An all-day lecture on Intel’s latest silicon showpiece, or being beaten unconscious with a boxed copy of Windows 98 SE? Strangely similar experiences, I grant you. But I’ll take the Windows-based brutality every time.

What, therefore, to make of the latest instalment of that biannual tech marathon in San Francisco, the Intel Developer Forum? It started just like any other IDF: Intel wheeled out the usual senior technologist suspects to stage centre. A familiar tale of market domination, mindless optimism and technological breakthroughs was woven. Jet lagged journalists snoozed at the back of the auditorium. And I began to lose the will to live.

But then Intel dropped the bomb. In a totally unprecedented move, the world’s largest chipmaker not only allowed journalists to benchmark a preview system powered by a next-generation Conroe processor; it also provided a rig running an AMD Athlon 64 FX-60 CPU overclocked to 2.8GHz and encouraged observers to conduct comparative tests. This from a company that has historically struggled to even admit to the existence of its number one rival. Ever since, I’ve been trying to make sense of what we all saw at IDF. And to get my filthy mitts on a Conroe chip and have my wicked way with it.

The fact that Intel allowed comparative benchmarks to be run on an unreleased processor is pulled eyebrow-muscle material. And the results themselves are the stuff heart attacks are made of – for AMD, at least. Those IDF numbers make the new Conroe chip, along with the remaining members of Intel’s new Core family of processors, look mean enough to pistol-whip Athlon 64s into burger meat. In a range of tests, a 2.66GHz Conroe thumped that trimmed-out 2.8GHz Athlon 64 FX-60 by 15 to 30 per cent across the board.

However, it would be grossly unfair to AMD to treat the comparison as conclusive. After all, both of the test machines were supplied and configured by Intel and all the benchmarks were performed under its supervision. You just know that if there are any holes in Conroe’s performance portfolio, Intel wouldn’t have taken the opportunity to show them off. And by the time Conroe appears in a PC near you (unlikely to be sooner than late August), AMD’s finest will have moved its game on.

But hold that thought for a moment. Because I pulled it off. I’ve had my own private party with a Conroe processor. Thanks to the cloak and dagger conditions under which I sampled the chip, I can’t say exactly how fast it was running, but I can pass on my impressions of how the chip performed. And it’s horrifyingly good.

One of the areas of Conroe’s performance I was most keen to investigate was SSE performance – one of the few remaining tasks for which Intel’s burnt-out Pentium 4 Netburst family of CPUs remains the weapon of choice. It’s also the Pentium M’s weakest link. And that’s a chip designed by the same Israeli team of engineers who knocked up Intel’s new Core architecture (see this month’s ‘Insider’, page 21). Intel’s claims regarding the doubling of streaming SIMD performance are right on the money. Clock for clock and core for core, the new Core architecture is twice as quick as a Pentium 4 processor for SSE number crunching.

If that alone doesn’t put the fear of God into AMD, this will. I have it on good authority that the fastest Extreme Edition Conroe processors will hit core and bus frequencies of 3.33GHz and 1,333MHz respectively at launch (the IDF chip sported a 1,066MHz bus speed). Forget doubts regarding Conroe’s ancient CPU bus technology and lack of integrated memory controller. It’s going to be an absolute monster.

All of which, you might think, leaves AMD looking near-as-dammit dead in the water. Granted, the Athlon 64 family is due to benefit from a new socket and added support for DDR2 memory at speeds of up to 800MHz, and yet another re-spin of AMD’s existing 90nm process technology will release a click or two of additional clockspeed headroom. But if Conroe really does launch at 3.33GHz, even a chintzed-out Athlon 64 X2 DDR2 support running at 3GHz is going to be beaten to pulp. Likewise, if AMD’s new 65nm manufacturing technology, which isn’t expected until early 2007, doesn’t deliver a serious clock speed bump, then you can kiss goodbye to the first half of 2007, too. In that scenario, AMD won’t be back in business until the mythical K8L revision of the successful AMD Hammer architecture shows up. But confident though I am that this is how the next 12 months in CPU land will unfold, for the sake of competition and the health of the industry, here’s hoping I’m wrong.