Intel Developer Forum, Spring 2004

Sean Maloney
Executive Vice President, General Manager, Intel Communications Group
Intel Corporation
San Francisco, Calif.
February 19, 2004

SEAN MALONEY: Well, good morning, everybody. Welcome to this last day of IDF. I suppose an interesting question to start with is to ponder for a second on why we are here. Now, I think that we're here, as a community, we gather around the world on a regular basis. We're here to develop technology that our customers, people across the world, will find entrancing and interesting to the point where they will start using it and weaving it into the pattern of their lives.

This morning I read a beautiful little article in the USA Today newspaper titled, "Wi-Fi Changes Virtually Everything. Users say they will never go back." This marks the continuing movement of Wi-Fi, which was a huge topic here two years ago, the movement of that technology into the mass consciousness.

The article makes three pretty interesting points. The first point is that Wi-Fi users use technology more than non-users. Now, we've known for some time that a broadband user uses substantially more computing resources than a non-broadband user. Recent data from Germany comparing broadband users versus non-broadband users shows that the broadband user uses the PC about three times as much as the non-broadband user. We instinctively felt that that would continue with wireless, because people will be able to use the computer even more as they walk around with it. That was the first point.

The second point is that people are doing more different things with notebook computers and wireless computers. They're extends the reach of PC technology into other parts of their life.

The third point, which is almost the most interesting, is that the content industry is already beginning to move and change its content now that there is this emerging group of people who are mobile. And it lists a number of Web sites that are beginning to cotton on to the fact that people are carrying their computers around with them and beginning to change their content.

This sense of optimism or kind of renewed optimism in the industry is really the theme of what I'm going to talk about here today, which is what we can call the broadband wireless era.

You can define the 1990s as being very much the Internet era, the last two years of a very rapid, severe correction in the industry. And what we're seeing now is a rising sense of optimism based around various forms of broadband wireless technology, of which probably the most controversial and fast-moving is Wi-Fi.

To me, this has picked up huge steam in the last four months or so, starting off in the telecom industry at ITU Geneva back in the fall, when, after two years of agonizing about, you know, "Have we spent too much money on 3G licenses?" "WAP doesn't work," and so on and so on, the industry kind of picked itself up, and the whole theme in the conference, this being the kind of most important conference of the telecom industry, $100 billion industry, the whole theme was about wireless. It was about Wi-Fi, WiMAX beginning to show its head, and 3G in various forms, whether it's UMTS and the global world of UMTS or also CDMA2000 and EV-DO.

So the telecom industry very much is beginning to rally around that. If you look at the equipment that's being deployed, overwhelmingly, the new equipment that's going in is for these high-speed wireless networks.

We had the Consumer Electronics Show just one month ago, which is for the consumer electronics industry. Again, that show was heavily influenced by broadband wireless, and the arrival in the home of a whole series of gadgets and appliances that will hang on to wireless networks.

The arrival of Wi-Fi has unwittingly made home networking happen. So, as you can see from the demonstrations outside and on the floor, there's a whole series of Wi-Fi gadgets that are rolling in. Wi-Fi is beginning to get what economists call increasing returns, more users, more uses, and producing a spiral of growth.

After that, we also are beginning to see broadband wireless begin to impact the content industry in a whole number of different ways. I'll focus on one particular example, and that is, we went to the Sundance Film Festival about three weeks ago. This is the largest independent film festival in the world. And what was interesting to us is that it's a community that is extremely picky about the video quality and extremely precise in the requirements that are needed to get a high-quality image.

What we did there was we streamed the world's first movie premier on Wi-Fi, and we used the Microsoft Windows Media 9 High Def which has really got an exceptional quality. And we streamed it live over Wi-Fi. I'll show you a little video clip of the response from some of the people there.

(Video plays and ends.)

SEAN MALONEY: Over the next year or two years, you're going to start to see more and more use of wireless as a transmission medium for high-definition content. It's something we should pay very, very close attention to.

The fourth area where we're seeing a growing role or involvement in wireless is to reach the next billion people. Now, that's not just people in emerging market segments, where you tend to see most of the press about Wi-Fi and WiMAX and so on. This is also in the mature economies. In the United States, 65 percent of homes don't have broadband. If you look across Europe, the numbers are very, very similar. And, you know, it's always better if you can get fiber to run to a home, because it's always better to have a connection that you can scale and put more and more bandwidth down.

Unfortunately, there isn't the money in the industry to roll out fiber to a billion people's homes. And copper has reached its limits unless we re-lay out the entire network and get everybody within 500 to 600 feet of a central office, which won't happen. So whilst fiber and copper will continue, there is now this chasm to go and reach everybody else who doesn't have the luxury of either having a high-speed cable or a copper line going to them. And that group of people is the overwhelming number of people who are going to get connected to the Internet in the next five to ten years. So that's an area where, in the last six to nine months, there's been a huge pickup in interest in how you can use broadband wireless to reliably delivery very high-speed data. I'll be covering that a little bit later on.

So what I want to do is spend a short period of time looking at some of the challenges that, collectively, we face in developing our technology to make it usable for that era, and I want to start with the first challenge, which is to improve the client, and to start with, just show you an example within about 20 miles of here of how a local police department is using broadband wireless and what they do with those client devices, how they use it and what they get out of it. So let's roll that video.

(Video plays and ends.)

SEAN MALONEY: Those computers are not doing anything new. They're accessing existing information. But the point is that because the computers go with the person, time is taken out of the cycle of work that those officers have in their lives.

In Hong Kong recently, we observed that as you walk into McDonald's, somebody is standing with a Wi-Fi PDA, and they take your order. And the order gets relayed back. So by the time you walk up and order again at the desk, your food is already under preparation. And they said it takes about one minute out of the cycle, which isn't a lot of time. But, of course, a minute by a billion burgers is a big deal.

All through the 1990s, what the Internet really did in business was it took time out of cycles. It enabled people to get information all around the world. It enabled people to get access to information much more quickly.

Broadband wireless takes this to another level, because the information will go with you. And that will be the common theme in our customers out through the next three to five years to seven years as broadband wireless gets deployed. Business model after business model, whether it's police, health care, or consumer retail, is going to get changed as people squeeze more and more time out of their cycle by using broadband wireless.

You also get some kind of unintended consequences of technology. And one of the ones I really like is the unintended consequences of having cameras on PCs.

I'm going to show you here a very interesting technology called DigiMarc that is developed here in the United States. Basically because cameras are on phones -- this is almost ubiquitous in Asia and Europe -- essentially everyone has a scanner. Instead of thinking of it as a camera on the phone, it's a scanner.

So what these guys have developed is a way of embedding a watermark image inside of an advertisement. The image is not really visible to the human eye, but it's visible to the camera and the computer.

You point your phone at the advertisement, and the phone recognizes this hidden symbol in there and immediately goes off and accesses information off the Internet to tell you about the product. So I'll give you an example here.

So this is a PDA with a camera. It could be a phone with a camera. And I click on the application, point it at the logo -- point it at the picture of the shoe. It found the logo, and immediately went off to index this content.

So these technologies are already being deployed in Korea by Korea Telecom and others so people can walk around with their phone, hold it, point it at an advertisement and then it will get indexed on the Web to get content from it. It's a fascinating unintended consequence of what happens by deploying technology.

So I divide up the challenges certainly as Intel sees them from a hardware point of view, divide them up into four areas, which is the memory, the applications processor, the graphics subsystem, and the communications capability. And so let me start by looking at the memory.

Computer memory in mobile devices needs some work. I'll give you a little example. When I turn on my Intel® Centrino™ mobile technology notebook, it takes about seven seconds from sleep to find and associate with an access point and be on the Internet.

When I turn on my phone, it takes 21 seconds before it's in a position where it's looking for a GSM signal, not uncommon with high-end phones now. One reason for that is the memory systems are slow and these phones need more and more memory. So there's a huge amount that needs to be done over the next two or three years to increase the performance of these devices so they come on instantly and you can search through information on the phone or delete it far more quickly.

One way we can do that is by shrinking the geometries, Moore's Law. There are some question marks whether or not you can do that on ETOX, the technology for Flash nonvolatile memory. We're very happy to say that we can. And we're announcing the world's first 90 nanometer NOR memory.

We obviously get cost benefits, we get increased density benefits, and we will be in production shortly on this.

This technology will enable us to make further strides, increasing density, reducing latency, and getting to much faster boot times and so on. We have some other tricks that we're going to be doing in this also to address those problems.

We get an additional benefit because we do the StrataFlash multi-bits per cell, we get significantly better increase in effective density, which again helps in a whole number of different ways.

The second area is in the space of the applications processor and the graphics. Six months or so ago, Intel talked about its first generation communications processor technologies. We said our business colleagues would be going into production during 2004 with those phones, and that's looking very much on track. We're also on track for our second generation of communications processors, which will be in the first half of 2005.

The other thing that we're looking at is getting much faster graphics performance into phones and PDAs.

I want to show you now a piece of technology which is designed for very low power, high performance graphics, and it's a device demonstration here we're calling Carbonado.

This is a specially designed low power graphics 3-D accelerator chip to go along with something like an ARM architecture compliant Intel XScale® processor. It's going into production this quarter and has an effective throughput of up to 3 million polygons per second.

A consequence of that is you can see I have two of them here, the one on the left is running streaming video and it has remarkably clear image for a handheld device. On the right-hand side there, you should be able to see the lights are bright so it's not clear, but there's a very unusually real 3-D feel for handheld gaming devices.

This is a very interesting category over the next three years. Those Wi-Fi networks are going to be increasingly common. You also have widespread UMTS and CDMA 2000. So there's going to be reasonably high-speed data available. The idea of handheld gaming devices and devices that also run short video clips is an idea that's catching on. I know many of you are involved in designing those.

More computer power helps for sure in addition to the graphics, and so we will be carrying on moving ahead with our next-generation applications processors and scaling them down through the geometries. And we're happy that there are lots of new architectural features we can put into those devices to help the graphics and video and all those other kind of applications as well.

In terms of the pure communications capability, we see a balance there between 3G, GSM, and Wi-Fi. You'll have some devices that have all of those capabilities, other devices that only have one of them.

Wi-Fi has been talked about a lot for handheld devices but hasn't been implemented a lot yet and one of the reasons is the power consumption used. So we will be going into production later this year with a specially designed low-power Wi-Fi device. I'll give you an idea of it.

This is the size of the kind of existing Wi-Fi module inside of an Intel® Centrino™ mobile technology notebook. This is the current state of where we are on our integrated Bluetooth Wi-Fi device, and we believe as we go into production we'll be able to shrink it down even smaller to be about the surface area of that little black chip there.

Again, putting things on small geometries, redesigning around low power and so on, all of these things are useful for extending the capabilities of the client. So the goal on that handheld Wi-Fi is high throughput and low power consumption and high integration.

So a final comment on handheld technology and mobile technology concerns the radio itself. The RF system which sits on a phone or on a broadband wireless device. If you open up a phone or any kind of device with radio capabilities, you see that there are lots of external discrete components -- amplifiers, filters and so on -- surrounding the core of the analog design. A lot of that really hasn't been innovated very much over the last five to 10 years. It means the designs are still complex with a lot of components on it.

Now, a year ago we talked to you about the work we were doing on MEMS technology, which is a way of using classic semiconductor production to do mechanical-based design. The same technology we're also very, very interested in using for shrinking down radio designs, and we've made a lot of progress since we spoke to you a year ago, and we believe we're going to be able to do something very interesting here, which will significantly reduce the complexity of cellular handsets and clients reduce the number of components, reduce the design issue and increase reliability. I'll be able to give you a much more detailed update on that at the next IDF.

Finally, on Intel® Centrino™ mobile technology itself, obviously, Intel Communications Group is the group that's producing that. We are focusing in the client on high throughput and reach. Reach is a very, very big deal. People want to be able to get a signal as far away as possible. And then also in speed of association, how quickly you associate with the access point. And of course compatibility.

We have a large, very well-established team on that now, working on their fourth generation design, and things are looking good in that space. So what I'd like to do, then, is move off the client and move onto the infrastructure side, and what we talked about last time were modular standards-based networks.

We talked about the progress that was being made on ATCA. ATCA is an open standard that the communications industry is adopting for multi-function devices for 3G base stations, for routers, for metro switches and so on. A lot of progress has happened in the last six months since we were last together.

I'd say that modularity is sort of happening. Pretty much all the equipment manufacturers have announced ATCA-based designs using either Intel technology or other people's technology. And this market segment as a whole is opening up to be a very substantial opportunity for equipment manufacturers and component manufacturers as the industry is coalescing around one standard for physical layout and for interconnect.

Now, one issue that we raised six months ago was the need for a very high speed interconnect so that these products like these next-generation switches and routers for networks where hundreds of millions of people are carrying data phones and people are carrying wireless notebooks and so on, all that traffic increase, we have to scale up the interconnect to handle that traffic.

We worked with a number of other people in the industry to establish the ASI SIG. That was about six months ago. Here was the list of the Board of Directors we announced at the last IDF. Substantial progress since then. A lot of other companies have come along and supported this standard.

Thirty days ago, the specification was finalized. I'm delighted to say that actually today, we can give the industry's first demo of hardware on this, and I would like to call up Sandeep Vij, who is a vice president at Xilinx. Sandeep, come on up.

(Demo begins and ends.)

SEAN MALONEY: So, those of you who have been involved in this have known how quickly this thing has moved in the last six to nine months. The industry coalescing around a standard, the definition coming out, and then, I guess, the message here today is this thing is real and it's arriving.

The time line is the design phase is well underway. Today you've seen the first hardware, the world's first hardware demo of the technology. And you're going to see the systems deployed out over the next year or two years. So I think the message is, if you are designing systems that need high-speed interconnect, this is real, it's here, and it's the new global standard.

Now, these standards are open standards. Both ATCA and Advanced Switching, of course, are open standards, using Intel and other technologies. Our design teams are focusing on using these open standards. And part of that has helped a significant buildup in momentum behind the Intel® XScale™ processor family moving from fourth position up to currently rated at the number one position.

The interesting thing here is In-Stat drawing attention to the world of the ecosystem. So it's much more than Intel involved in this. It's a whole ecosystem of companies focusing on using technologies based around the Intel XScale processor.

So we're obviously going to keep that moving forward. We're moving into our second and third generations of designs, keeping code compatibility, obviously, as we go, increasing the number of packet engines. That means that we can increase the amount of traffic that can be handled moving up from one-gig speeds up through 10-gig speeds and above. As the greater number of packet engines and faster frequency means we can do more and more things, more and more things intelligently, on the traffic.

This also raises the issue of higher-speed I/O and 10 Gig, which has been pretty slow over the last two years in taking off. We're beginning to see signs that 10 Gig is kicking into more life.

Up here, on stage, we have a next-generation 10 Gig and one-gig switch. This is an Intel product which uses 90-nanometer technology, 220 million transistors, with a large number of 1 Gig and 10 Gig ports going into it.

We also have over here our second-generation 10 Gig Ethernet card. The first generation was larger with a lot more components. This one, now we've managed to shrink this down with a much smaller 10 Gig optical interface on it and the 10 Gig optical transponder on the edge here.

As one gig has moved into the networks, as more and more devices come on, the traffic level moves up, and at the aggregate point, the industry is beginning to deploy 10 Gig.

You will see again more and more 10 Gig products coming out over the next year or so. In addition to that, we have a need to drive down the cost of optical interconnect at slower speeds. So today we're also announcing two new products. These are four-gigabit per second optical transponders. And they are very small devices, highly integrated devices, both SFF and SFP package-compliant.

So the message there is that modular networks, modular design is definitely happening in the infrastructure business. ATCA is strongly picking up momentum. The switching devices are making substantial progress. And then also, the cost-reduced optical components and on-board components are also making significant progress.

So the third and final area I want to finish on is on building the next-generation wireless infrastructure, and specifically focusing on the world of WiMAX, which is like the big brother or big sister to Wi-Fi, and what WiMAX will be able to do for our customers.

I'll start by showing you something. This is a map of the world with countries with more than 50 percent broadband deployment highlighted. This is not a very good indicator. This is not a particularly good thing. The only two countries in the world that have more than half their homes connected to broadband are Japan and South Korea.

As I mentioned earlier on, there are probably insurmountable challenges in connecting everybody else up with fiber or getting the copper runs short enough that you can get the kind of bandwidth that you need to do smooth video and kind of next-generation Internet experiences.

So along comes WiMAX. Now, wide-band radio has been under development, really, for about 10 years or more. Lots of technologies have been tried, and millions of units have been shipped. But we haven't had a kind of vertical liftoff. And the reason is, there hasn't been an IEEE standard finalized that the whole industry can agree on.

That has changed with the arrival of 802.16 and WiMAX. There is now a single global standard which more and more companies are beginning to focus around. We see it as basically going in three visions.

Intel will be making components and modules for this. Other folks will be doing the same thing as well. We see the outdoor install in the first half of '05. The application is broadband to the home, so people putting WiMAX into existing cellular base stations or putting up new WiMAX low-cost base stations. Then a consumer would hang a simple antenna or something on the outside of their home.

Second phase we see in the second half of 2005, which is indoor install, where the link budget and the technology will be mature enough for you to be able to just take a device in the home and the signal will be able to penetrate the wall and give you broadband performance into the home.

Then the third phase, which is the portability phase, we see in 2006, where the power consumption gets down into an acceptable threshold and the specification, 802.16e, which has the portability and quality of service features, will be available.

We think this is a big deal. We think it's very good news for the industry. It's good news for people developing clients. It's good news for people developing servers. It's good news for the semiconductor business, because it will enable people to use the technology more and more.

Clearly, the goal here is to take our little area and get all of the world or as much of the world as possible, connected on high-speed broadband.

Thank you for being here today.

* Other names and brands may be claimed as the property of others.