen to the PC when it becomes available. If USB 1.0 was liberating, LightPeak can take us to the next two generations.

Of hot rods and chipsets—Nvidia can’t win

2009-10-15T15:27:56Z

What an incredible week this has been for Nvidia. It started out with them introducing the most revolutionary new architecture in a decade and ended with stories about them getting kicked out of the chipset business, and giving up the gaming market.

Fermi is an amazing piece of technology—period. The Nvidia haters (they are legion I’m told) have been quick to point out that it is overloaded with silicon not needed for gaming, and is big, and won’t be available in time for this year’s holiday madness. And actually much of that is true—but wait—there’s more.

Back in the days I built hot rods we would buy the biggest engine car we could afford—and then strip everything off it we possibly could to focus its horsepower on the issue at hand—being first off the line and to the finish line. We didn’t start with a tiny engine and try to build it up. So off came the air conditioning, out went the power steering, the bumpers were pulled, the back seat thrown out, and we’d even pull the gas tank and replace it with a can—this car only had to go 1,350 feet.

Fermi is the same kind of thing. If there’s stuff in it that’s not needed for gaming don’t you think Nvidia is smart enough to know how to make a stripped down derivative? But before you can start stripping down, you gotta have something big and powerful to work with.

Getting out of the game business—I don’t think so. Getting out of the lime light since they won’t have a new product for the holiday—you betcha. Don’t read too much into those tea leaves.

Chipsets

And then there’s the sad news about them being driven out of the chipset business. This story has more spin on it than quantum procession processor. The first words began to appear in January 2008 about the rift between Intel and Nvidia over chipsets. When Intel reveled their high-speed (25.6 GB/s) HT-like QPI CPU serial interface they told Nvidia to not even think about making an X58 clone part because Nvidia didn’t have a license to the new interface. The two companies flustered and blustered in the press and on the web until Intel filed for a declaration in May that prohibited Nvidia from saying they had a license. That was an unusual move, but seemed to be effective and the noise level from Nvidia quieted down except for a few “yes we can too” murmurs.

After Computex, in August 2008, “Nvidia hater” Charlie Demerjian, working for the Inquirer at the time, declared that Nvidia was exiting the chipset business, had shut down development in their Shanghai lab (that they acquired from ULi in 2005) and reassigned the troops there.

In December Nvidia renamed the FSB-based GeForce 9400M as Ion and went after Atom sales.

Last week statements from various Nvidia people, mostly PR, set the web hive buzzing, every Webster was trying to be the first to post the most shocking story about Nvidia’s abandonment of its chipset business. Boy I’d hate to be an inside PR person at AIN (AMD-Intel-Nvidia), running around all the time stomping out fires, issuing corrective statements, trying to get the spin right—awful ulcer creating job, but I digress.

Somebody at Nvidia was asleep at the switch that’s for sure. Letting word get out about halting any business or product is the same as the Osborn effect (announcing a new product before you have it, effectively killing your current sales.) What’s the first thing consumers do when they hear such a story—avoid buying the product because it’s dead. What’s the first thing OEMs do when they hear such a story? Stop ordering parts. Gwak! So if you’re going to EOL a part, you do it quietly and in private with your OEMs—not in public forums that get multiplied by the minute.

After Nvidia realized the flap the hard working and underappreciated PR folks leapt into action and started issuing corrective statements.

On Intel platforms, the NVIDIA GeForce 9400M/ION brands have enjoyed significant sales, as well as critical success. Customers including Apple, Dell, HP, Lenovo, Samsung, Acer, ASUS and others are continuing to incorporate GeForce 9400M and ION products in their current designs. There are many customers that have plans to use ION or GeForce 9400M chipsets for upcoming designs, as well.

We will continue to innovate integrated solutions for Intel’s FSB architecture. We firmly believe that this market has a long healthy life ahead.

We expect our MCP business for both Intel and AMD to be strong well into the future.

We expressed our view on this in March. And to anyone reading this who’s interested I’ll let you have the report at a 50% discount, give me a call (415-435-9368) or send an email jon@jonpeddie.com

Nvidia just needs a little time, and a little less press exposure (they could help that by shutting up for a while.) Things are in motion, Fermi is coming along, it will be a good game engine. Nvidia will still make chipsets for as long as AMD or Intel make them. And they are going to take some lumps until then.

Epilog

Ion 2 should launch in a few months, or Q1 2010. It should have much faster graphics and much more shaders. Since the current Ion has 16 shaders the new one may have more than 32. It should be faster, and may be built in 40 nanometer. This will improve gaming for netbooks and nettops and Cuda applications will run faster. Apple will use it with CULV notebooks no doubt.

If a dog ever caught a truck, what would he do with it? - Or ... be careful what you wish for…

2009-10-06T12:07:11Z

Remember the old joke, we lose money on every unit, but we make it up on volume? Sometimes, I wonder if that’s what we’re doing in the electronics business.

Last week, at two conferences, on two different coasts, we got the lowdown on big to really big systems and the really small systems. At both conferences a senior person commented on the scale of things.

Numbers impress folks, and big is always better. Somehow investors, fans, bloggers, reporters, and product managers translate the bigger numbers in markets to automatic growth. Look at the color of the grass over there they say, reminding me of “Pay no attention to man behind the curtain,” the basic move of all magicians.

And in addition to moving into what they call adjacent markets, they want to continue to dominate the ones they’re already in, but which somehow no longer represent the pot of gold they once did. PCs are nice says Intel but look at those GPUs and mobile phones. Games are OK says Nvidia, but look at those supercomputer and mobile phones. CPUs are fine says AMD, but what about graphics and CPUs together, wouldn’t that be better? Mobile phones are great says ARM, but have you seen those netbooks? And all of them look at the staggering large number of embedded systems and wonder how they can either enter that market or get more of it—tens of billions of units they drool, TENs of billions, and then they fall on to the floor shaking with wild wide open eyes. Tens of billions, every year, holy cow.

Who are the semiconductor companies in this tens of billions market? Well of course Intel is there, as is Samsung, Renesas/NEC, Freescale, TI, and Microchip to name the top players. And many of them use ARM microcontrollers or processors. All these companies and more have a market worth about $30 billion or less. Microchip for example, one of the big names in embedded is a $900 million company. The combined sales of Renesas and NEC, both losing tons of money BTW, for embedded is about $10 billion. Freescale gets about $4 billion of its revenue from embedded. Samsung maybe gets $3 billion out of embedded.

Embedded systems include everything from your watch to your dashboard, your refrigerator to your alarm system, toll booths and ATMs. The categorizations of what is and isn’t an embedded system vary from company to company depending on how good they want to look in that market. For example, are SoCs in STBs embedded systems or CEs? Are SoCs in handheld military systems embedded systems or government & military?

The point is the Embedded Systems market is so vast and so ill defined you can make it whatever you want to hide loses, build up sales on your financial statement, or start a new business. You might just as well call it Electronics, it’s about as meaningful. But there’s tens of billions of them, whatever they are.

And, the tinier and more ubiquitous these things become, the smaller the ASP and subsequently the margin. Now why would any sensible CEO go after a market with lower ASPs and margins – because they can make it up on volume? Did I mention there are TENs of billions of them!?


Device	Annual shipments	Direct connect display	Semi ASP	GM
Supercomputers	1,000	1 to 3 (more remote)	$15,000	35%
PCs & servers	300,000,000	1 to 2	$400	25%
Handheld devices	1,000,000,000	1	$100	20%
Embedded devices	10 to 15,000,000,000	1	$10	15%

On GPU computing—it’s not about languages

2009-09-21T11:52:38Z

GPU computing has come into its own this year and is, I think, well understood and accepted by industry and to a large extent many consumers and scientists. Although I still encounter people in all three categories ( poeple who have been using GPGPU, will use it, will never use it) who say, “Huh?” when I speak to them about it.

Aside from that right-hand side of the Gaussian distribution of adopters, the important part is that all of the early adopters get it, many are employing and/or using it, and almost all (with the exception of some notable high visibility status developers) are planning to use it.

The obvious benefit of using a massively parallel processor that is ridiculously inexpensive relative to other less powerful processors, combined with the OS and programming tools support, is almost overwhelming.

And it raises some interesting questions. Even with the ubiquitous amount of vaunted x86 code, cracks in the technology’s monopoly are starting to show. Highly parallel or multi-threaded programs that were written to run on an x86 are now being re-ported to the GPU and the x 86 versions of those programs are being relegated to legacy code and already described as the “dusty decks” of the 21st century.

Intel hopes to blunt this trend by offering a highly parallel x86 chip code named Larrabee, and AMD thinks they might offset it with Fusion. But it’s doubtful Larrabee will ever be able to compete on a FLOPS basis with the GPUs, and Fusion will just be an integrated heterogeneous solution still employing a GPU.

The advancement in discrete GPUs is astounding. The current ATI GPU will bring 1,600 32-bit double-precision floating point processors to developers and users with something north of 2.5 TFLOPS for under $500. Nvidia will later in the year offer a GPU with not as many processors but a claim to richer implementation of IEEE 754 floating point specifications, and also in the $500 price range. Furthermore these massively parallel processors will have a minimum of 1GB of tightly coupled GDDR5 memory running at outrageous speeds, almost double that of the DDR3 system memory of conventional x86 processors used in PCs and servers.

But let’s not get taken away with the numbers. Remember, the GPUs will only run vector-based SIMD code. Parallel versions of the x86 such as Larrabee will be able to do that and run the more complex MIMD code sets. And the x86 is a 64-bit double-precision floating point processor with IEEE 754 support, plus a small, dedicated SIMD engine, and running at speeds greater than 3 GHz. For single stream, (currently) up to eight multi-stream, and MIMD applications with huge datasets, it will be difficult to beat the venerable x86.

Winners and losers

And who wants to anyway—if it ain’t broke don’t fix it. The GPU isn’t going to make Excel or Word, or your browser run any faster. So contrary to the desires of the press and fund managers there won’t be a winner and a loser with the expansion of the GPU for computing operations. The press and fund managers bring a sensationalist football game mentality to technology. Rather than invest the time to understand it, they want it to be a repeat of the CD killing 8-track, or Blu-ray defeating HD-DVD. And in all fairness, it’s not like there aren’t plenty of examples to call on, but this time it’s different. In the case of GPU compute, it’s an augmentation not a replacement, and that really confounds the press and investors—neither think they can make any money on that kind of development.

It also seems to confuse the players, Intel sees the discrete GPU as a threat, and the GPU suppliers seem hell bent on proving the x86 is a waste of silicon. Fact is they need each other and if we’re going to move the industry forward they’ve got to stop wasting time sniping at each other.

At IBC last year I surveyed most of the software application and tools companies and asked if they were using GPU for compute, if not did they plan to, and if not why? A very small number said they were, a larger number said they were planning, and about a third said no, never.

This year almost everyone reported they were or were planning to use the GPU for compute in some part of their application or pipeline, and several had examples running. They were showing speedup of 10x to 200x over just an x86 CPU on some functions. Matrox and a couple of other companies that don’t have a GPU were showing acceleration using FPGAs.

The time has come

The point to be understood by the press, investors, and the players, is that we now have an environment where the OS can do a pretty good job of load balancing jobs to a heterogeneous group of processors. And these heterogeneous processors can do certain jobs better making the overall results a faster operation. Not only that, with the different processors doing more specific jobs for which they are better suited, each processor is now offloaded and allowed to apply more resources to the things it does best—that’s a win-win situation.

And that’s where we are now—GPU computing has arrived, it’s never going to go away, applications that can be multi-threaded will take advantage of it, and we users will benefit greatly from it. Debates will continue about what programming paradigm to use, Cuda, OpenCL, or DirectX11 compute. If you’re not personally doing the programming this is a useless debate—you might as well invest your time in fantasy football. The choice of programming tools is a marketing debate between the GPU suppliers, and other than your team loyalty burden it has no impact on you personally. Each side will tell you how their programming tools are better and how the other team can’t do certain compute functions. You know what? You don’t care. You’re just a simple user and when and if benchmarks are ever available, you’ll see very little variation because of those programming differences. It’s the heterogeneous processing paradigm that’s making all the difference.

GPU compute is being applied to scientific, academic, and commercial applications, as well as some video-based consumer applications. So soon GPU computing will be used so widely and transparently that in a year’s time it won’t even be discussed—welcome to the future.

Do OEMs, ODMs, and/or consumers value technology?

2009-09-01T12:33:16Z

If company A builds a chip and it is better and newer than company B’s, shouldn’t company A get more market share?

Suppose company B introduced a part in January and it got a Pmark score of 100 and it sold for $150. Then in June, company A introduced a part and it got a Pmark of 150 and sold for $100. Company B, reacting to the competition, drops the price of its 100 Pmark part to $99 and the Pmark for their part went up to 120, and takes market share away from company A. Remarkable as it sounds, this is actually a true story—only the innocents have been changed to protect the names.

One explanation is that company B is better at marketing than company A. And that’s certainly a factor. But it’s not enough.

What does that leave? That the customers don’t value the technology—case in point is On2 discussed earlier in this issue.

Or, maybe they don’t believe the technology messages. Will my application REALLY run faster with company A’s part? Does it REALLY support more applications?

We asked a few people about this, and they generally said, a few more points of performance don’t mean that much to them—the “good enough” scenario that’s fueled Intel’s IGP success. That’s understandable for the low-end Value and Enterprise segment, but for the Performance segment and the Enthusiast segment that is heresy, the unthinkable, the pixel god will come down and smite thee for saying, for even thinking such things. Why your entire family could be turned into turnips.

Brand loyalty is another factor (maybe it’s a subset of marketing skills) and if things are close enough then the brand loyalty influence will kick in. We asked a few folks about that too. And when brand loyalty was a factor, the difference had to be 2x the performance for the same price. Those are customers you’re never going to get. Brand loyalty is largely due to people having their image and ego too tightly associated with a product, letting, or maybe hoping the brand identifies them.

Here again we find a strange anomaly—the brand outweighs the technology.

Consumers, OEMs, and ODMs, seem to assume that of course the technology will be better with every generation. And of course there will be new feature sets with every generation. And certainly the price will be the same or go down. Without necessarily invoking the mantra of Moore’s law they live it, understand it, and expect it. And so technology gets neutralized. Technology isn’t the differentiator the developers had hoped it would be. It’s not that technology is not a factor; it’s just not THE factor.

But you can’t quit. Company A can’t stop innovating and rest on their laurels, assuming they can just rebrand and reprice last year’s parts. Eventually the industry will run away from you. A new operating system will show up, or a new API, or a new game, and you won’t be able to support it. You’ll be left with the long tail customers and even they will one day abandon you. And if you couldn’t get paid for your technology when you were a leader, what do you think your selling argument will be with the long tails?

So many pixels and so much Andouille

2009-08-17T19:55:10Z

This was my 30th Siggraph, I was 10 when I attended the first one… Almost all of them have been fun, educational, and informative. If measured by attendance, this year’s event had the poorest showing since my first Siggraph in Seattle in 1980. However, since the first New Orleans event in 1996, the ratio of exhibitors to attendees was higher this year (1.7% compared to 1.1% then.) Now, that may not be what an exhibitor wants to hear. They want the ratio to be smaller, and the lowest it ever hit was 0.79% at the glory year of peak attendance (49k) in Los Angles in 1997. This year it was just 11K, although it really felt like more.

Siggraph is part of the Association for Computing Machines (ACM), which was established in 1947. By the 1970s it was (and today still is) the oldest computer society. By 1973 the ACM members who grew up with the developments of the SAGE systems and the early CAD programs, as well as Ivan Sutherland’s famous SketchPad graphics drawing program, knew that computer graphics were a powerful tool for all disciplines of science, including medicine, manufacturing, and design. Thus, ACM created the Special Interest Group (SIG) for Graphics. Siggraph is now the largest conglomerate of academics, businesses and industries dedicated to computer graphics and CG animation in the world.

That said, there are only 8,000 registered ACM Siggraph members. Membership costs anywhere from $50 to $150 depending on where you are in your career—I’m pleased to say I’m in the $150 category. And when offered a 66.6% discount for being a member of the old farts club (Siggraph Pioneers) I declined. I’m one of those people who thinks we should pay taxes and I’d like to see the ACM thrive, but that’s a rant for another time and venue.

Andouille is a spiced, heavily smoked pork sausage. It’s unclear whether it originated in France, where the name comes, or in Germany, or maybe Canada. Nonetheless, the French colonists of Louisiana, and the Cajuns (descendants of evicted Acadians from Nova Scotia) made this stuff from meat and fat, seasoned with salt, cracked black pepper, and garlic, and smoked over pecan wood and sugar cane for up to seven or eight hours. It is the base for almost all soups, Jambalaya, mixed with shrimp, chicken, even crocodile (if you came to our luncheon you got to try alligator [tastes like crocodile]) and almost everything else that dares call itself Cajun.

And here’s the point - Andouille is to New Orleans as Pixels are to Siggraph – you can’t get enough of either one, and you’ve probably never met one you didn’t like, and in both cases the really good ones make your mouth water.

How I spent my summer vacation

2009-08-06T17:16:24Z

School runs, in most parts of the world, from September to June, with a few breaks for religious and political celebrations in between. And school’s out in the summer, and August is as summer as it gets in most places in the northern hemisphere.

In most parts of the U.S., other than San Francisco, it’s usually pretty hot, pretty damn hot, in August.

ACM’s special interest group in graphics, commonly known as Siggraph, has always had (and I hope always will) a strong link to the universities. The next time you go to Siggraph, spend a little time looking at the posters, and emerging technologies to get a feel for just how Siggraph is influenced, if not driven, by academia.

Therefore, Siggraph the congress/convention is held in the summer so students can attend, look for work, and deliver their papers.

In order to make the conference accessible, Siggraph has kept the entrance fees low. Back in the glory days of the late 1980s when Siggraph attendance and the number of exhibitors were high the organization made a little money. But basically it’s been a hand-to-mouth operation relying on volunteers to get it put together.

Siggraph has had to find a low-cost venue to hold the conference, and because many people bring their families with them, the site selectors try to find a fun place—usually. And with one or two notable exceptions (1980—Seattle, 1985—San Francisco) the places they’ve picked have been HOT AS HELL! Who picks these damn infernos? I remember one year in Dallas and one year in Las Vegas—Las melt-my-shoes-in-August Vegas when I thought I was going to die if my clothes didn’t catch on fire first.

This is the third time Siggraph has been in sweat city on the Mississippi. Obviously there’s no community history in Siggraph. The volunteer kids who run it get jobs and don’t come back for a few years. The exhibitors have been dropping like flies and don’t seem to care, why the hell should they, they’re inside all day in an air-conditioned exhibit hall.

The last time Siggraph was in New perspiration Orleans the air conditioners on the mile long building broke down—boy was that pleasant. But no one seems to remember this stuff except me.

Siggraph began in 1975 with the first conference in Boulder Colorado. I didn’t attend that one, but I think that’s a really swell place to hold the conference—it’s never been back.

Starting with Atlanta in 1978 Siggraph has managed to find the hottest possible places in the United States for the conference 85% of the time.

And global warming, which regardless of your religious position on why it’s happening, is not going to make this situation any better. We have got to relocate Siggraph to the North.

Now, if LA is the big attraction because of its association with the movie business, then how about Vancouver?