We’ve only just began to experience the power of the 45 nm CPUs and now, Intel plans to shrink the manufacturing process under 10 nanometers. Intel expects that, when the semiconductor industry transitions to 450mm silicon wafers around 2012, the number of companies that run their own fabs will drop into the single digits. These announcements were made by Intel's Pat Gelsinger in a San Francisco preview of celebrations surrounding the chip giant's 40th anniversary this month.

Allow me to remind you that Intel was founded July 18 1968, when Fairchild Semiconductor physicists Robert Noyce and Gordon Moore struck out on their own to found Integrated Electronics, Intel for short.
How much longer is Intel going to shrink and miniaturize things? I guess nobody can answer this in exact terms, but speaking of Moore’s Law, Gelsinger noted that there was a time when he and his Intel colleagues wondered if they'd ever be able to scale chips below 100 nanometers.

According to Crn.com, Gelsinger described the elemental hoops Intel has had to jump through to achieve each "tick" milestone in the chip maker's relentless pursuit of Moore's Law, pointing out that while each new process adds materials used in novel ways, modern processors are still built on “the ancient silicon scaffolding."

"We are putting more and more of the periodic table onto that silicon scaffolding. Today we use about half of the elements on the periodic table. When Robert Noyce and Moore started, they used six elements," Gelsinger said.

"We replaced the gate with high-K, we put metal on top of it, but it's still, quote, silicon. [The process of getting smaller] keeps moving forward. It may be carbon nanotubes next or it may be spintronics. But we'll keep moving forward."

We all know that Intel Atom , the last CPU for ULCPCs from the Centrino series is a hit, and there were even concerns that Intel won't be able to keep up with the demand, soon after its release. Obviously, AMD couldn't simply sit back and watch Intel get all the cake, so another round in the AMD vs Intel Centrino match started. For a while, only Intel played the ball, but now it seems AMD has a surprise ready...

Before moving to the AMD vs Intel Centrino part, let me give you a hint - Gainward, notorious manufacturer of NVIDIA-based video cards, has just started to release ATI graphics cards. Is this cool, or what?

OK, now...it seems that some AMD partners are already testing a low power & low cost CPU targeting the same market as Intel Atom - the ultra low-cost personal computers one. The pictures available at this time are showing a 3.5-inch mainboard, and the rumor goes like this - the new K8 BGA processors will have a TDP of only 8W, and the chipset for it, the AMD M690E + SB600, about 10W-11W, so it should go head to head with Intel's Atom N-series CPU and Intel 945GC chipset, both having up to 26W.

For now, AMD didn't say anything, so it's pointless to get into additional details. We'll surely do that when the moment of truth comes, and the product is officially revealed.

Nehalem really fed the IT rumor mills in the past months, as fictitious specs circulated all around, a delay was prefigured, and some preliminary tests proved it can be faster than expected. Now Intel comes along once again and announces the final specs and settles the release issue once and for all… or not?

According to Fudzilla, Intel’s higher clocked codename Bloomfield will work at 3.20GHz. It will come with 6.4 GT/s QDI memory speed, but it still integrates only 8M of total cache. It’s still a quad-core CPU (what, did you expect it to somehow turn into an octo-core over night?). Still, it can manage that eight simultaneous threads and officially supports DDR3 1333. This is pretty awkward, since DDR3 memories have already reached 2 GHz standard and can go beyond that in OC mode. Intel claims, however, that any 1333+ memory matched with this CPU will have to be overclocked in order to make it work. Great, so now they fully encourage overcliking methods.

Moreover, preliminary test have shown that the triple channel memory support will require three memory modules for best performance, but this could actually turn to be a good thing, lowering the memory prices even more.

Nehalem has 130W TDP which looks a tad better than that of QX9770. The sad part is that the new Bloomfield will only work with a brand new LGA1366 socket motherboard. Not to mention the integrated memory controller, QDI, and some more stuff you already know. Intel also claims that we will most certainly see Nehalem coming in Q4 2008. Right… for $2,000?

AMD Fusion is not something new, but it has been around only in theory, so far. Fortunately, this doesn't seem to be one of those technologies that remains stuck in the rumor stage, but I can't bet on that until I see it. Anyway, the good part is that, according to the latest official reports, coming straight from AMD, its next generation notebook platform, codenamed "Shrike," will be the first Fusion product to be released, and the performance improvements expected may turn the tide...but I wouldn't bet on that, either, since there's plenty of time left until we'll see Shrike in action, and Intel won't sleep in this period, that's one sure thing!

The Shrike will be the first platform to have hybrid processors, combining a general purpose x86 dual-core processor with a new GPU featuring DirectX 10.1 support and UVD 2.0. To go with the flow, the Shrike platform will also have a built-in memory controller with support for dual channel DDR3 memory. The most interesting part is that PCI-Express for discrete graphics is also expected to be part of the platform, and I wonder how small will they manage to make those notebooks, considering the space needed for an additional video card, but since there's enough time to come up with slim cards, I have no worries regarding this.

AMD expects the new platform to boost CPU performance with 20%, and GPU performance with 35%, when compared to the existing Puma platform . Since notebooks have a higher market share in many areas than desktop computers, if AMD Fusion starts well with the Shrike, that would be a great breakthrough for the chipmaker, since currently almost all ultra-portables are powered by Intel platforms, exceptions being a few Fujitsu-Siemens Lifebook models, and various small mobile computers sold only in Japan, using AMD Turion processors.

Since ATI HD4870 is killing the GeForce 280GTX already, I think AMD can have a comeback in the notebook platforms market, too. The Shrike looks great on paper, but there's a long road from theory to practice, and this wait is killing me...

Remember Intel's Nehalem? Oh, wait, they didn't release that one yet! Anyway, there are a lot of things known about it already, like the fact that it may be faster than expected, will have support for triple channel DDR3, and - that's no good news for AMD, but it was expected - sooner or later, Intel's Nehalem will be a direct competitor to the AMD Fusion platform, although these two only have a few things in common. This is not about head to head racing, but about survival(AMD's, of course), but it seems Nehalem may not arrive in time, and so here's another opportunity for AMD to catch up!


Ehhh...did anyone say "delay?" Here's what Doug Freedman, analyst at American Technology Research, says - "Our checks suggest Intel’s Nehalem server parts will initially launch in single-socket only, which we believe will be viewed by customers and investors as a delayed ramp of the company’s second generation 45nm server architecture due out in late 2H ’08. We believe this may be viewed as the second launch delay in recent weeks and is likely to call into question Intel's solid record of on-time execution."

I don't think that many users can't get enough with the latest Intel processors, or even with AMD ones, and so probably hearing that Nehalem may be delayed isn't such a big disappointment, but it's funny to see that, after claiming last year that octa-core Nehalems will be shipping in 2008, Intel's CEO Paul Otellini is now saying they will only be able to produce the Bloomfield quad-core Nehalem this year!

Well, if the Nehalem chips really get delayed, and AMD manages to release the Shanghai 45nm server chip, we may see things slowly getting back on track for AMD, but I wonder why does Intel have to lie to its users? I always think it's better to say "I'll get there, but I may be late" and arrive on time, rather than doing it the other way around, as Intel seems to have done with the Nehalem chip...