Technology: September 2008 Archives

If it seems like only yesterday that chipmakers were pumping up 45nm processes, it was. But this is now and the talk is about the next generation and the next generation and a half after that. There used to be some breathing space between semiconductor processes: about three years or so. Now that time has shrunk to two years, largely because all the other tricks used to reduce the size of transistors on chips have run out of steam.

So, just months after 45nm processes started to ramp, companies are now touting the 32nm process and its little brother, 28nm. Curiously, a bunch of them picked the same week to do it. How odd. TSMC has decided that 32nm is a process for wusses and that the action is going to be 10 per cent smaller at the 28nm node. The Common Platform gang, who have a conference today in Santa Clara to talk about it, are concentrating mainly on 32nm, but say they will have 28nm ready about the same time as TSMC.

In some respects, you can look at the Common Platform and TSMC moves as signifying a bit of a slip in the TSMC deadline. At first, TSMC was not going to follow Intel and introduce a process that used high-k dielectrics and metal gates instead of the silicon and silicon oxide-based gates that have served the industry faithfully for 40 years. At the International Electron Device Meeting (IEDM) last year, IBM began to talk up high-k, metal-gate processes as well: expecting to introduce it at 32nm, in contrast to Intel's relatively early move at the 45nm node.

Then, earlier this year, TSMC told customers that it would offer both conventional silicon gates and the metal versions in parallel processes at 32nm. This has since mutated into metal gates at 28nm. Or, as history seems to be repeating itself, the process formerly known as 32nm has had a bit of a rebadging. Similar to the situation last year, when a process presented at IEDM as 45nm suddenly morphed into a 40nm offering, TSMC has a paper all about metal gates on - wait for it - 32nm when, apparently, it has no such process in development. The commercial 32nm process, which only has silicon gates, has been shoved into the shadows and emerge meekly later on next year. TSMC wants to shine the spotlight on 28nm.

Is TSMC's 28nm really a 32nm process? Frankly, it's anybody's guess right now. The company will not release details of the two things that would indicate whether it will really have a process that allows for smaller transistors than what IBM and the Common Platform gang plan. Those measurements are the contacted gate pitch - this tells you how far apart you can space transistors, so is a good marker for density - and the minimum SRAM cell size. However, if these numbers were good, I would expect TSMC to be shouting them from the rooftops.

Instead, what TSMC has done is shown how nervous it is about Common Platform even though its sales of foundry silicon dwarf those even of the Common Platform companies combined. Chartered Semiconductor Manufacturing is the third-placed foundry, but lies some way behind TSMC and UMC. IBM's foundry sales have slipped since 2005, according to IC Insights. Samsung, although the number-two chipmaker, has tiny foundry sales. Then again, it only got started a couple of years ago but has grown to be a top-ten foundry.

At 32nm, Common Platform has placed a big bet: that metal gates are not only good for high-end processors - this is why Intel jumped early - but for low-power, mobile devices. The deal that the companies tied up with ARM was all about pushing into the nascent market for mobile Internet devices. Buoyed up by bullish projections over growth in this area, the Common Platform people reckon they have a very strong position.

The reason why is that, although metal gates potentially push up cost, you can make some trade-offs between power and speed. You can design a slower circuit and get a big drop in power. TSMC's foil to this is to push the silicon-gate process, claiming that if you run it a bit slower still, you can undercut the power consumption of a metal-gate process. What you do, in effect, is make the gate longer so that it switches more slowly and leaks less juice.

TSMC's claim that the power advantage of a silicon gate is inherent to the process on the basis that the conventional design suffers fewer parasitic effects in a low-power process. Although there is some truth to this, if you are willing to slow the circuit down to the right point, it ignores the fact that you have other options with a metal-gate process. TSMC's argument is somewhat undermined by its use of a graph that contains entirely arbitrary units on both axes that compares what it does to what IBM has claimed. It's not even clear how TSMC computed the values reported for the IBM 32nm process as they have seem to have been derived from a couple of graphs in a VLSI Technology Symposium paper.


Yay, arbitrary units'r'us. Is this operating power, leakage power or some synthesis of both? How is performance measured? You decide. However, notice that the speed of the TSMC silicon-gate process is lower than that of the IBM one.

The problem for TSMC is that companies are now so worried about power consumption - just look at the problems of the iPhone 3G - they are more likely to go for something that gives them low power and decent speed than what they used to pick: low power and low cost. TSMC majors on the latter and says it will have the former soon. IBM is promising potential customers the low power and speed option, plus underlining the idea that there is more than one source for it. With the larger fabless and fab-light chipmakers the phrase "not single-sourced" has a reassuring ring to it. TSMC tries to get around this by operating parallel fab lines: but it is still the same company.

Although TSMC currently dominates 65nm foundry production, there is potential for the situation to swing round dramatically based on which process works for the bigger fabless chipmakers at 32nm. And it provides the best explanation for why the world's biggest foundry was keen to tout a sub-32nm process one day ahead of a seminar run by the competition.

The future is in...November

26 September 2008

Fed up of not being able to getting all your Singularitarian braindumps in one place? Then prepare for the moment we all - well, OK, the people with money - fast-forward into the future at the Convergence08 not-a-conference conference.

I can't find chief Singularitarian Ray Kurzweil on the headliners list at the mid-November unconference but maybe, as a technologist with more than a passing interest in populating virtual worlds observed recently (as reported to me by a regular reader of this blog), the format maybe isn't quite Ray's thing:

"Yeah, Ray downloading himself and becoming immortal. It's not the moral side that bothers me, to be honest - it's the shutting Ray up. Imagine him, talking at us for ever more, never sleeping, never stopping."

Apparently our observer did emphasise "at us".

Too much nitrogen

19 September 2008

One thing that was bugging me from the news today was why anyone would try to adulterate food with melamine - a material that makes hard-to-break plates but which needs to be kept out of food itself. Given that melamine has turned up in a number of contaminated food scandals, particularly pet food, a poison seemed an odd choice as an adulterant.

The news reports explained that melamine makes it look as though there is more protein in the food being tested. But melamine itself is no protein. It's not even close. How come it comes up positive?

The answer lies in the test itself. Because the process has to be carried out so often, you don't go looking for protein, or the amino acids that join up to form it. The test works on the basis that carbohydrates and fats don't have a lot of nitrogen in them: they are generally pretty much all carbon, hydrogen and oxygen. The protein test simply uses nitrogen - there is at least one atom of nitrogen in every amino acid - as a proxy for protein content. And melamine is stuffed full of nitrogen - about five to six times as much per molecule than an amino acid.

Short of lacing the food with cyanide, urea or some ammonium salt, you couldn't get a more evil or effective "protein substitute". It's only when you perform a more sophisticated test that melamine turns up. But, can such a test be administered as cheaply as the flawed nitrogen-content check?

Sooner or later it was going to come to this. Having taken a firm grip on the horizontal, a big company now wants to make sure it controls the vertical.

A few days ago, I got a message from Taiwanese foundry TSMC's European PR agency asking if I wanted to do an interview, under embargo, about an "advanced process announcement".

This could have been a number of things, either some extension to the existing 45nm processes or the advance information on 32nm*. When I got an email from the Common Platform Partner group later that day, the penny dropped. On the 30th, IBM and the gang have a seminar touting the benefits of their upcoming 32nm process. TSMC's embargo would lift the day before. Ah, it's the old "anything you can do..." PR gambit.

OK, I can live with that, I thought. So, I started to put in motion the plans for a story that would, in all likelihood, take in both. It wouldn't necessarily turn out that way, but I might as well plan for it. Given that both camps have said they will have high-k, metal-gate processes on 32nm, doing it as a head-to-head story for print, if not online, would make sense.

So, I start contacting some people, as I can't be in Santa Clara for the actual Common Platform seminar where these companies are planning to talk about 32nm. I hadn't seen any communication from IBM or Chartered at this point, so I don't know if they actually planned to time their own announcement for Monday as well, which could be possible.

However, in doing this, I break an agreement that I didn't actually knew existed until I get a second email, telling me (with my emphasis):

"The release is going to be available just beforehand and TSMC makes a total embargo request to everyone we're approaching, ie the subject matter should not be shared with anyone (including analysts or competitors) before the publication date even as part of any background research."**

When the PR rang to ask about a slot and whether this was OK, I said it was a problem as the subject matter, as far as I understood it, had already been shared. And, frankly, I wasn't about to stop trying to get hold the Common Platform people simply because TSMC didn't want it's shiny announcement tarnished.

Let's put this into perspective. TSMC holds a commanding market share in the foundry business. It has steam-rollered the competition out of the way at 65nm. The 45nm process from the Common Platform people is practically missing in action and number-two foundry UMC is, well, I'm not entirely sure where it is with an actual production-ready 45nm process. It's all gone quiet over there.

TSMC, market-wise is in a position of strength but has, it seems, in one of the cheesiest acts possible in the land of PR, scheduled a spoiler announcement ahead of a similar one from a group of smaller competitors. It's not uncommon, but I don't recall the last time anyone said, in effect: "Don't upset our spoiler by calling anyone else, not even an analyst."

What exactly is TSMC worried about? That an analyst or competitor might explain that doing something in a certain way is a bad idea? And, don't forget, this is going to be at the level of detail you can get from a half-hour phone call - you could end up with less than has already been discussed at conferences such as IEDM and VLSI Technology.

The problem is that, by putting such heavy restrictions on research - next they'll be telling us not to put certain keywords into a Google search - all you can do is parrot the spin. You can't call anyone to check whether a certain technique is viable, because that's verboten.

I should point out that I'm pretty careful about what does get relayed to a competitor while doing research under embargo: I try to keep it all very general but simply use the information I have to construct the questions. Also, checking out things ahead of the embargo lifting can help the company. I have significantly changed my construction of a story scheduled for Monday having spoken to a couple of people. The company that imposed the embargo didn't know certain things were happening and so didn't mention them, but the actions of one or two companies around this new product indicate how they feel about the launch.

But the kind of demand that has come from TSMC leads you to the point where the concept of doing an interview under embargo becomes worthless. You can't research the story in the way you know it needs to be done. So there can only be one answer.

* OK, they may be doing something crazy like skipping 32nm, but the actions of TSMC tend to suggest that the company is more concerned about looking behind the times with its announcement rather than leading the way. If you were going to scale right down to 27nm or 22nm, you would hardly be concerned what IBM and Chartered might say.

** Blacklist, here I come.

Update: TSMC has rethought. Normal embargo terms have been restored, apparently.

Nominally, today is a 50th anniversary for the integrated circuit (IC). But it's a troublesome anniversary, which is why the celebration is a bit muted. I could argue that any one of three or four dates could serve as the true beginning of the IC industry.

On September 12, 1958, Texas Instruments engineer Jack Kilby demonstrated to colleagues a tiny chip that carried more than one transistor. Some six months later, TI filed its first patent on IC manufacturing, making the first move in an acrimonious war over who really invented the IC. Filing in February 1959, TI easily beat the application from Fairchild Semiconductor's Bob Noyce, which was filed in July 1959. Although Noyce's application was published first, TI argued it had the prior art and won, only to lose ten years later.

Here's the problem: Noyce had the basis for the technology that is used today. Kilby used wires to connect the transistors together. Noyce realised he could get layers of silicon deposited on the surface of the chip to do the same job. The reality is that, if Kilby's technique was used today, we would still regard the IBM System/360 as the ultimate computer. Noyce's realisation made possible billion-transistor chips. However, the discovery that would usher in that kind of integration had to wait for almost ten years after the initial inventions.

When I saw iTunes 8 needed about 200MB of free disk space to install, I thought I'd have a go at graphing the software's recent high-protein weight-gain programme. Up until version 6, it seems that Apple's flagship software package was a mere stripling, with downloads checking in at well under 15MB up until the end of 2005. Then it was time for a new look iTunes - I think it was about then that the colour of the icon changed from green to blue.

Since version 6, iTunes has definitely been at the cheezburgers and piling on the pounds. Given the way it now glitches and succumbs to the beachball, it's not clear how much is digital muscle and how much is blubber. Either way, its download size is currently doubling every 1.5 years.


After starting it up, I turned off the grid pretty much straightaway. I'm trying out the Genius function but it's still crunching away after an hour. To be fair, it has 20 000 tracks to work through.

The download size estimates are from Versiontracker, by the way.

I've got a tip for Joe Eschbach, marketing veep at Plastic Logic. If you're so worried about how much paper you're carrying around, try losing the ring binders, particularly if they are half empty like this pair from his part in the performance at Demofall 08 yesterday.


Eschbach joined chief executive Richard Archuleta on stage to show off the latest prototype of the company's plastic ereader, using the pile of paper to show why you might want one. The demonstration showed that the company has some problems with yield on its display and more than a little chutzpah from Archuleta. Not only did he claim that the company plans to launch the first ereader for the business user - iRex was there several years ago - he boasted about the first "commercial scale" plastic electronics plant being fitted out in Dresden right now. I think Innos in Southampton, which Polymer Vision bought last year to make the displays for the Readius, might have something to say about that.

Readers of Robert Scoble's blog will probably be wondering why there are two plastic electronics companies unveiling stuff at the Demofall08 conference this week. Despite visiting their website to have a bitch about it, Scoble somehow thought Plastic Logic was actually called Plastic Electronics. Then he wondered what plastic electronics was all about. Well, the clue's in the name. At least what Scoble thought was the company's name.

Making ebook readers has been Hermann Hauser's dream ever since he departed Acorn some 20 years ago and before he headed into venture capital. The Active Book Company then had to make do with the ropey battery life and bulky electronics of 1990s chips and LCDs. As a lead investor in Plastic Logic, the situation looks a lot better for his dream of a mass-market ebook reader. The eInk-based display is bistable, so there's no need to keep powering it to display an image. Up to a point - until the tiny black and white pellets get jiggled around too much - whatever you program the display to show will stay there. And the company is readying a plant in Dresden for volume production of the thin display modules.

There's only one teensy-weensy problem: Amazon, Sony and a few others got there first. The Amazon Kindle has demonstrated that there is a pent-up demand for e-readers. For its part, Sony will probably demonstrate the limits of that demand by by forcing users to buy and store ebooks in proprietary, non-portable (in an IT sense) formats. I know the Kindle enforces worse limitations right now, but Sony has bags of form in the lock-in business.

No suits. We mean it

4 September 2008

The conference happened a few months back but I spotted this while doing some background research on a story:

Proper FCS attire

Participants in the 2008 IEEE International Frequency Control Symposium, Waikiki, HI, are strongly discouraged from wearing jackets and ties in any and all of the Symposium activities and functions.

Wouldn't dream of it.

Sophos has noticed the UK Government's brilliant idea to let its old domain registrations lapse. It turns out that when the National High-Tech Crime Unit (NHTCU) was disbanded and its responsibilities shifted over to the Serious Organised Crime Unit (SOCA), the civil servants in charge thought they would not need the old domain name. So, the people who hoover up old domain names have swooped in, bought the domain for a song and now looks to be gearing up for a bit of advertising and affiliate-marketing action.

It could be worse: the site could be doing drive-by downloads of Trojan-horse malware. But, as Sophos points out, the appears on a fair few news stories about internet crime. Yahoo counts around 1800, so it's not as many as thought it might be. But the very high-traffic BBC news site has a lot of links to it.

It makes you wonder about the "high-tech" bit in the name of the old operation when it seems nobody thought it might be a good idea to stick a redirect, at least, on the old site and hang on to it. Did they never check the server logs to see if people were still going to it?

Remember, a domain name isn't just for Christmas.