by Anna Filatova
09/19/2007 | 02:00 PM
Day one of IDF Fall 2007 is over and I guess we can say that the 10th anniversary seems to be really remarkable. Of course, the news of the past day has truly contributed to it, but also a record-breaking attendance: they registered over 5000 people this time and claim they’ve got tremendous response from the industry. Well, judging by the crowd everywhere, I can say that it is probably true.
Today’s morning keynote was devoted to mobile revolution and breaking the barriers on the way to it. Dadi Perlmutter, Senior Vice President and General Manager of the mobility group spoke of the today’s changes in the mobile market and the new objectives on the way to further technological progress in this field.
The mobile revolution is here:
We see continuous increase in the share of notebook platforms in the market, and according to the chart above they will even outgrow the PCs in the nearest future. As defined at the previous IDF, there are four vectors of mobility that include performance, wireless computing, battery life and “cool factor” (small weight and size). Has anything changed since then?
The data on the chart shows that four out of five vectors of mobility remained the same, however, people start worrying more about data protection and antitheft security.
Despite what people tell you, that performance doesn’t matter any more, it is wrong: performance does matter. We may believe that when it comes to mobile devices performance becomes secondary, but we are mistaken. Life and internet become more sophisticated. It is now all about rich media and video playback. We were offered a very illustrative demonstration of how the new 45nm Penryn processor with SSE4 instructions will be used to meet the new demands for higher rich media and video playback performance.
There were two notebook platforms participating in this demo, two identically configured HP systems, only the processors in them were different. One of them was equipped with a 45nm Penryn and another one with the Merom processor.
The demo showed that Penryn was indisputably a way ahead.
This so-called system refresh, when u can get a new processor into an existing platform, is great and gives users a lot of flexibility.
So, performance matters and Intel claims they do care about it. They are improving it year after year. But performance alone is not sufficient. With today’s computer challenges you need graphics and media. Now Intel is pushing hard to use their processor technology, design and architecture leadership to improve the graphics and video performance by the factor of 10 by 2010.
Cantiga chipset and Montevina platform is already 2 years ahead of the plan: the performance improvement by the factor of 10 will be reached by 2008. But this performance will be better received when it comes with sleek thin notebooks, like those several Santa Rosa platforms you can see below:
Among them are 17-inch HP, 15-inch widescreen Asus, a solution from Sony, Lenovo and Dell. Intel refers to the exterior design and looks of the notebooks as a “Cool Factor” and it does matter a lot for the platform’s successful adoption by the consumers.
And here Intel implied not only the size of the actual device but the size of the packages. Santa Rosa platforms were sleek, but Intel tends to go smaller with the upcoming Montevina platforms that are going to be smaller than a penny.
But physical sizes are not the only determination factor. Power matters as well. Next year Intel promises to take the power down to 25W. This will allow building even thinner machines. So, with the new packaging technology and as a result of reducing the power Intel intends to achieve 25% thinner processor packaging and 60% smaller mainboard size:
Here is the first working notebook on Montevina platform with Cantiga chipset, solid state hard drive:
Let’s talk about battery life. There are a lot of things that can make battery life better. Intel claims that their innovation and fruitful partner collaboration will help increase it over time. In the past Intel talked a lot about SpeedStep technology that steps down frequency and voltage. They also introduced power management technology.
This year in Santa Rosa they brought back power management capabilities. In particular they announced that they did a lot of managing with partners from Toshiba and Matsushita Technologies to control the display power. Also, a lot of effort was invested into managing the CPU and graphics power, so that the power could be balanced between these two subsystems depending on the workload at a given moment of time. They claim they have 6 other new technologies that will be brought into Montevina.
They introduced the Deep Power Down State that is taking the idle core of the running processor to 0.
They tried to balance efficiently between voltage reduction and exit/enter latencies. The main constraint in this case was data retention, because they had to make sure that the data is retained when the voltage is lowered. How doe it work? The processor content is saved into on-die memory which is always powered on. And since this operation can be performed very quickly, Intel engineers got the ability to play with the voltage and power as they wished until they found the most optimal ratio.
There is also very tight communication between the mobile chipset and CPU. The chipset protects the CPU from waking up, because wake-ups consume a lot of power. To avoid numerous wake-ups the chipset stores some frequently requested info and retains the knowledge of what is going on. Sometime user interrupt requires the CPU to wake up and perform a few instructions. In case the frequency and power can be balanced so that the CPU reassigns these instructions to the powered core, the net power saving is tremendous.
Here is what it looks like in action:
As you see from the pictures above, in Penryn Intel takes it one step forward: they take the cache power off completely and reduce the core voltage way down.
In Nehalem they promise to go even beyond that thus increasing the battery life and driving the power savings down even more.
90% of notebooks today come equipped with WiFi. All Santa Rosa platforms on display that we have already mentioned above are running streaming videos on 802.11n wireless mode.
WiFi is widespread. There are computers in home, office, cafes, campuses, hotels, convention centers, etc. But it is not sufficient. We need broadband coverage. 802.11 standard provides high data throughput. And we were offered a very vivid demonstration of the advantages of WiMAX and ultimate personalization of wireless internet: a guided electric smart cart with embedded Santa Rosa laptop, a scooter with music and video playback showing some virtual tourism, a golf cart enabled with Santa Rosa functionality. Pretty funky means of transportation :)
Cellular technology was a breakthrough in its days, but it was designed and optimized around voice, while WiMAX is designed and optimized around data. It fundamentally changes the communications landscape.
Intel is committed to take WiMAX beyond the goal of 150 million population covered in North America. Intel and KDDI in Japan joined the WiMAX initiative adding to the WiMAX community.
One of the most interesting things about Montevina is that they will have Echo Peak WiMAX solution that will launch in Montevina next year.
Three years ago Intel challenged their designer team to go make a significant reduction in power compared with the best platform at the time that was Centrino. They came up with a wonderful small technology that can run way below 1W. It was designed with MID (mobile internet device) in mind, but it will obviously find application beyond the original thought:
In his keynote presentation Anand Chandrasekher, Senior Vice President of Intel Ultra Mobility Group touched upon a very acute topic. These days the amounts of data transferred back and forth and access at a given moment of time keeps growing. Data will consume bandwidth and the killer application for data is Internet. So, it is extremely important to be able to unleash the internet experience and he was going to talk about what it takes to make that happen.
The picture below shows the growth of internet traffic over time:
And the remarkable thing is that according to recent studies, 25% of traffic is contributed to by social networking. That’s a staggering number. There are about a billion subscribers to social networking sites at this time. That’s 1/6 of the world’s population. 154 million people are online daily. 3 billion minutes per day is being spent on social networks.
Yes, it is obviously a fantastic business potential. But Intel claims that it is all being done on PCs: over 95% of this action is happening on PCs. Why so? – Tom Conrad, the CTO of Pandora put it very nicely: “In general the whole mobile space is a mess. The whole mobile development thing feels like 1984”. The handheld mobile experience today is not that great. You don’t get adequate performance from you mobile devices, they are not compatible with certain applications. And even though there are numerous mobile handsets today that claim internet options support do not make up for that drawback. Take a look at the chart below, and you will see that these features are hardly used at all:
So what’s needed to be able to unleash internet?
Let’s see what Intel is offering to do about it.
Their innovation imperatives are simple: without sacrificing performance and compatibility to deliver low thermals, lower power and small form-factor design. Menlow platform based on Silverthorne CPU is implementation of these objectives.
In 2005 Intel promised to lower the power by 10X by 2010. However, as we have already mentioned before they are two years ahead of schedule delivering on the promise of 10X power reduction and compatibility. To prove it we were offered a demonstration. A system was running Silverthorne 2008 architecture. On the screenshot you see the top reference line is the power consumption of Dothan processor, while the bottom line is the 10X smaller target value.
The green line is the power consumption of the Silverthorne processor, as you can see the goal is achieved!
Next they run power virus routine in order to increase the workload, but as we see Silverthorne’s power consumption is still below the red line.
At last they run a standard flash application on both: Dothan (blue) and Silverthorne (green). As you see, the latter is still low, even lower:
So, although Intel has obviously hit the target, their job is not completely done yet.
As I have already mentioned before, it is important to stay always on, always connected. This is why we need broadband connectivity. At the presentation Intel showcased the first device in the world to run WiMAX on Linux (device from Compal):
Bringing products like this into market requires partnerships. 6 months ago Intel announced Mobile Innovation Alliance:
And here are some of these devices life:
Intel claims they will start shipping silicon for these solutions in 2008, the partners are also getting ready for the market. So, we should see these platforms next year. But there is more to come.
Moorestown is coming after:
Intel will take the power even farther down, and the idle power will drop down by factor of 10, which will dramatically improve battery life by 2009/2010.
This is the kind of device Moorestown will go in: thin, slim, all screen, it will unleash the internet and make innovation possible.
Full internet in your pocket is not possible today but is a big transformation and one of Intel’s primary goals for further.
We have talked about the barriers that have already been broken, but there are even more emerging barriers to be taken into account these days. One of them is affordability barrier. Intel talked a lot about digital devices, miniscule amount of students around the globe who do not have PCs at school and about their desire to fix it with solutions like those discussed yesterday. Notebook is a great solution for application fields like that: great battery life, small size, low power, etc.
Another barrier for internet on the go is architectural consistency: if all software developed for PC will run on mobile devices the problem of architectural consistency will be solved, as currently most web-sites are not optimized for mobile environment and hence drive poor user experience.