Friday, March 30, 2007

The Next Intel CPUs are the old AMD CPUs!

Intel introduces its next-generation of processors…

And it is….


Congratulations, Intel.

It is nice to see Intel finally, finally, see the light and commence the development of AMD chips.

The announced chips will be extremely innovative, using technology delivered by AMD back in 2003.

In fact, it is extremely disturbing to see that it by the release of Nehalem in 2008, it would have taken Intel only five years to replicate the efforts of AMD in the server/desktop GPU spaces. Disturbing to holders of INTC, not I.

In a release a couple of days ago, Intel introduced their roadmap for the next two generations of processors, codenamed Penryn and Nehalem, respectively.

Unlike the mainstream media harlots outlets ho's, I came away from the entire announcement totally underwhelmed by the ‘advances’ promised by these two processors.


I had the feeling of déjà vu, you know, like I had heard or seen it all before.

Guess what?

I have heard, and seen it all before! In fact, I already have it! And will have it soon, in Q3 2007.

It is called Opteron, and Athlon, and Torrenza, and Barcelona, and AMD Fusion!

What is Penryn?

Penryn is essentially a die shrink of Core 2 Duo, period.

SSE4? Deep Sleep? Bigger cache? Better virtualization?

All those fancy words do nothing. See, ‘Double-Secret Sleep’ or whatever the low-power state is called is nothing new.

All you have to realize is that it is a die-shrink.

Nehalem, is the next rev of the Core 2 Duo architecture.

It is viewed within Intel “as the first true dynamically scalable microarchitecture”. From Intel, that is. Totally discounting AMD Fusion.

It will also have an integrated memory controller, a la Opteron. Among other things.

No one was able to get Gelsinger to admit to the integration of a GPU with the CPU.

Also mentioned was the fact that there would be versions of Nehalem designed expressly for mobile devices.

How incredibly not new!

All of the ‘innovations’ to be delivered by Penryn and Nehalem are already here.

Furthermore, all the talk about performance gains due to process improvements are a chimera. Looking back at Intel CPU evolution through several process shrinks, most recently the 130nm-90nm-65nm-45nm processes, where have the performance gains come from? Not the process shrinks. It has come from larger caches, higher frequencies, and in the case of Core 2 Duo, a new microarchitecture. The only nugget in the announcement, IMO, about driving the power of the CPU/GPU down to mobile devices, is also made somewhat irrelevant by AMD.

Look no further than the announcement by AMD at the last GDC that they have succeeded in driving the XENOS GPU in the Xbox 360 down to mobile devices. Now, that is an announcement, and a revolutionary one at that. I had expected the technology behind the XENOS to be driven down to the desktop, but the AMD wonks pushed it down waaay further. This should bring gaming-class 3D graphics to mobile devices: another way to annoy your fellow passengers on flights.

The Intel announcement, while informative about their technology roadmap, was an exercise in intellectual masturbation; a lot of effort being spent on meaningless but exciting words, and an empty feeling later, after the euphoria evaporated, of having accomplished nothing significant.

© 2007, John Obeto II for®

No comments: