Don’t bother getting out of bed. That’s the message Intel will be sending out to AMD on 23 July when it rolls out Core 2 Duo, the desktop processor previously known as Conroe. Yup, Core 2 Duo is a killer and will recapture the PC performance crown for Intel in just about any 32-bit desktop application you care to mention. And there’s absolutely nothing AMD can do about it. Right?

Not so fast. Granted, the first comparison tests pitting Intel’s new baby against AMD’s finest makes for extremely ugly reading. Core 2 Duo pummels the latest AM2-socket Athlon 64 FX-62 chip to within 65nm of its life in everything from gaming to media decoding and old-school integer number crunching. But is there hope for the processor industry’s perennial underdog?

If Intel has 32-bit processor performance sewn up with the impressive Core 2 architecture, early indications are that the gap is much smaller in a 64-bit data environment. Add in superior scaling in multi-CPU configurations and AMD’s server and workstation Opteron chips, at least, will be more than competitive enough to keep the cash flowing in. Back on the desktop, however, only new products will do. Does AMD have anything that can truly compete?

AMD’s processor riposte will come in two stages. The first, codenamed Brisbane and also known as revision G of the K8 architecture, is a 65nm die shrink of the existing Athlon 64 X2 dual-core processor. Brisbane is a dead ringer for the recently-released 90nm AM2 CPUs. With little or no architectural advancement, therefore, it’s all about launch dates and clockspeeds – when will it arrive and will it clock high enough to keep Conroe honest?

More intriguing is the K8L processor, the first major revision of AMD’s K8 architecture since the original Sledgehammer cores were unveiled back in September 2003. AMD has begun feeding info-hungry analysts and journalists with the first morsels of K8L’a architectural detail, and the overall flavour is reminiscent of Intel’s Core 2 design. Think shared cache memory, wider instruction handling capability and a focus on more work done per processing cycle rather than higher headline clockspeeds, and you’ll get a good idea of what K8L will offer. But there’s an extra twist that builds on AMD’s traditional strength in CPU I/O and includes the next version of HyperTransport and support for a new generation of off-chip co-processors that could deliver hardware acceleration for tasks as varied as Java, XML and media processing. Tantalising stuff, but when will K8L arrive?

Well, AMD is keeping its cards very close to its chest, but there are two basic theories being traded among the intrigue-addicted trade and tech press cognoscenti. The first, more optimistic take on AMD’s roadmap sees a limited batch of so-called “boutique” Brisbanes, released towards the end of the summer, running at over 3GHz and with a mission to prevent Intel’s Core 2 Duo X6800 Extreme CPU (to be launched at 2.93GHz) from dominating the enthusiast sector without a fight. Brisbane production would then ramp up to mass market volumes towards the end of 2006, with mainstream Athlon 64 X2s reaping the benefit of the new 65nm manufacturing process in terms of clockspeeds and power consumption. Next – in early 2007 and no later than the CeBit trade fair – AMD will unleash the first K8L dual-core chips with quad-core models to follow. That scenario would be a real result for AMD, keeping it within spitting

distance of Intel for six months or so before possibly pulling ahead with the early arrival of K8L.

However, there’s another, altogether more pessimistic possibility. It starts with the first Brisbane chips appearing in January next year at the earliest and ends with dual-core K8Ls arriving no sooner than the fag end of 2007. Given the headroom Intel has with the Core 2 architecture (overclocking escapades with early samples suggest Intel can release Conroes running at well over 3GHz) and its plans to release the quad-core Kentsfield revision of Core 2 in early 2007, this adds up to an unassailable advantage for Intel at least until the end of next year. Kiss goodbye to the brand equity built up over the past two and half years by the Athlon 64.

But which version of events is more likely? Sorry AMD fans, but it’s the latter. Historically, AMD has been at least a year behind Intel in terms of processor technology. Intel rolled out its first 65nm processors at the beginning of 2006, which places AMD’s initial 65nm efforts squarely in early 2007. If AMD really was just months away from releasing 65nm chips, then engineering samples would be leaking into the wild in ever greater numbers. But they’re not, so AMD ain’t.