CHIPMAKER Intel's CPU lines still dominate the performance as well as overall product spread in the PC market, from ultrabooks to supercomputers. However, the regular yearly tick-tock rhythm seems to have slowed down somewhat over the past year or two.
Remember the 2008 roadmap slides - the 2008 Nehalem was to be followed by 2009 Westmere, then 2010 Sandy Bridge and 2011 Ivy Bridge. Of course, Sandy Bridge became a 2011 product, with the flagship Sandy Bridge-E, that is, Core i7 3960X only arriving at the end of this year. The full-functionality version, the eight-core Sandy Bridge-EP or Xeon E5 2600, will be coming out in the market only at the end of first quarter of next year.
Then, the 2011 Ivy Bridge is now likely to become a May 2012 product launch even for the initial mainstream quad-core parts for the LGA1155 socket, with the flagship E/EP parts likely somewhere down the road in early 2013. Let's not even try to guess what impact it will have on the Haswell launch after that.
What is happening? Well yes, the chippery does get more complex with each new generation. The eight-core Sandy Bridge-EP stretches the limits of the 32nm process with its 20MB cache and four memory channels, while having to address a wide range of power and performance points with a single die.
The yet uncharted waters of the 22nm tri-gate process could also pose some challenges for Ivy Bridge, the first CPU to use it. You've also seen the initial teething problems of the Patsburg chipset, including its desktop version, the X79. All that takes some time to resolve.
However, no one at Intel seems to be bashing their heads against the wall over this. Well, why should they? The only x86 competitor in town at the high end, AMD, didn't exactly impress the crowds with its initial Bulldozer launch. The real meat there will have to be in the upcoming Piledriver core update, as well as more software that will be re-compiled to support Bulldozer's instruction extensions and features better, and new sockets with more memory and interconnect bandwidth to prepare for future high-end Fusion APUs as well.
Until then, even Intel's current Westmere cores at the high end, and the mainstream quad-core Sandy Bridge, do their competitive jobs more than well enough.
One thing did bother us a bit, although we have to classify it in the rumour department. Just before the Bulldozer launch, Intel was seemingly quite firm with its plan to have the first Ivy Bridge parts sometime in February. However, after that launch and the Bulldozer performance reviews came out, the updated roadmaps all over the web suddenly showed Ivy Bridge sliding towards May.
Could Intel really think there's no point in rushing that launch when the existing line can still be the king of the hill and provide safe, high ROI for a while longer? As we said, the real reasons could be many, however one can't help but raise the question.