by Anna Filatova
09/18/2007 | 02:12 PM
IDF 2007 in San Francisco started this morning. The members of the press have been crowding in front of the keynote hall well over half an hour in advance, trying the get the best spots: the presentation promised to be very interesting: Penryn, Nehalem, the first working 32nm wafer, innovations in ultra mobility and power efficiency as well as breakthroughs in WiMAX technology. Everyone waited.
And then the doors opened, and the stunned journalists realized, there is no one to check their badges and scan the bar code before admitting them in: so everyone literary ran inside, stumbling over cables and carpet edges people ran to jump into the first row seats, to be closer to the ultimate technology.
I guess that’s the beauty of being a pretty woman: I didn’t have to run. I already had a choice of two or three spots reserved for me by fellow journalists :) That was a very pleasing start for the day. But let’s get to the news of the day now.
Pat Gelsinger opened the day by sharing their particular excitement about this Intel Developer Forum. He started off by remembering the exciting news of the past announced at the previous IDF events during the past 10 years: first 500MHz processor, the launch of the Pentium 4, introduction of multi-core architecture, etc. Today is the 10-year anniversary of IDF and everyone at Intel believes that it is going to be the best IDF ever. Pat suggested a new reading for the IDF abbreviation - the Industry Developer Forum stressing how crucial the partner participation is to the success of what Intel does. And he invited everyone to celebrate the innovation and the things they have done together to literary change the world.
So, let’s join the keynote speaker, Paul Otellini, President and Chief Executive Office of Intel Corporation and see how the Intel world is changing these days.
Paul Otellini opened his presentation by revealing the main topic of today’s discussion: he was going to talk about extremes - extremes in products, technology and usage, and how it is Intel’s jobs in the industry to drive the technology from inception into widespread adoption. There are three aspects that need to be taken into account when talking about widespread adoption of new technology:
So, how did Intel get where they are now? There were a number of technologies that made it happen. First, we are talking about the relentless pursuit for Moore’s law in processor micro-architecture. Then at some point Intel realized that connectivity was crucial and started working in this direction. In early days Intel was the leader in semiconductor memory. As things got smaller power efficiency became more important.
Intel didn’t invent everything in that picture, but surely a lot of it, thus sending a message to all of us in the industry. The innovations that Intel is making today are the basis of the computer environment. The first is silicon process technology, the second is micro-architecture and the third the ability to create markets.
We are all familiar with 65nm silicon process technology:
But at some point Intel realized there were problems with it, because there were signal leakages. They started to look for new solutions and eventually offered a breakthrough: a new material-based insulator called Hafnium. They replaced silicon dioxide with this new gate. As a result, they reduced the gate leakage by the factor of 10:
This is what Paul called “the magic of 45nm”.
It is the latest silicon technology Intel is going into production with today. It will make Intel unique in terms of the power, performance, transistor density and the yields in the transistor industry. Thanks to this new technology Intel is making great headway on the next generation now and that why they were ready to demonstrate the world’s first 32nm wafer.
The product is fully functional, it is a 32nm SRAM chip, each die features 1.9 billion transistors.
Intel is constantly monitoring the market segments in order to identify user needed features for the given markets and then incorporate them in a complete platform solution.
The platform approach is very efficient. There is also low cost associated with it. Besides, the platform approach makes job for software developers a lot easier. The software Intel uses and the nature of the architectural compatibility is really available to you as you design new devices. So, you can have high predictability on your development around Intel development.
Now let’s talk about a few new solutions that will become the basis for the new platforms aimed at satisfying the demand for certain features in given market segments.
It is based on the same Core 2 micro-architecture shipping today. It is going to be a large quad-core configuration, featuring 410 million transistor dies. Penryn will launch officially on November 12, 2007. Intel is very excited about customer adoption of this technology. And they are already in production of 45nm and Penryn today getting ready for the November 12 launch. However, they feel that it is not just silicon technology that has to be pushed into the envelope. It is also packaging technology. The CPUs should be about 60% smaller, and as a result it will also drive lower cost. Smaller is better, smaller is cheaper.
But small and inexpensive are not the only advantages from innovating into the packaging technology. They are also focusing on building cleaner products. EarlierIntel announced that their products were 100% lead free. Today they are also announcing that all Intel products will be halogen free by the end of 2008. As a result Intel is going to be even friendlier for the environment.
Intel is going to announce 45nm Nehalem next year. It is a very dynamic design from numerous prospectives. We can change the configuration of core, cache, IO to be able to meet the needs of diverse segments. From the developer standpoint it is also dynamic. You can configure real time needs of the system by turning threads on and off and optimizing the performance for the given task. The largest configuration will be an 8-core die supporting 16 threads.
Nehalem design is already complete. This is the first wafer that was complete about a month ago. Each die has about 731million transistors.
Intel wanted to build the highest performance core that would be used in all types of platforms from laptops all the way to high-performance servers. Nehalem is exactly the core like that. It is designed to be a basis for scalable multi-core systems. Besides, Intel also paid special attention to improving single-threaded performance of Nehalem, so that it could be fit for even more applications. So, it will be available in wide range of systems and configurations.
Nehalem has several new important capabilities: high performance integrated DRAM controller that provides very low latency DRAM and scalable bandwidth for memory intensive applications. It also provides simultaneous multi-threading that is very power efficient.
Nehalem also acquired Quick Path interconnect that allows building multi-socket systems linked together.
Now that we talked about the new microprocessors – Penryn and Nehalem, that will be part of the new systems, let’s talk about the chipset platforms.
Intel talked a lot about platforms in the past.
They introduced in Q3 2006 VPro for desktops, then they launched Centrino Pro for mobile. So far Intel shipped 5M units and this number is going to grow higher thanks to broad OEM support, as all OEMs are now shipping Pro versions.
Intel’s efforts in helping create new market have been very successful lately. And the major three components of success here are scale, capacity and diffusion. Let’s get a little bit more into detail on the three of them.
Scale is important because it allows Intel to ramp very rapidly.
The lower and the steeper the drop, the better.
Capacity factor is also evolving dramatically. There are two factories running full capacity 45nm production process today, and two more in Israel and Mexico will come into operation later in 2008.
So, what’s going to be the next mainstream?
A lot of people on the go want to stay connected as much as possible in real time wherever at. Prior to Centrino notebooks made only 20% of the market. After the successful Centrino launch this number increased to 40%. And today Intel expects notebooks unit volume to cross over the desktops by 2009.
Mobility has become mainstream, no one will argue that. However, mobility still has room to improve. The industry needs new devices, services and products. And since the demand is pushing the development, the future of ultra mobility is closer than we think.
New ultra-small ultra-low-power microprocessors and chipsets that will have the ability to connect to the internet all the time are right around the corner. But solutions like that obviously need the network. WiMAX is the network. So, let’s talk about it a little bit.
A year ago Intel talked about trials in WiMAX moving from fixed WiMAX to mobile. Today Intel announced that Lenovo, Acer, Asus, Panasonic and Toshiba were committed to integrate WiMAX into notebooks with the new platform launch. It means WiMAX is moving mainstream.
On the microprocessor side what is needed is to enable low-power. Was going to introduce an x86 CPU with 0.5W power consumption only in 2010, but now it looks like this objective has been achieved sooner. They promised that Silverthorne will come out in 2008, which is 2 years early.
And as for 2010 there is a new goal now: to drop the idle power by factor of 10, which will happen with the introduction of 32nm process.
In the notebook field Intel has already shipped over 10 million Santa Rosa platforms. In mid 2008 they will launch Montevina with integrated WiFi and WiMAX, HD DVD and Blu-Ray video formats support, Cantiga chipset with higher level of integration and 50% smaller form-factor. Intel is prepared to start shipping this solution in May 2008.
Beyond laptop there are other devices such as WiMAX enabled MID – mobile internet device. Intel presented a life demo of a MID like that transmitting data from a real time base jump in a Zion Utah national park right to the keynote presentation floor. It was a great demo of what’s possible and what will soon be made from possible to probable.
That revolution is starting to happen in our home with evolution of consumer internet. And one of the most illustrative field of showing the evolution of consumer internet is the area of games. Gaming drives much of the internet growth today. And according to Intel, PC is where the gaming is going vs. console.
Rich gaming experience requires great processor power and great graphics.
Today we witnessed a demonstration performed by Charles Wirth, owner of XtremeSystems.org. He presented a platform on quad-core Yorkfiled processor overclocked to 5.56GHz running with phase-change cooler and set three benchmark world records in a little over 2 minutes.
There were no voltage modifications made to the board. The cooling system was a three-stage refrigeration cascade cooler. The thermometer reading showed 162.3F (109C) below zero. There were three refrigeration compressors working that allowed the system to reach these temperatures.
Speaking about the extreme gaming platform Intel also spoke of X38 chipset. It will boast unlocked bus ratios, doubled PCI Express bandwidth and will have incorporated new software tuning utility that allows you to tune directly to the bios.
Beyond processors graphics is increasingly important in the visual computer environment. Integrated graphics from Intel can allow you to play comfortably all popular games and deliver Hollywood quality video playback. Next year Intel will bring 65nm production process into the integrated graphics solutions. And in the second part of that generation, in 2009, Intel promises to move integrated graphics to 45nm production process, which will give the industry more than 6X+ performance improvement compared with the results of last year.
In 2010 microprocessors and graphics will be designed from scratch at 32nm, and performance will be up by factor of 10 compared to the today’s level.
However, that is not enough. New product is already under development and will be demonstrated in 2008. It is called Larrabee.
With this solution Intel will deliver teraflops of performance, and its unique advantage is that it will scale easily for software developers offering easy programming model for the developing community. But it also got one more significant advantage: graphics and it is not dependent on the new software paradigm.
Now let’s shift from gaming to consumer electronics. There is much change coming there too.
Intel’s and industry’s main objective here, from global standpoint, is to build internet capabilities in every consumer electronics device on the planet. Interactivity requirement is going to be a predominant factor, as well as content portability.
In the computer industry, the system development cycles are 6-12 months. In consumer electronics field this interval is much bigger: between 24 and 36 months. AT this time however we see the cycles shrinking down from 24 to 12 months. To meet these growing needs Intel offers a new class of products:
Intel is investing heavily here, and although they refused to demonstrate any actual prototypes today they promised to show some at CES and to deliver this solution codenamed Kenmore in 2008.
One of Intel’s advantages is their ability to provide wide arrays of technologies in parallel. Two weeks ago they introduced the dual-core Xeon. 45nm versions of this processor are going to start shipping in next couple of months.
Also the collaboration between Intel and other partners takes place not only in the technology field, but also in energy efficiency field, which is good for the enterprise and friendly for the environment. They have recently launched their Climate Savers initiative that is driving energy saving not only in servers but also in PCs around the world.
According to the recent research, 1.5% of national electricity use is utilized by data centers, and Intel and partners expect it to double by 2011 because the demand for datacenter services continues to grow. They tried to figure out a way to show how you can use today’s standard technology to make rack servers a lot more efficient from the power consumption prospective. They came out with a basic set of components for the system. For example, they suggested running DC throughout the rack itself, they tweaked memory to eliminate energy drain, etc. As a result, there is no sacrifice in performance and it doesn’t cost any more that the off the shelf equipment. It has 18% efficiency improvement.
The industry finally hit the billion connected computers on the internet. So how do we connect the next billion of people after that? And then how do we connect the third billion? Intel is committed to this. They have currently three programs effective in this area.
The first one is government assisted PC progress. They work with governments, their internet service providers and local PC manufacturers and so far they created 170 bundles. This program is active in 60 countries around the world it will bring 10 million people into Internet this year.
The next program is Intel’s teaching program. It already trained 4 million teachers around the world, so that they could bring technology integrated seamlessly into the classroom for the learning progress. By 2010 they expect to have 10 million teachers trained.
The final initiative is the new category of laptops in cost effective fashion that is obviously demanded for the success of the above mentioned programs. Intel built the classmate PC, which is in OEM production by partners nowadays. It is made for kids, with teacher control software integrated. Intel also joined “one laptop per child” foundation. Besides, the industry is responding to this initiative with their own solutions, such as ASUS EPC: sub-$200 fully functional notebook setting new price points for dedicated machines.
Intel has obviously matured. But they can only capitalize by introducing new products, as growth only counts on new products. Paul Otellini shared his excitement about the new opportunities with us today leaving a stable impression that much of what they do is really good. But he admitted that it doesn’t come for free. It requires persistence, innovation, collaboration, vision. So, good luck, Intel!