You’ve heard plenty about Intel’s shiny new Nehalem microarchitecture over the past year â€“ the modular multi-core design the shared slab of Level 3 cache the integrated memory controller and graphics plus all manner of silicon smarts that result in more processing muscle while generating less heat and drawing less power.
Now that Intel has officially launched the platform Nehalem lines up as the successor to the long-lived Core architecture and will in turn provide the foundation for a new wave of processors. But unless you want to buy a server or an Extreme quad-core desktop you’ll have to wait until this time next year before enjoying the fruits of Intel’s labour.
That’s how long it will take before Nehalem-class processors appear in mainstream desktops and all notebooks. It could even be early 2010 before laptops with the new sixth-gen Centrino platform codenamed Calpella and due to share the Nehalem silicon hit the shelves. And that’s a long time from the launch of the much-vaunted microarchitecture to it becoming a day-to-day reality for PC buyers.
After all Intel’s trumpeted ‘tick-tock’ model is based on a steady annual cadence which sees a new processor architecture introduced one year and a broader-scale platform microarchitecture the following year.
â€œFor the beginning of each silicon generation we have a processor that’s derived from the prior generation and that’s the tickâ€ explains Stephen Smith Vice President and Director for Intel’s Digital Enterprise Group. â€œThen we have a product that’s the larger step what we call the tock which is a new microarchitecture to really bring on a leadership level of performance.â€
The theory is that by overlapping new processors and new architectures in alternate years both the processor and microarchitecture streams each build on and then feed back into the other in a state of constant forward movement.
The first ‘tick’ saw Intel’s processors reduced to a 65nm die with the following ‘tock’ ushering in the 65nm-based Core microarchitecture. This was followed by the Penryn-class processors downsizing to the 45nm with today’s unveiling of the matching 45nm Nehalem architecture.
â€œThe Penryn design and a family of products based on the Penryn are derived from the Core microarchitecture and provide the tick for the 45nm processâ€ Smith says. â€œThen Nehalmem 2008 provides the tock. It’s the large microarchitecture step and the biggest leap in system design since the Pentium Pro about ten year ago.â€
â€œWe’re ready to roll with Nehalemâ€ says Intel VP Kirk Skaugen â€œand here’s a whopping great Nehalem wafer to prove it!â€
But here’s where tick-tock appears to come unstuck. Penryn was â€œlaunchedâ€ on November 13 so it just squeaked into Intel’s 2007 target even though the first processors didn’t land until January 2008.
Today’s kick-off for Nehalem repeats this arriving with less than eight weeks to spare. The first Nehalem-based Core i7 systems are due in December but even those will be limited to high-end processors. At last week’s Intel develop forum in Taipei Intel veeps Anand Chandrasekher and Kirk Skaugen both told APC that mainstream desktop and notebook processors won’t arrive until the fourth quarter of 2009.
That’s potentially a full year since the Nehalem clock hit ‘tock’. In fact according to Intel’s own timetable that flood of Nehalem 45nm systems will arrive just in time to see the introduction of the ‘Westmere’ 32nm processor technology slated for late 2009.
And while processor update appears to ramp faster than the microarchitecture by the time most Westmere 32nm desktop systems roll out through 2010 we’ll be rolling out the red carpet for the matching 32nm Sandy Bridge architecture â€“ Intel’s next next-gen microarchitecture which succeeds Nehalem.
So is Intel’s tick-tick clock running a bit slow? The lines between each swing of the pendulum appear be not only increasingly blurred but overlapping.
â€œTick-tock is a very aggressive model there’s absolutely no question about thatâ€ Kirk Skaugen told APC. â€œBut Nehalem’s been in production since the third quarter and we’ve been very clear in differentiating the volume production of the processor from actual launch dates.â€
So volume production of a processor does not equal the launch and that in turn doesn’t equal broad availability of systems?
â€œThe launch gets determined by holidays and customer financial quarters and overall OEM acceptance in the market and things like thatâ€ Skaugen says. â€œWe chose to put the launch date in November for a bunch of marketing reasons.â€
â€œWhat’s important (for Intel) is hitting the tick-tock rate for getting volume and in some cases when we get to the launch date we’re going to have all sorts of products in the pipeline. Right now we’re on track for Nehalem and we’re also on track for Westmere next year.â€
David Flynn attended IDF Taipei 2008 as a guest of Intel.