Nehalem: Intel gives the industry exactly what's needed

Posted by Rick C. Hodgin

Opinion - I have been very surprised reading the commentary from other websites regarding Core i7. Many of those authors believe Intel fell short with this Tock cycle, essentially stating they did not deliver a commensurate successor as with Core. I honestly have to ask the question, are those reviewers looking at the same chip I am? Do they not recognize what Intel has done with this iteration? I for one am most impressed with Intel's design and goals. In fact, I believe Intel has delivered exactly what the industry's needed - a high performing platform to steer the course between here and the next big thing.


There will come a point and time when the heterogeneous nature of processing will surpass x86's abilities and attractiveness, and most likely by orders of magnitude. I honestly believe Intel realizes this today, but recognizes the industry can’t just jump from here to there in one go. It’s why their focus is on Larrabee and Terascale-like processors, and primarily on x86-design. These hold a single replicated core repeated time and time again for massively parallel computing, albeit with some specialized design considerations. But the truth being revealed by the industry and the many solution forms cropping up, seems to be that homogeneous specialized core designs won't be our final destination.

I believe it's inevitable that fat or wide heterogeneous cores will dominate. They must because we simply can't do what's needed in serial today. And in truth, even if we pushed x86 as far it could go, it still wouldn't be enough. And of course Intel realizes this, and that’s why we have Core i7 exactly as it's been produced.

Intel: The Next Generation

Core i7 brings us to the next generation of x86 architecture. It injects life into the upgrade path and undoubtedly has internal design considerations which, while not necessarily providing greater performance for this iteration, serve as a stepping stone necessary to get from here to there (not just Sandy Bridge or Larrabee, but far beyond).

Nothing AMD has today competes with Intel's highest end offering in any measurable way, at least not in terms of performance. And even Intel's previous offerings are all but left in the dust by comparison. Intel has done this deliberately, releasing Core i7 at this performance level.

Intel delivered a processor which keeps the industry interested and focused on x86 for the time being. And with the constant thought running around in the back of all our minds that barring some kind of unforeseen technological advancement, by the year 2020 the CPU advances we know today will come to a halt. At that point, the industry will be hit by some very real, fundamental physics barriers. These will prevent increased performance and process shrinking, and are the very laws of nature itself.

As such, we already know something more and different is needed. And Intel knows this as well. And yet, they are still delivering increases in performance, albeit at a cost premium.

But this is the thing. Isn't what Intel is doing also exactly what this industry needs? Don't we need an Intel out there charging the high dollar fees so they can innovate and develop and produce the next big thing in the timeframes we desire? Hasn't Intel become the stand-up company in doing this, the one taking the exact steps necessary to ensure the future paths of continued processor development? In truth, aren't we exactly where we need to be to keep things moving forward?

Conclusion

Intel has delivered a very high performing CPU, more so than Core 2 and anything AMD64 (or anything I believe some of the other reviewers recognize). Core i7 is, without question, the best performing x86 CPU on the planet. Hyper-Threading, on-die memory controller, large shared cache, new SIMD instructions, QuickPath interconnect, and initial clock speeds of up to 3.2 GHz which have already been overclocked to 3.8 GHz. According to Intel‘s own documentation, there is also much room for advancement thanks to Hi-K/metal gate technology.

If you consider our near future, Intel is already retiring some 45nm cores. It won't be too long until they retire many more. And at some point after that, our only choice for upgrade will be something Nehalem-based. At or before that point, prices will come down to something closer to commodity levels.

And all the while, those willing to pay the big fees up front (mostly universities, governments, big businesses, gamers and enthusiasts) will have paved the way to keep things moving forward. The rest of us will be getting the benefits of Intel’s innovation. We will just slowly fall in line behind everybody else as things line up in whatever direction the technology leads.

Intel has given this industry exactly what it needed - an injection of life. There's now a clear upgrade path for those desiring something more. And until such time as the next big thing is developed, it is very likely Intel will continue to innovate and release products at the forefront just like these, and just like they've done for the past 35+ years.

Until then, we don't have to buy Nehalem. We can buy $60 processors from AMD and have more than enough computing power. But thanks to Intel and Nehalem, when the time comes for us to upgrade ... we will have options. And I, for one, am very much looking forward to seeing what those options are.