Their main message was that we do have choices when it comes to processors and what sort of systems you can buy nowadays. For many years, there was really one choice, but now they are making a pretty good run at it. You have a lot of options. Whether you are in the value space market for some of the lower processor speeds in the 600 range, or at the very high performance end, AMD has got a solution for you.

Alex was in the middle of a scandisk that hung on him and asked that we bear with him for a second. While we waited, he wanted to know how many people have an AMD system right now; he was impressed with the response.

At this point the questions started: What are the prices now based on speed versus price? What is the speed of the Athlon in megahertz?

Alex explained that the Duron processors are fairly new. They take the place of the K6 line, and should be in computers within the $800 to $1200 range running at about 600 MHZ to 750 MHZ, the high end for the Duron. In the Athlon there is a wide range depending on the peripherals and everything else; those computers are about $2,500 and up, or slightly below that.

The Athlon processor family is a family of high-end processors that range from 600 MHZ all the way up to 1.1 gig, which was just released this week. The Duron processor family is what AMD terms as a value, low-cost solution; and those range from 600 MHZ to 750 MHZ, priced at about the sub-$1000 level for a complete system.

AMD, he said, emphasizes processor technology, and that they spend a lot of money on R&D. He first of all just wanted to give us an overview of who AMD is. Most of us know, he opined, that AMD provides processors for personal computers and also communications devices for networking as well. Last year, they had revenues of over $2.9 billion and had a record quarter in this last quarter. He pointed out that they are the world's second largest supplier of Windows-compatible processors, and that they are sold out for three years.

Alex mentioned FAB 30, which is in Dresden, and which was in full production this year. He indicated that it now is able to build processors using .18 micron technology. It is also ready to use copper and is moving towards .13 technology. Copper technology is very important when you are moving into microns of.13 and .18, which are the actual gate measurement of the transistor; the smaller the width, the faster the processing speed, or the faster the electrons cross through the gates.

AMD also has FAB 25 in Texas, which has been up and running for many years. AMD shipped more than 18 million units of the K6 and Athlon chips last year. At FAB25 .18 micron technology is used but using only aluminum, while in Dresden, as he mentioned, both copper and aluminum are used. With the two together, they have the capacity to ship 11,000 eight-inch wafers per week.
Q:How many do you get out of a wafer?
A:Jeff wasn't sure, but though that there is a variation in the yield.

A rather lengthy discussion followed concerning issues of yield with the K6 processor, but Alex pointed out that the Athlon has something like 90 to 100 percent coverage. When a wafer fails, they can test the die to isolate the failure. This approach was not available for the K6. On the other hand, the K6 was not designed to run above 300 MHZ. The design of the K6 was improved, and was then able to run at from 500 to 550 MHZ. The K6 and the Athlon are, however, two different beasts, he explained. The Athlon was started from scratch and was designed for the test, which is why AMD is not seeing any processor-related problems with the Athlon. The interconnects in the Athlon are copper; all the Athlons coming out of Dresden are copper.

Trends, of course, in the electronics industry mean that smaller is better, faster is better. We should notice, he said, that their K6-2 started out at .35 microns and now is down to .18; the Athlon started at .25 and is now at .18 microns. The die of the Athlon is much bigger than the K6-2. Now with smaller technology, he said, we can have higher front-side bus speeds, and we can run up above a gigahertz as we move forward toward smaller technology.

(The front-side bus, he explained, is the interface between the processor and the northbridge; the northbridge has the memory controller, the video controller, and the link to the southbridge, which is your link to the PCI bus. The front-side bus is at a 200 MHz data rate. A backside bus links to the L2 cache.)

When AMD first introduced the Athlon, it came in a cartridge or module and it came with two discrete 512k caches. This was the external cache. When it was first introduced it was running full speed at 500 MHz. He had a demo here that was passed around. As the speeds got faster, it surpassed the standard SRAM devices with regards to latency, so they have decided to integrate the SRAM into the die itself.
Q:Don't the chips require some type of microcode or software to run. Doesn't each company have its own microcode or proprietary process to get complete compatibility?
C:He is talking about the 8086 command set compatibility with the Intel chip.
A:You mean like X86 applications?
Q:Is the Athlon completely compatible? If the microcode is different, how could it be?
A:He is asking if there is proprietary microcode that allows a processor to run, so you could run X86 applications. Is that the gist of your question?
Q:It is a compatibility issue between Intel and yours, whether each have proprietary microcode. I do not know, I am just asking.
A:The processor is compatible with X86 applications. It uses the X86 instructions, if that is what you are asking.
Q:Does it use its own microcode?
A:I don't know the answer to that.
Q:I think he is asking the question because both of you, Intel and AMD, use microcode for your systems, and legally your microcode couldn't be the same as Intel. Because you have different microcode, would you use the same subset control codes for the applications that run on Intel X86 systems, run on your machines?
A:Intel has their own microcode. AMD has their own microcode right now. Years and years ago, Intel licensed microcode to AMD, and then there was ... a problem down the road .... That was many years ago. So if we are talking about the chips that are out nowadays, Intel has their own microcode within their processors. AMD has their own microcode. But, the end result is compatibility in those applications. The only thing I can think of where you might think it is an incompatibility issue, is that the software may be looking for a CPU ID or a manufacturer ID, and if the application does not recognize that, it might think it is an Intel processor, or it might say it is an unknown processor. But our job is to keep up with all of the applications out there and say they need to incorporate our ID, our manufacturer ID.

He continued to talk about platform trends where graphics is very important. Graphics are becoming more and more intensive. In the future, there is going to be something called UMA, Unified Memory Architecture. Basically, memory is taken off of the video card and incorporated into the system memory on the motherboard. Motherboards will be developed incorporating this technology.

Alex then explained how AMD is working side-by-side with companies like nVidia. He showed the complexity of the processor and the complexity of video cards, and video chips, and the number of transistors that the Athlon uses as well as the video chip manufactured by a company like nVidia. AMD's job is to keep up with this technology, because as graphics become more intensive, they require a lot more processing power.

One of the things AMD likes to emphasize is the data bandwidth deficit that they realized on previous processors. It is also important to note that they found the bottleneck in the previous processors before the Athlon was at the system bus, which is the front-side bus.

In his demonstration, Alex pointed out the AGP card, the system memory, the PCI, and the legacy devices. His purpose was to show that by adding up the bandwidth for each of these links, you would get a total of 1.4 gigabytes. And considering the Athlon's front-side bus at 8 bytes, at 200 MHz, bandwidth would reach 1.6 gigabytes per second: plenty of bandwidth to accommodate current requirements, but we are at the border of current and future bandwidth requirements. But the point is that there is plenty of bandwidth on the front-side bus to accommodate all of the current interfaces. Now the bottleneck is at the system memory. In the future we may require a total bandwidth of 3.2 gigabytes per second.

Alex then turned to ways to increase the performance of the Athlon. One of them has been carried out already. AMD has provided a full speed on-chip L2 cache and has been talking about raising it up to 2 megabytes of L2 cache at integrated full speed. Currently they have a 200 MHz front-side bus. It is a 200 MHz data rate, which is not to be confused with a 100 MHz clock, but it is 100 MHz double-edged, which means that data are pumped on the positive and negative edge of the clock. In the future, we will see a front-side bus of 266, which is 133 double data rate.

Next we will see what they call PC2100 and PC1600 DDR. 2100 indicates 2.1 gigabytes per second, and PC1600 1.6 gigabytes per second, which utilizes the standard SRAM interface, but pumping data on both edges of the clock, which is how they get the bandwidth with a current SRAM.

You can see what is up for the future visiting their Web site. You will find a presentation at the microprocessor forum. It talks about LDT, which is Lightning Data Transport. It will be moved away from the actual PCI bus. It will bump up the bandwidth to about 3.2 gigabytes per second.

Alex then brought up the subject of multiprocessors. AMD is going to enable some of its partners to develop chipsets to support Athlon in a multiprocessor environment.
Q:Intel is selling several boards now that have dual processors. Have you got any manufacturers supplying AMD dual processors in a high-speed range?
A:We are enabling partners to develop multiprocessor systems.

When you compare the Intel processor to Athlon here are a few things to consider, he said. Again, the front-side bus is 200 MHz. AMD has a proven technology from DEC that is scalable to 400 MHz. It uses a point-to-point dedicated architecture that provides a dedicated 1.6 gigabytes of bandwidth for each link, for each processor, as opposed to a shared link, or shared bus; it is scalable.

As to platforms, in 1999 they had the K6, the Super 7 platforms with the front-side bus of 100 MHz using PC100. Then they introduced the Athlon in a slot configuration with the front-side bus of 200 MHz, using PC100 and 2X AGP. This was called an ADM-750 chipset, which is the northbridge. After that, VIA was taken on as a partner, who has the technology to develop a northbridge; they came out with the KX133. The difference is that they use a PC133 interface, and support the 4X AGP. VIA developed another chipset, called KT133, which supports AMD's socketed processors, basically the same thing as the KX133; the only difference is it uses the socket processor. AMD thought that they would have their own flavor of the socketed platform using the AMD-750 chipset, the current chipset. They do not see any value in it, since it only supported 2X AGP, and since VIA was doing so well with their KT chipset, AMD thought they should focus on what they call IGD4, which is the AMD-760 chipset, which supports 4X AGP. It is going to have a front-side bus speed of 266 MHz; it will support DDR memory. We should see this soon, he reported, and see very good results and performance.

AMD will have a socketed motherboard that will tap into both performance space and the value space and a platform that can support UMA (Unified Memory Architecture). You could, though, override it and plug in a 4X AGP card and have it in the performance space. This board could be used with a Thunderbird with an Athlon or Duron. AMD also has another flavor of the motherboard that will move further down into value space, which will only have Unified Memory Architecture and will not have a 4X AGP override.
Q:For people who may not understand you on that, you are saying it is going to have something like an integrated graphics chip put on the motherboard that works in tandem with the chip rather than plugging in an AGP graphics card?
A:Actually, that is correct. The video controller will be integrated into the northbridge, and then it uses the system memory for video memory, so if you have UMA integrated into your northbridge, you have to buy more memory for your system. Is that understood? UMA architecture is actually video integrated onto the northbridge or onto the motherboard. So you don't need a video card. You do not take up a slot.
C:That's sharing.
C:On a Compaq or Hewlett-Packard, one of the two, they share.
A:You are going to add to your system memory, because now you are using...
C:You need more system memory.
A:Right, you need more system memory for the video. Cards nowadays have 8 MB of memory, or 64, so you would have to add to that when you are using Unified Memory Architecture.

He further explained that further down the road, in the workstation and server space, they will have another platform that will support dual processors, their AMD-760 MP: a socket-only, front-side bus, 266 MHz with DDR technology. They are working closely with all of the different motherboard, chipset, BIOS, and graphics vendors. The list is growing tremendously, he reported.
C:I saw a newsbite today that some of the specifications for the new K8 are supposed to be announced within the next few days.
A:For Sledgehammer?
C:They did not say a name. Just 64-bit architecture.
A:Oh, yes. We are going to enable our partners to go ahead and develop applications for x86-64, and a big part of that is going to be Linux.

Among their processors, everybody knows about the K6-2 and the K6-3 [K6-III] with the Super 7 technology. There is the Athlon at .18 micron slot configuration in late 1999. There is a new flavor of the K6-2, which is mobile only, called the K6-2+ and K6-3+, also .18 technology. It has additional features for power-saving mode, which is why it is only enabled for mobile. It has an integrated L2 cache running at full speed. There is the Athlon, both slot and socket and, of course, there is the Duron. When it first came out, or was talked about, it was called Spitfire. When the Athlon was discussed in the performance space, it was called Thunderbird. AMD didn't want to lose the name Athlon, and kept the Athlon name with Thunderbird, the Thunderbird for the performance space. This is the AMD Athlon with a 384k on-chip cache that is really Thunderbird. They want to keep the name Athlon so that their users and customers wouldn't think they had a new, different chip.
Q:Your socket is how many pins?
A:462
Q:One other pertinent question I think is, what is the power consumption of these different chips?
A:Right now we are looking at around 50 to 60 watts I believe. You are going to see that is going to get lower as we move forward. Duron is slightly lower than that; it is probably in the 40 range.
Q:What is the heat compared to the Intel chip?
A:Very hot. You are talking about 20 million transistors, active transistors, I mean. Well, 15 million out of the 20 million are active. This part runs very hot, and we have very strict requirements on heatsinks, and I could show you why there is a strict requirement on heatsinks. Everyone asks, "Why can't we use the Intel heatsinks?" You can't. Different material, different surface area, different number of fins, different clips. The clip is probably one important thing about these heatsinks, and I pointed that out to a couple of our partners, and I could share that with you today as well.

In the workstation service space, he went on to explain, they have what they call Mustang, which is targeted toward the high end or the enterprise. It is going to have a very large on-chip cache, L2 cache for a socket. AMD is planning to align Mustang with the multiprocessor environment. But you could take either one of these processors and plug it into any socket 462 motherboard. The pins are compatible. The difference is that the mobile product will have additional pins for power-saving technology and in the non-mobile product, those pins are not connected. They will also have Palomino, a flavor of Mustang, that runs at lower power with the power-saving technology added.

In the future, there will be Sledgehammer, their eighth-generation processor. It will have support for 64-bit applications and will be backwards compatible for 32-bit applications as well.
Q:Can you tell us if your support technology for laptops will increase battery life, by basically clocking the computer down when it is not doing anything?
A:Yeah. That is what I mean by power-saving technology for our mobile products. It is what we call Gemini.

Alex said that he couldn't leave without telling us about K6. It is still going strong and is sold out for the rest of the year. It is going to sustain its competitive position, as is the K6-2+. K6-2+ has PowerNow!, their power-saving technology. And here is a summary of the differences between the K6-2+ and the Celeron. They say that it supports L3 cache, but he would not recommend that for a motherboard because you don't really gain much performance if you have an L3 cache. For embedded applications that are cache-intensive, this would be ideal, he reported.

As time has gone on prices have gone down, especially since the introduction of Duron. Jeff reported that the K6 has not gone down recently because it is in high demand and there is not much supply. Prices have remained in a range, while Durons and Athlons have dropped down. As of yesterday there was a big price move down.
Q:All of my computers are K6-3. Why did you eliminate the K6-3?
A:It wasn't selling as much because K6-3 with the on-die cache was at 400 MHz. Then we had the K6-2 at 450 and 500; people were buying frequency. They were looking for higher frequencies, and they didn't understand that the on-die L2 cache on the K6-3 was actually higher performing.
C:Everyone that is intelligent knows it.
A:Well, it wasn't moving. We were not selling enough. Then Duron came along, and we decided let's not have so many flavors of the processor, let's stay with the K6-2, people were all buying the K6-2, lets move to Duron and sell those.
Q:A K6-Duron is much better than a K6-3?
C:Oh yes. Much better.
A:You've got on-die full speed L2 cache; you are running at 600, 700 MHz.
Q:And a Duron is your current product, or not?
A:Yes.

AMD has processors for every segment of the market. Alex did not show the enterprise because they don't have the solution for that yet, or for the platforms to support the workstations and server environment, but they soon will. They do have Athlon, Duron, and K6 to fulfill the performance and the value space.
Q:Lets assume I've decided to go to a new system tomorrow; what would I choose between a Duron or an Athlon? How would I determine which one I really want to use?
A:What system do you have now?
C:K6-3.
A:You would have to change your motherboard.
C:Oh, yes.
C:He wants to know why should you buy a Duron, why should you buy an Athlon?
C:That's correct. You got it.
A:It really depends on what your application is. Lets see, how do I explain?
Q:Which one is the gamer machine, and which one is the...
A:For gaming, I would go Thunderbird, I would go Athlon.
A:If you are just looking at clock speed, the Duron goes from 600 to now 750 MHz. But the Athlon, being the high performance, goes all the way actually from like 700 all the way to 1.1 gig.
Q:And the Athlon has a larger cache?
A:Right, and that is what makes it higher performing too. I don't know if that helps, but it helps you to define I guess where you should be, depending upon your application. But it doesn't matter what motherboard you buy. You could buy an Athlon motherboard now, a socketed motherboard, you could plug in either one.
Q:What about Dragon Naturally Speaking, or something like that?
A:Yeah. Either one. But what I would do, I would recommend you go with Duron, and that gives you the upgradeability in the future to move to Athlon without changing the motherboard. Or just skip a step and just buy an Athlon.
Q:Is there any software at present that supports either 3DNow! or Intel 3D, or whatever it is?
A:Are there applications that support 3DNow!? I think there are quite a few. When you play games, they use Windows DirectX technology, 3DNow is supported in DirectX, so all of your games will be accelerated automatically for you. But if you talk to me about things like PhotoShop, no, there are not accelerators for PhotoShop.
Q:What is the price difference between the Duron and the Athlon?
A:Well, when you are looking at a full system, or just the processor itself?
C:Just the processor. Same speed.
A:Well, yeah, same speeds, right now it's hard to say off the top of my head, I don't have the price lists right in front of me, but there would probably be, guesstimating, a $50 difference.
Q:You are saying that the same motherboard will run both chips?
A:Yes. Keep in mind...we have passed around a sample of it, and...in the stocks is a slot. What we are talking about now is socketed parts; we are talking about the size of [a] little sticker...in a socket, just a chip in a socket....
Q:How is the heatsinkability of that product compared to the socket?
A:As long as you are using AMD recommended heatsinks and fans, there is no problem. I have a heatsink for my slot processor; it is an active heatsink, but it has got two fans, and those work really well.
Q:What is the top temperature on a processor, how hot can you run it?
A:For some reason, 90 degrees Celsius comes to mind.
Q:When you say active, are you talking about a thermocouple type of system?
A:No. Meaning it has got a fan attached to the heatsink. The fan and the heatsink are not separate.
Q:There was a recent article in either ZD or PC World that said that the Athlon 800/ 850 ran relatively cooler than the competition.
A:I don't know about that.
Q:What is your power voltage on both of them?
A:They vary, they go from 1.6 to 1.8 volts. As far as recommended motherboards, recommended heatsinks, fans, so on and so forth, that list is growing as we come out with new chips and new technology. So, the best source is always to go to the AMD web site, AMD.com, click on processors, and then choose what processor, whether its Duron, Athlon, K6, whatever, and then find the recommended motherboards, heatsinks, and other solutions and questions and answers online as well. Additionally, for instance the motherboards, the list is not as long as you would expect it to be. It is only because the validation team is swamped with motherboards. I've seen them do their work, I've done the work with them, and they do intensive validation work to get these boards on the Web site. Your best bet is, if you buy any motherboard out there, you can rest assured that motherboard is sitting in Austin getting validated, and most of the issues that come up are BIOS-related. You can rest assured that those problems will be resolved, it is just a matter of time.

Alex ended by pointing out that K6-2 is for Socket 7, which is not incompatible with the Pentium IIIs, but the number of pins is the same. The Duron has 462 pins, which is why it is socket A; you can't just move from K6-2 to Duron; you have to buy a whole new motherboard.

"In some cases it may be a full body scan," said Trapani. "It could be a facial. My wife is a flight attendant with United; in coming back from overseas, they go through a handscan. You have seen the James Bond-type situation where you have the retina or iris scan. We are only concerned with the fingers, in probably the simplest form of biometrics."

Digital Persona was started by the core individuals responsible for Logitech. The company has had a product on the market for about two years called U.are.U Deluxe, designed to work with Windows 95/98.

In April of this year, Digital Persona offered another product called U.are.U Pro. "It works with Windows 95 and 98, NT 4.0, and Windows 2000. It works on a local workstation and it works in a server environment. You can authenticate a user on a local workstation or, in an enterprise environment, you can authenticate on the server. It will run on a Windows NT server; it will run on a Windows 2000 mixed mode; and we hope in about two months to have support for active directories in a native mode."

The virtue of a biometric device is that it allows the elimination of passwords. " We all hate passwords," Trapani noted. Recent studies estimate that it costs a single individual $100 for password management during the year. A Gartner Group study on their own and they says the cost is $340 a year for one person.

"It is a very expensive part of support for a company. It's a lot of overhead, and it's a lot of overhead for the individual user. Not only do you have to remember your password for NT, but also for other applications you work with on your desktop, or for Internet accounts. By using our device, you can virtually eliminate the use of passwords."

The other positive aspect of biometrics is security. "We all know that passwords are passed around, they are stuck underneath the keyboard, they are put on the sticky note on the monitor itself. Everybody knows what everybody else's password is."

Trapani argued, "With biometrics, when a user places a finger on that scanner, you know for certain who in fact that person is. It is a person that has already been enrolled, and has been authorized to gain access using that workstation to that NT environment. It is not going to be anyone else."

"It is the motto of our company to remember to give your computer the finger," Trapani said. "You can use one finger, or you can set it up to use two fingers or three fingers in a sequence. If you were doing a two-finger sign-on you have a one in one billion change of a false acceptance."

Digital Persona's scanner uses a USB connection, and the company wrote its own drivers for the device. The only risk not covered by the device, Trapani said, is what they refer to as a live finger detection. We currently have a patent pending on that particular situation," involving a finger cut off. But he added that those who are extremely security-conscious could use two or three fingers, or require a password in the middle of a sequence, to reduce the risk even further.

Because the device takes a 3D impression of the fingerprint, copies of a fingerprint will not work. "We do not take your fingerprint, we take information from the fingerprint itself. We copy minutia points; from the minutia points a digitized template is created. The fingerprints as we have scanned them are thrown away and the template is encrypted. If someone were to break that encryption and try to re-engineer it, it would not show up as a fingerprint. There is no information there."

The information can not be used to compare to other fingerprint records, however, because the system is not compatible with the AFIS standard used in law enforcement fingerprinting. Using a floppy disk to boot a system, he noted, will not put a user into Windows NT or 2000.

"Now, Windows 95 and 98, we all know are not very secure. So we have created something called Private Space. What we have done is to allow you to carve out a virtual portion of your hard drive, either local or on the server, where you can drag information, you can pull applications down, you can have multiple Private Spaces that are available only through the fingerprint device.

If a person had a blister, or cut a finger, Trapani said, "We take approximately 75 minutia points to create the template. We need approximately 30 to 50 percent of the finger in order to make an authentication. But we also suggest that when you enroll, you should probably enroll two or three fingers from either hand so you can have that versatility. A cut would not make that much difference, but a bandage obviously would. That is why we suggest you actually do two or three fingers on either hand."

Two people can have a common session under NT with the device, Trapani pointed out. "My wife and I use it that way. I've enrolled my right hand, she has enrolled her left, so we enter into the same session anytime we want." You can always offer yourself a password as a backup, he added. You could enroll with a password and a finger, enroll with a finger only, with a password only, or any combination.

The rate of false negatives, which block a legitimate user, "is something akin to about 0.05 percent, requiring you to log on twice with the same finger."

"Let me show you some of the functions we have here. Let's say that I am not the correct person and I put my finger on here. ‘This is Michael's session' [is displayed]. It knows right away that I used a finger that I don't have enrolled.

"If I were to use another finger it says [displays] "Hello Michael," so it knows who I am. I can immediately, by clicking the icon at the bottom of the screen, step away and now I can go to lunch with the most sensitive files open and nobody can get past this. It is keyboard- and mouse-sensitive. If someone steps up to this and puts their finger on it and trys to get in, it won't let them in."

If the system is installed on a server, he said, a user can "move around the domain to other workstations, and get on to my session, as long as the other person [at that workstation] is logged off. If you ran upstairs to do something and all of a sudden you had to get some information, because it's part of the domain, you could put your finger on there and it would recognize who you are and let you back on."

Whatever you set up. Now this has a Microsoft management snap-in console, when you load on the server, its going to go out and find everybody within the domain, it is going to locate the groups, etc. I will show you the administration. Yes, it works entirely with the NT security.

It works entirely within NT security, and thus provides all the administrative tools available to an NT administrator. "I'm going to go under digitalPersona, and administration console, and only since I'm the administrator can I get in this in the first place. Here it shows all the people that I have on the NT environment; here is whether or not they are enrolled. If I want to enroll this person I can double click. I've got an enrollment area which allows me to select to enroll by a finger or enroll by a password, for the individual that asked about that.

"Now I am going to choose the particular finger that I need to enroll with. It is highlighted and it brings up this little frame. I'll put my finger on it; it shows my finger and the little guy jumps up and down. That was a good scan. I'm going to do it again. There is another good scan. We are going to do this four times, and now I have a good scan."

The administrator can set a policy for the user that specifies which fingers and how many are required to enroll.

A log captures when everyone accesses, and logs on and logs off, and illustrates what fingers they used.

With regard to Internet uses, "We have something that we refer to as one touch. Basically this will run with an application on the desktop or it could run with the Internet. When the login procedure is requested, you put your finger on the control key and the authorized finger on the scanner and you get a window. In the window you type the necessary information for the specific field. In this case it is the user ID. It could be the password or the account number."

These accounts could include desktop applications, e-mail accounts, online brokerage accounts, and similar uses. "The next time I open this up I'm going to put my finger on the scanner, this window will pop up. I tell it which one I want to use and it fills in the fields."

In response to an audience question, Trapani said "Does our software allow more than one user to be in a session? No, because NT does not allow for that to happen. Don't forget, we are here to protect NT, we are here to protect your session in NT. So one session per user, or one user per session."

The entire system is priced at $149 per workstation, which includes the sensor and the software, or $1499 for a server-based system.

"We will also have something called U.are.U Online. You can look into it at www.digitalpersona.com. U.are.U Online is going to allow people to authenticate over the Internet or over an intranet, using the device."

The system is a USB only interface, because Digital Persona prefers not to "get involved with the fight over the parallel port. The USB gives us the bandwidth that we want to support the authentication. It is a sub-second type of authentication and USB gives us what we need in that."

Making the system available for Linux users is on the future projects list, but is not currently available. There is also only minimal support currently for Novell and none for Macintosh.

Digital Persona, Inc.
805 Veterans Blvd, Suite 301
Redwood City CA 94063
General (650)261-6079