Interview: Intel discusses future technology

Eamonn Sinnott, GM of Intel Ireland talks about the 'Ivy Bridge' tri-gate transistors
Author:
Publish date:
3_eamonn s.jpg

In a nutshell, the tri-gate technology will make computing devices more powerful, but also be less of a drain on power – as in batteries and electricity supplies. Would you say this is more of push towards economical power consumption, or a drive for more performance?

Think about it as the choice you’d like to make. If you wanted to make the latest and greatest cloud based server, or a computer that needed a lot of horsepower, or you want to have a handheld device that lasts all day or all week – this processor technology allows that choice during the design phase.

So the structure of the new transistors can be bent to either power saving or higher performance?

The structure allows the product designer that choice. Do you want to cluster them together and arrange them in a way that you can produce a server product, or produce it in such a way that it’s a lot smaller?

Those little microscopic building blocks – and it is hard to imagine 60 million of them standing on an area the size of the dot at the end of this sentence – imagine those microscopic switches. You’re a product designer and you’re saying I’d like something to power a phone or a tablet, or I’d like something that’s really powerful that’s going to drive a server farm. You use those building blocks in different configurations to give you what you’re looking for.

That’s the bit that’s kind of magical for me. It doesn’t really matter to the end user, this innovation, but I think they’d be excited to know that there’s an innovation there that’s going to continue the pace of products that they’ve seen hit the market. I mean, who would have thought a decade ago that we’d have the sort of performance on our smartphones or laptops or tablets that we do today? So just let your mind wander, and you could argue we’re limited only by our imagination, now that we have a core capability that’ll give you the choice of how you want to deploy it.

You mentioned that this has been in development for about ten years?

Yes. Intel first announced a 3D tri gate transistor in 2002, an experimental one.

That’s quite a long league time, and something that’s pretty unheard of in the tech industry. Are the sorts of problems a company like Intel was trying to solve back in 2002 still relevant now? Or are these projects systemic to computing?

It wouldn’t have been introduced as a solution to a problem at the time. If you could reverse your thought process, ten years ago there were guys at Intel talking about solutions to problems that exist today.

The way I think about it, those guys in pathfinding, they’re not thinking about the products that we’re manufacturing today and tomorrow, they’re thinking about products that are going to be manufactured in a decade. And they’re extrapolating Moores Law and they’re saying, ‘well that’s what we have today, what we’re going to want in the future is a level of performance like this. What do we need to invent?’ A great example of this is the tri-gate transistor, which has been part of the industry lexicon for a decade.

So this is just one of a list of long term R&D projects that’s been selected to be made into a tangible product years later?

You can imagine that that’s one of many. It’s almost like a menu – they say ‘for this node these are the things that we need.’ And I would just repeat what Mark T. Bohr (Intel senior fellow and director of the Technology and Manufacturing Group) said, and Mark is an industry titan, he really is, he understands these things better than anyone else. What he said, which I thought was very telling in the presentation, was that we have our 22nm, tri-gate 3D transistor, and we think that will scale well for the next node 14nm. Then he said kind of quietly, ‘and after that we’ll need more invention.’ Now, those guys are working on that today. They won’t necessarily say what it is because that’s a competitive issue.

In terms of that competitive element, you now estimate you have a three year technological lead over competitors, is that right?

We think we have that today, and we think this will extend that lead.

So you think it will take companies like AMD three years to come out with an equivalent tri-gate product?

We think we have a significant lead in our processor technology today compared to anybody else. We have an incredible worldwide scale, and the capability to deploy these products we think is unparalleled. And we also think that this new announcement will extend that lead even further.

And how do you think that will work on a tangible basis in the market place? In terms of your traditional competitor AMD, and latterly ARM, is this going to increase your market share as well as your lead in R&D?

I think the main emphasis on what the technology allows us to do is to maintain what we have, and where people are really happy with the stuff that we make, they’ll want to continue to buy it. But, for the first time, it also gives us a competitive offering in the much smaller, more talked about, ultra-mobile space.

So you’ll be going into further competition with mobile chip players like ARM now?

We would certainly see this as giving us an ability to have competitive products across high-end servers, or a very small form factor device. So we have ambitions to be very competitive in all of those sectors. And to emphasise the point, this technology allows us to prepare those building blocks to give us that choice.
The way it will go is that we’ll introduce it first to the areas where we have the most sales. And then we’ll deploy it in other products later.

There seems to lots of talk about powering super high-powered server farms, and then right the other side to mobile devices like smartphones. Is this relevant to everyday PCs?

The intention is to communicate the entire continuum, its not one or the other.

But the launch presentation did seem to emphasise both those areas…

The intention was not to communicate that that section of the marketplace [standard PCs] was not as relevant, the idea is that Intel wants to compete right across what we call the computing continuum, whether it’s a small device or a high powered server. That’s not meant to exclude the bits in the middle. So to your point, absolutely this processor technology is targeted at all sectors, including the traditional desktop PC.

It’s a bit fuzzy on when this tech is going to arrive, is beginning of 2012 about right?

We’ll be ready for high volume production at the end of this year, then it will flow through the factories and into consumers hands.

Related