Dadi Perlmutter, general manager of Intel’s Architecture Group, spent most of his presentation showing off Sandy Bridge and it really looks like it’s going to give AMD – and Nvidia – some severe headaches.
The first Sandy Bridges – the 2nd Generation Intel Core chips – will be produced using the 32 nanometre process in early 2011 and in the previous keynote from Paul Otellini it’s clear that Intel is clearly on track to produce 22 nanometre chips as early as the second half of next year.
It’s also pretty clear that this news is going to give AMD some headaches. The two Fusion chips it is showing off this week look like they’ll be pretty nifty and give Intel a few worries at the netbook/notebook level, but the Llano chip is delayed and furthermore it will be produced at the 32 nanometre level, is based on a new and improved K8 architecture but will leave AMD a clear 18 months to two years behind on the processor front.
Yes, it’s true that AMD has a billion or two floating round in its coffers, but it is optimistic to think that Fusion is going to dislodge Intel from its pole position in the graphics market. After all, it has well defined routes to market, it’s already dominant in the integrated chipset market and in its own language, it likes to execute and has shown over the last few years that it can do so.
There is, I think, always going to be a place for AMD and customers like HP and the rest don’t like to rely on one supplier – who does? But if AMD is in slightly dire straits, surely Nvidia is one step beyond that. It’s been clear for some time that Nvidia needs X86 very badly in order to compete with both AMD and Intel. There were rumours – that we reported in our former incarnation at the INQster – that Transmeta would be the magic pill that could turn round Nvidia’s fortunes. There have also been persistent rumours more recently that Nvidia has a secret project and is developing X86 technology. It forged a partnership with plucky little Via that has so far yielded little because the terms of the Taiwanese company’s licence meant that it was unrealistic for a convenient merger.
So is it possible that Nvidia could turn round the rather depressing situation by forming a closer alliance, or merger with ARM. ARM is allegedly totally disinterested in such a deal. The British company licenses its IP and the pound sterlings just roll in.
We chatted to Fuad Abazovic (Fudo) of fudzilla.com earlier today and asked him what he thought about all of this. He said one of the crazier rumours doing the rounds was that AMD could strike a deal with Nvidia. Nvidia is not saying much.
Intel’s Perlmutter crowed about Sandy Bridge during his keynote speech this morning, possibly with some justification given that his company has pulled a magic rabbit out of the hat.
Perhaps he shouldn’t have crowed quite so much. In Augsburg, in 1990, Intel claimed that by the turn of the millennium many of the functions in motherboards, including graphics and the rest, would be integrated in a single die. Given that, it could be argued that Intel is 10 years late and still not yet quite there. Intel, like other individuals on this planet – let’s not forget a company is an individual so it’s it, and not they – is subject to the slings and arrows of outrageous fortune. Just like AMD and Nvidia, I guess.