Nehalem and mobile platform: a reminder

Nehalem and the future

You may recall, the Nehalem core is a light evolution of the Penryn core, with the return of multi-threading, various small improvements and the addition of some new SSE instructions. This is on the "uncore" side, a term used by Intel to describe the processor structure, that there will be something new. Roughly, we can say that Nehalem gets its inspiration from the structure of the Athlon 64 and other Phenoms. Thus, the memory controller makes its way into the CPU and connectivity evolves from an antique FSB to a more modern bus, QPI (Quick Path Interconnect). Nehalem also gains in terms of modularity. Thus, the cores, memory controllers, L3 cache, QPI lanes, PCI Express lanes, and integrated graphic core are all modular elements. In its first incarnation, the Bloomfield, Nehalem will have 4 cores equipped with 64 KB of L1 cache and 256 KB of L2 cache which share 8 MB of L3 cache and a 3-way memory controller. In other words, a Core 2 which will no longer leave any structural advantage to AMD’s Phenom.


Nehalem was omnipresent and functional at the expo. Samples ran at a frequency of 3.2 GHz, but real performances could not yet be measured. In terms of the chipset, it will be reduced to the strict minimum given the integration of the memory controller. The northbridge, IOH HEDT Tylersburg, is nothing more than a QPI / PCI Express 2.0 converter which provides one 4x port and two 16x ports that can be divided up into four 8x ports using switches on the motherboard. The southbridge will be the ICH10, which compared to the ICH9 doesn’t change much except for the fact that it is more environmentally friendly as it no longer contains halogen gas.


And the future? Concerning the post Nehalem, we didn’t learn too much more. New architecture, Sandy Bridge, will arrive in 2010 and will add a new instruction set, the AVX, destined to replace SSE. Contrary to the latter, which functions with 128 bit registers, AVX will rely on 256 bit registers and will thus enable processing larger vectors and added flexibility. Intel specified at this IDF that the first implementation of AVX will be done with "full performance" conditions. This means that Sandy Bridge will have 256 bit vectorial calculation units when in the beginning 128 bit SSE was processed in several cycles by 64 bit units.


Finally, Intel indicated that the next generation arriving in roughly 2012 will integrate FMA, Fused Multiply Add, or the MAD which is often a topic when we talk about GPUs. It’s actually the core calculation unit of a GPU which gives the advantage of being able to process two operations at a time. This is a combination that is often useful in graphic computing. Of course, its integration to the heart of CPU processing units is an example of the convergence of the two architectures.

The mobile platform

While Intel obviously referred to its future Montevina mobile platform, or the "first version" of the Centrino 2 which will arrive in the second half of the year and that will mainly add the use of a new chipset with a class G45 graphic core, they also briefly mentioned the Calpbella. This 2009 mobile platform will be based on an entirely new architecture because it will be built around a CPU derived from Nehalem architecture that should normally be a dual core CPU integrating a graphic core, memory controller and PCI Express support.

Besides the gain in performance, we can easily imagine that on the energy level this solution should add significant advantages; however, Intel didn’t mention anything concrete in this area about its platform.

The Larrabee and integrated graphics

Larrabee

The Larrabee has probably been the most intriguing chip for the past few years. What will be its potential? Its use? How successful will it be? At this time, all of these questions remain unanswered.

The Larrabee stems from the tera-scale project which consists of the development of massively multicore processors destined to eventually replace generalized processors in the long run. In the mid-term they are meant to serve as versatile accelerators for tasks that are well suited to parallelization. Graphic rendering is of course one of these tasks.


While there is nothing new and concrete concerning the Larrabee, the battle with Nvidia, and to a lesser extent with AMD is now clear. Pat Gelsinger, an emblematic technical figure at Intel, gets straight to the point. He says loud and clear that 3D rendering as it is known today has come to a point where it’s now too limited and doesn’t have much more room for evolution. This leads us to ray-tracing, a discourse the company has already had for some time now. There is one small evolution, however, as Intel changes the name “ray tracing” to “Photorealistic 3D”. Of course it’s a term that means nothing, Nvidia and AMD have been talking about it for a while, but it's easier to "sell" for marketing.

While this method of rendering evidently has numerous advantages, it remains very resource heavy. Also, despite the uncontestable improvement in rendering speed, we are skeptical about the possibility of seeing it widely used for some time. This is all the more so true that Intel hasn’t shown anything new on the subject. Moreover, the ray traced Quake 4 demo is starting to look a little outdated compared to current games just like the questionable artistic slides used to speak of the visual revolution it entails.

Integrated graphics

More pertinent, Intel wants to affirm loud and clear that its integrated graphic cores aren’t as bad as Nvidia and AMD claim. And it especially wants to counter Nvidia which has just launched an unequivocal offensive against Intel. The CPU maker goes of course a bit too far in saying its products display better video quality than competing graphic cards.

DirectX 10 compatibility since the 965G was always referred to in a vague manner. It has now been made clear. More or less, except the 965G, whose first versions probably suffered from a bug on this level, all of Intel’s recent chips will support DirectX 10 via a driver that is planned in the near future. So the 965GM and G35 will support this API as well as the G45 which is about to arrive. This last will add some improvements though with the same overall architecture. The increase in frequency (to 800 MHz) will produce a gain of 25% in terms of brute power. Other gains will come from small details like a higher geometrical processing speed and transcendental functions (SIN, COS etc.) that are implemented more efficiently.


The most important innovations for the G45 will involve video with native Displayport support and especially complete support of video acceleration as defined by the Bluray standard. This should be something to bring Intel up to the level of the competition. Another help in putting them back on track will be in terms of reliability in games. Intel seems to be trying to work more and more with developers to prepare for the future and also to improve support of current products while at the same time heavily insisting on its market shares.


The roadmap for integrated graphic cores mentions the Havendale, a dual core derived Nehalem which has a graphic core. However, Intel specifies that there will be partial integration to the CPU and we will have to wait for Sandy Bridge architecture in 2010 for total integration. Our interpretation of this information is that the GPU will be a distinct component in the Havendale while in the Sandy Bridge the graphic core will use the CPU’s calculation units.

Inexpensive desktops and laptops

Inexpensive desktops and notebooks

What was probably the most revealing in this edition of the IDF concerning Intel’s future strategy was the subject of entry level machines. These were originally intended for developing countries but quickly imposed themselves in more developed countries.


While Intel didn’t really reveal anything too new and concrete (the Atom was already discussed before), its strategy is now clearer: create a real distinction between the "inexpensive" and the market that has been Intel’s classic domain of activity.

For a manufacturer like Intel, the arrival of machines such as the EeePC initially wasn’t good news because there was potentially less money to be made in a PC costing 300€ than a 1000€ (or more) model. These inexpensive products thus represent a threat to Intel’s business; however, they are indeed here to stay and will evidently become increasingly widespread. Intel plans to take the bull by the horns with the Atom which is a very low cost processor to produce while it is enough for basic use.

Because it was derived from an entirely different and much more limited architecture, Intel makes a clear division in the market with the Atom: there will be low priced “limited” PCs and “classic” models. In this way, the traditional Intel market is protected and the manufacturer is ready to swallow up another new market with enormous potential. The separation is much easier to manage than with currents solutions that are based on the classic entry level. Also, as long as they are selling a product, it might as well be profitable. In Intel’s own words on the subject : "Game changing : Intel's purpose-built solution for basic segments to minimize cost and maximize margin".


A new segment is synonymous with a new name. "Inexpensive" or "More expensive" wouldn't have been practical. On the desktop, the basic PC will be called, “Nettop”, and in its mobile version, “Netbook”. The reference to “net” is obvious because the main function of these machines will be to connect to the Internet. For us, this more concerns secondary posts, for example, to equip the children’s room (gaming power is poor, but for homework and chatting on MSN it’s enough).

These same machines will also allow the arrival of numerous affordable offers in developing countries and represent crucial help in the classroom. Moreover, Intel presented at the IDF the second version of its Classmate PC which is actually very similar to the first one. We will have to wait for the end of 2008 to see the arrival of a real new version based on the Atom processor.

Conclusion

Conclusion

As you may have guessed, for any respected technology journalist, the first subject that interested us was the Nehalem and Larrabee. Intel didn't unveil any new details in this area and probably for several reasons. The main one was that they gave the latest details just before this IDF took place.

However, we also got the impression that the manufacturer hesitated in saying more and probably held itself back. In the current climate, Intel largely dominates with no pressure on the CPU market due to weak competition from AMD. Therefore, there was no reason to hurry and give details too early. Concerning the Larrabee, Intel is probably waiting to know more on the potentials of its chip so as not to have to retract any claims in the case of problems as well as observing what Nvidia does.


For Intel, AMD is under control. Whether this is an underestimation or not, based on the evidence, AMD will probably continue to be noteworthy competitor as much on the GPU as CPU level; however, it will not be a force that will be able to significantly affect the market. In short, AMD does not scare Intel.

On the other hand, Nvidia is a confirmed threat. If this manufacturer looks like David compared to Goliath, the determination of its management probably creates some doubt for Intel. The latter intends to keep Nvidia out of its backyard and do everything to prevent this outsider from having too much to say on the computing market. In the process, Intel is planning on repositioning itself in terms of graphic rendering, a domain which is very profitable because it’s possible to justify almost any supplementary calculation power. This is contrary to the classic CPU market in which it is increasingly harder to show the value and interest of a processor priced at 1000€. Of course, Intel plans on changing this; however, it will not prevent them from positioning themselves in two areas.

Nvidia attacks Intel, Intel attacks Nvidia, and the battle should be interesting to follow. Besides the small offensives and (supposed) low blows which are more and more frequent, we should be able to observe a Nvidia offensive that will gain in ampler in 2008 and then a massive response from Intel in 2009.

Finally, in ending, we should briefly mention a domain that is demanding in calculation power and which will become increasingly important in the future: robotics and artificial intelligence. This is probably a long term strategy in which Intel has an enormous advantage: its R&D capacity is incomparable with other rivals.


On the left, an artificial hand equipped with sensors capable of detecting close objects and therefore avoids crushing a human hand, for example. On the right, a Core 2 inside vacuum cleaner… oops…we mean a well behaved robot.