Intel recently introduced its latest flagship model, which also happens to be the fastest quad-core x86 processor. THG ran the new processor series through its paces, carrying the internal designation G0 stepping, and tested its overclocking potential – with unexpected results.

Depending on how quickly AMD can bring its own quad-core processor to market, Intel may very well release another, faster version of this CPU, running at 3.33 GHz, with the putative name QX6950.

The QX6850, with its four cores running at 3 GHz, is the current high-end model in the x86 arena and has a price tag to match. Its main new feature is the increased front-bus-speed of 333 MHz (FSB1333). The L2 cache remains unchanged at 8MB. It seems as though Intel may be postponing the introduction of its 45nm Penryn processors, in effect a die-shrink of the current Core 2 Duo line, taking the extra time to fine-tune the production process and the processor itself.

AMD, on the other hand, is still working on its first quad-core part, making it increasingly difficult to counter rival Intel’s offerings, where both clock speed and performance are concerned. If nothing else, the market is waiting for a breath of fresh air. However, if recent rumors are to be believed, the first samples of AMD’s upcoming Barcelona design are on par with the performance of Intel’s Core 2 Duo chips. For the time being, though, AMD has only one option to stay competitive with Intel, and that is to slash prices, sacrifice margins and introduce new model numbers.

Intel’s current, relatively comfortable situation may explain why the company has optimized the production process of all of its new dual-core and quad-core FSB1333 processors in order to reach higher frequencies. As we noted in our previous quad-core test, the B3 stepping was unable to reach clock speeds beyond 3 GHz due to its thermal envelope of 130 Watts.

With its FSB1333 processors, Intel is now introducing the new G0 processor stepping, reducing thermal power loss and paving the way for processors beyond the 3 GHz barrier.

Looking a little further into the future, Intel has also announced that it will introduce a refreshed version of its current processor lineup later this year, with models incorporating up to 12MB of L2 cache built on a 45nm process and code-named Penryn. At this year’s IDF, we were already able to witness some early samples in action in a few demo-PCs.

Of course, this brings up the question of why Intel would invest more time, money and effort to further optimize its 65nm processors, which are already outclassing the competition, when Penryn is literally just around the corner. One possible answer is that the transition to the 45nm process may not be going quite as smoothly as anticipated.

At any rate, AMD’s introduction of the Barcelona processor is only days away, and AMD’s Opteron product manager Steve Demski will reveal new technical details about the highly anticipated chip at an upcoming press event.

Judging by AMD’s current share price, a lot of people have high hopes for the upcoming line of processors, which prevents the stock price from slipping any further. Could the same be true for Intel? Is the processor heavyweight possibly worried about AMD’s new quad-core solution and has thus chosen to optimize the current 65nm CPU core for these newly released models?

We were astounded by the high overclocking potential that the new FSB1333 with the G0 stepping. Compared to the previous models in our lab, there was a marked difference. We decided to take a closer look at the actual overclocking potential and conducted extensive benchmarks. We also made detailed observations of how the jump from FSB1066 to FSB1333 affected performance.