If you aren't excited about computers these days, you probably never will. Personally, I can't recall a time in which it was more intriguing to follow the events in the micro processor industry than today. The fact that AMD was able to sneak up on Intel and slap the giant left and right has brought us an unprecedented phase of innovation and value: I'd argue that there never has been more incentive to buy a new PC than later this year or in early 2007.
At TG Daily, we typically try to stay in touch with key players in the IT industry. Besides your everyday phone conversations and conversations at trade shows and conferences, we occasionally visit firms to talk about new trends and technologies. A few weeks ago, we visited Intel at its headquarter in Santa Clara, California, and got a chance to interview senior vice president David Perlmutter, who then shared the lead for Intel's Mobility Group with Sean Maloney.
Since then, Perlmutter has received sole responsibility for the business group that has developed the foundation for an architecture which could be considered the savior of Intel's microprocessor business: Core not only restores Intel's competitiveness, but it will also be the source for the lion's share of Intel's micro processor revenues over the next few years.
Perlmutter has not been a very public person so far and I welcomed the opportunity to learn more about his perspective on the microprocessor industry. Join me in a conversation about Core's potential, a sensible number of cores, power consumption trends, future architectures and products as well the importance of graphics. And don't forget to share your opinion on Perlmutter's answers with other readers in our forum.
TG Daily: The Core microarchitecture, most likely Intel's most important product in over a decade, is out of the gate. You have launched the Woodcrest Core as Xeon 5100 about a month ago; the mobile Merom and the desktop Conroe core are being introduced as Core 2 Duo. The question really is how they will stack up to AMD's processors. Somehow I feel you won't like this first question, because you probably have heard it many times: I'll ask it anyway: For the near future, how well will Core perform against AMD?
Perlmutter: I think we will fare very well. Assuming that AMD will come out with a new architecture in the second half of next year, it obviously will be tighter than today. We will be using a new process technology in 2007 that will allow us to keep the lead. Also, in the second half of 2008, we will be introducing a new micro architecture on 45 nm, Nehalem [as replacement of the Merom core family]. Overall, I think Core will do very well.
I actually like that question. I would have hated it two years ago. Today I am much more confident with predicting how well our product will do.
TG Daily: Would you go as far as saying that Core reverses the competitive landscape in the micro processor industry? Does AMD now have
Perlmutter: Yes. I think Core is changing the game. That does not mean that I am underestimating the competition. I believe that they will do everything they can to be better than we are.
TG Daily: In which way is it important for Intel that Merom is superior? Does it matter at this point?
Perlmutter: Well, I am a technologist. The nuances of the architecture and the performance are very important to me. In general, you always need technology leadership in order to be successful in the market place. And that leadership depends on two things: First, are you answering a need of you customers, depending on the market segment, and second, are you better or equal than your competition in this niche? I think these two factors are interrelated.
And, of course, there is also the question whether you are creating growth and a shift in the market place. Good products always create a shift in the market place. Mediocre products miss the growth of the market. For example, the Ipod and Centrino are products that really created a big change.
TG Daily: Several of our readers, some working for large corporations, have told us that they like the direction Intel is taking with Core. But they aren't completely convinced that you will deliver on your promises, for example on power efficiency. Do you think that Intel has lost some trust over the years with previous architectures and do you think that you can regain that trust?
"Yes, you can see some skepticism in the market. To a certain degree, this has to do with the execution at Intel over the past years."
Perlmutter: Yes, you can see some skepticism in the market. To a certain degree, this has to do with the execution at Intel over the past years. And yes, I think we have to regain that trust. Regaining this trust by the way is something you have to do continuously and if it is something that has been used up, it is going to take a little longer. But I am ok with that.
I am looking at how we executed on our mobile technology in the last three years. We worked very much like clock work. I feel very confident about our ability of gaining back that trust.
But if you are coming back in six months with the same feedback, then something is not right and I would be concerned. It would mean that we haven't convinced the market about how great this product really is. We will have to be very precise in what we say and what we do.
TG Daily: AMD claims that its current and upcoming architectures will scale very well, especially in terms of power consumption and performance. They also claim that Core could be in scaling trouble very soon and put you in a 'Netburst trap'. So, from your view, how well will Core scale?
Perlmutter: Core scales and it will be scaling to the level we expect it to. That also applies to the upcoming generations - they all will come with the right scaling factors. But, of course, I would be lying if I said that it scales from here to eternity. In general, I believe that we will be able to do very well against what AMD will be able to do. I want everybody to go from a frequency world to a number-of-cores-world. But especially in the client space, we have to be very careful with overloading the market with a number of cores and see what is useful. I believe '2' is a good number. '4' will be an interesting number for the high-end. Will we see eight cores in the client in the next two years? If someone chooses to do that, engineering-wise that is possible. But I doubt this is something the market needs.
TG Daily: That basically means that you consider dual-cores as the sweetspot for now?
Perlmutter: Yes. You will start seeing four cores on the high-end and they will be going into the mainstream, but it will take a while. I think that it will be two or three years until you are going to see four cores entering the mainstream.
TG Daily: What do you consider as Intel's most significant advantage over the competition at this time?
Perlmutter: The main advantage we always had is process technology. I think we have about a year lead. There are many aspects to designing processors for power efficiency, where I also believe we have an overall advantage: There is a lot of experience in designing mobile products. Also, I believe that we have more elements of the platform that will help us. For example, we have Centrino. We can optimize the CPU, the chipset and other components to optimize the battery life. It's not just a CPU game anymore.
"Looking at Netburst, I wish we would have had a good enough mid-step in the 2003/2004 timeframe - which we really didn't have."
TG Daily: Intel appears to be on a two-year-cycle for the transition to new process technologies and the transition to new micro architectures. In 2007, you will switch to a 45 nm production process. Will there be a new micro architecture in 2008 that replaces Core?
Perlmutter: "Replacing" is a hard word. We have developed a cadence. Every two years, we will have an architecture enhancement. For 2008, there will be a big improvement over Core, an enhanced die. It's a continuum of Core that will be better than Core.
TG Daily: Looking back in time, Netburst was introduced with the Pentium 4 in 2001. This architecture lasted about five years. Would it be correct to say that the life cycle of a certain microprocessor architecture will be shorter in the futurethan it was in the past?
Perlmutter: You have to differentiate between basic classes of micro architectures - which are defined for example by the number of pipe stages and basic structures - and core enhancements. From that perspective, Core will live way more than just two years. Looking at Netburst, I wish we would have had a good enough mid-step in the 2003/2004 timeframe - which we really didn't have.
If you have a new design or a micro architecture improvement, it does not mean that you are completely redesigning the micro architecture. Look at AMD, they carry along a lot of K7 pipe stages, they improve interconnects, their caching, and the number of cores. Our next architecture will not change in its basics, but it will be a re-implementation with a lot of other good stuff.
TG Daily: You are hinting at new features. Any details?
Perlmutter: I can't really go into the details, but there will be something new in the second half of 2008.
TG Daily: Let's keep it general: Where is room for improvement in Core?
Perlmutter: Well, we can improve the power/performance aspects and we can do better on system interconnects. Also, we can improve the basic performance: Integer and floating point performance will improve significantly.
"The reason we took mobile at the center is because we realized that power/performance efficiency is the important factor. We simply see better scaling of performance over time using the mobile core."
TG Daily: With Core, Intel introduces the first microprocessor architecture that is really based on a mobile processor at its heart. Previously, mobile processors were developed as derivates of desktop processors or separate from desktop processors. Are we seeing a new trend here? Will future processors be based on mobile processors that are optimized for desktop and server versions on the side?
Perlmutter: There is really nothing optimized on the side. It's about developing everything in parallel. The reason we took mobile at the center is because we realized that power/performance efficiency is the important factor. We simply see better scaling of performance over time using the mobile core. In general, you have to develop a lot of features in tandem. If you don't, then you don't get all three versions at the same time. Developing for one market and then retrofitting later is very time consuming. With Core 2 Duo, we optimized the core to a wider spread of applications than we did with the Pentium M architecture - which had very much a mobile focus. We learned from that mobile lesson and optimized from the beginning, going from mobile to desktop to server. What we decide to bring out first is based on things like business needs and customer readiness.
TG Daily: A year ago, it appeared that Intel processors were caught in a spiral of ever increasing power consumption. With Core 2 Duo, you cut power consumption in half. Now it looks like that certain pockets of power consumption have been defined, around 25 watts for mobile, around 65 watts for the desktop and 100 watts and higher for high-end desktop systems, even if the Core 2 Extreme is only rated at 75 watts right now. What is the trend on power consumption? Do you see certain pre-defined pockets or is power consumption still moving down - or up again?
Perlmutter: I will be steady for a while and then will go down.
TG Daily: What exactly does that mean? A few years out, you would processors expect to consume less power than today?
TG Daily: What is the key to reach this goal?
Perlmutter: Well, it certainly will get harder to do that. From a market perspective, this is what needs to happen. I believe that power will decrease for a very simple reason: For sure in the notebook and it's also becoming obvious on the desktop because of smaller boxes. In the server you need more density as well. You need to condense more and more computing power in any given space.
TG Daily: Are we talking about a gradual decrease - or huge jumps?
Perlmutter: In notebooks, it's probably going to be dramatic. I think we will be seeing much smaller form factors than today that will require lower power designs.
TG Daily: Let's look at the other end. Graphics appears to become more and more important in the mobile space. Last year, Intel projected that its first dual-core mobile platform Napa would be a "hell of a gaming machine." How important is graphics in the notebook today and in the future?
Perlmutter: I don't think that it is very surprising to you that I say that graphics and video capabilities will be more important over time. If you had asked me five years ago about graphics requirements in notebooks, I would have said you need the basics. That is still very true for corporate users, but as notebooks become more mainstream, we definitely cannot have the mobile lagging behind the desktop. Do we believe that you will need a full graphics card? I don't think so, but I believe people will want very good graphics machines going forward. I think that's going to be an even greater focus for us in the future.
TG Daily: Thank you for the interview.
Complete Core 2 Duo launch coverage:
TG Daily interviews Intel: "Core is changing the game"
Intel is back: Core 2 Duo launches
Up to $16,000: Core 2 Duo computers flood the Net
The long road to Conroe
Tom's Hardware: Core 2 Duo smokes AMD's Athlon 64 X2
Intel to launch Merom, Conroe on Thursday
Four AMD dual-core prices now at or near Intel price/performance curve
Technology Background: Will Intel's Core Architecture Close the Technology Gap? (Tom's Hardware)