Making the APUs pay
APU targets previous-generation Intel prowessI was invited to visit AMD after Intel's Developer Forum (IDF) had run its course. AMD knows that Intel's Sandy Bridge architecture is potent, good enough to keep the mega-company's vast range of partners happy, as it promises double the integrated graphics performance over previous generations, along with a boost in the CPU, yet, for all that, AMD must know that that 2011 is a critical juncture for its fortunes. The year marks the introduction of the Fusion APUs, which bring together a new energy-efficient CPU architecture, Bobcat, and Radeon 5000-series-class integrated graphics.
The first iteration of Fusion GPUs will feature what AMD terms the Zacate core. These use a dual-core Bobcat core - a brand-new architecture, remember - that's married to the most-impressive integrated graphics AMD has ever put in a chip. Crucially for AMD, these low-power chips - split into 18W and 9W TDPs - will enable partners to build $500 thin-and-light laptops that offer acceptable CPU performance and, going by history, a decent multimedia experience. Indeed, AMD has gone deliberately GPU-heavy on Zacate, apportioning practically 50 per cent of the silicon for IGP duties. The approach makes sense, especially as the Internet continues the transition between 2D and 3D - putting a greater onus on a computer's graphics, if you will.
Zacate's good enough; now over to the marketing men
AMD's Zacate has pretty much all the technical goodness to be a good CPU/GPU fit for a $500-plus (£350) laptop, I believe. I'm also adamant that dual-core Bobcat will give Intel's Atom a good whipping - the chip that powers most of the Intel-based netbooks priced at under $500 - and it should be good enough to challenge the more-potent CULV portion of Intel's line-up. Investing a large transistor budget into the GPU aspect of Zacate also makes sense - why waste all the learning emanating from designing discrete GPUs, right?
AMD, then, has no real excuses in 2011 as far as mid-price laptops are concerned. Its new mobile chip Zacate is decent enough on paper and it will be bolstered by the mainstream Ontario part in due course, which promises a heap more performance for both the CPU and GPU: it will be the natural competitor for Intel's mobile Sandy Bridge parts. Nowadays notebook success is built around adequate CPU performance that's backed up by above-average GPU credentials and outstanding battery life. Heck, AMD's claiming potentially near-10-hour usage on laptops outfitted with low-power Zacate parts.
No excuses in 2011
The Austin outfit has a number of ducks in a row. The plan, no doubt, is to increase market-share a few percentage points in 2011. The best method of doing this is to go after a segment of the market where the competitor isn't at its strongest. This happens to be at the same $500-$600 market that AMD's newest mobile chip is designed for. If AMD cannot gain momentum with the technical proficiency that's inherent in the Zacate chip, then it will may well just snatch defeat from the jaws of a minor victory.
AMD's Zacate presages much of the company's work in 2011, so the denouement of this story will be known in less than a year. I must remember that temperance is a good thing, however, because the cynical pragmatist in me cannot help but feel that I'll be taking the same plane journey from IDF 2011 and thinking that AMD missed the market. Will time prove me wrong?