feature article
Subscribe Now

The Road Ahead

The ITRS Updates The Forecast

Each year the International Technology Roadmap for Semiconductors (ITRS) does an update. In case you haven’t seen it, it’s a comprehensive review of where the industry thinks the industry is going. It covers everything from materials to geometries to packaging.

2007 was a weighty update. It seemed like the whole technology roadmap got a good scrubbing. If it’s the first update you’ve seen, you might get the impression that things change a lot from year to year and that visibility is limited. Which might not be that big a surprise – how many of you can predict where you’ll be in five years? Well the ITRS forecast goes out to 2022. Yeah. Feels particularly ironic now; everyone’s wondering whether they’ll have an income in three months, and yet we’re predicting out 15 years. I guess it’s like forecasting the weather: long-term trends are easier to forecast than tomorrow’s rain.

So, if 2007 was indicative, then 2008 should show another massive reshuffling of the numbers, right? In fact, given the new facts on the ground this year, all bets should be off, right? Well… not so fast… In fact, very little changed at all in the forecast. In most cases, either numbers stayed the same or missing numbers were filled in. The basic geometry forecast remained intact (although they did complain about marketing announcements blurring the definitions of nodes). But in terms of changes, the only one that felt somewhat significant was the fact that VDD will remain higher than forecast last year – drops below 1.0 V have been pushed out by about five years to 2020 for high-performance circuits; the drop to 0.5 V for low-power circuits has been pushed out four years, also to 2020.

Of interest, however, were a couple of additions to the document: one up front in the summary, one at the end in the glossary. (Yes, a glossary. Reading through this is like dashing through a WWI battlefield – arcane terms and TLAs whiz by you thick and fast, and you hope, you pray, that you’ll get to the end of the paragraph without being blown away by some obscure nomer for some obscure technology used by some obscure industry in some obscure corner of the world. Or, worse yet, that you won’t get tripped up by some highly convoluted name for something that turns out to be entirely familiar.)

The first was a section devoted to energy. Clearly the technology sector has taken on a more verdant hue (especially since that’s where a lot of investment dollars are going – yeah, I can be crusty and cynical that way sometimes…). They outlined four different ways in which the work of the ITRS affects energy usage.

The most fundamental energy factor comprises the materials and processing used in electronics. Some obvious examples of this are the reduction of leakage through the use of high-κ dielectrics, Si vs. SOI substrates, and new transistor configurations, such as multi-gate. Jumping up a level brings us to the design realm: this is where the basic materials and processes are molded to create actual working circuits. Power-reduction design techniques have been widely explored and expounded for years, even in these pages, so this is well-traveled ground.

A less-obvious level above design is the application of electronics. This is where ready-made devices can be used to reduce the energy consumption of other things. More intelligent thermostats, more efficient automobiles, that sort of thing. While completely valid, there’s an aspect of this that feels almost defensive – like we’re having to justify a bit too hard just how much we’re doing to save the planet.

The last, and no less important, aspect they address is the use of energy in the manufacture of electronics and, in fact, the creation of unfriendly waste in the process. We’ve come a long way from the wanton creation of Hulk-producing underground plumes in Silicon Valley, but there are still plenty of heinous-sounding substances in use. It’s likely that, until we find a way to make electronics out of water, lemon juice, and olive oil (plus some beach sand, and toss in some oily cold-water fish for Omega-3 fatty acids), there will be opportunities to improve the toxicity of the manufacturing brew.

When it comes to energy usage, theorists will ultimately look to the efficiency of computing – how much work is done for a given amount of energy. What’s scary is the suggestion of a rather unholy “rapprochement” between two disciplines that should never be in the same state, much less the same conference room. You’ve heard me bemoan before the fact that the simple digital world has been invaded by complicated analog considerations, not to mention outright analog circuits. Those of us that think in black and white, one and zero, are being forced to consider such decidedly undigital concepts as real numbers, calculus, convolution.

So imagine the suggestion that, in order to assess the ultimate limits of what can be calculated for a given amount of energy, the science of information might have to consult with <shudder> thermodynamics. How humiliating is that? Here tens of thousands of engineers successfully kept their eyelids raised long enough to check off the box on the list of degree requirements that, for some reason, presumed it was useful to have some thermo in the curriculum. Over and done, right? Not so fast, apparently… It may come back to haunt us again. We can only hope that it falls to a few mutant polymaths to do the requisite work reconciling the two disciplines, and then they’ll just tell us what to do from there.

Meanwhile, the glossary entry attempted to define the euphonious, if imprecise, concepts conveyed in the terms “Moore,” “more Moore,” and “more than Moore.” In other words, technologists are having to expend extra energy unraveling the work done in marketing. Not sure where this fits on the thermodynamic energy scale, but it does feel necessary, since such phrases, once coined, get thrown about with increasing abandon, in an “of course everyone who’s anyone knows what this means” fashion, daring, just daring you to ask the question, “What the hell does that mean?” At which point they can righteously respond with, “Everyone knows what that means! Where have you been hiding??”

So here’s what that means. “Moore”… well, everyone knows what that means. Right? What?? You don’t? Where have you been hiding?? This refers to the doubling of computing power or whatever every year-and-a-half to two years.

“More Moore” refers to continued scaling through three dimensions: “geometric scaling” – the simple reduction of sizes, doing everything else the same; “equivalent scaling” – doing non-geometrical things like changing transistor shape or processing methods to improve performance for a given geometry size; and “design equivalent scaling” – using design techniques to improve performance for a given process and geometry.

“More than Moore” refers to trends that go beyond scaling to add value and functionality. They give as examples the integration of analog onto SoCs and the availability of new design and verification tools and techniques, all of which are completely orthogonal to any scaling developments.

All of this brings to mind a trend I’ve noticed over the years, and one that I’ve finally formulated into what I call Moyer’s Law: “Every technology-related paper or presentation must mention Moore’s Law at least once to achieve legitimacy.” Even if it’s more or less irrelevant to the topic of the presentation. Seriously, go to any conference and see if you can count the number of references to Moore’s Law. In fact, we should make a drinking game out of it. Wait, no, then too many people will want to go to conferences.

Of course, there’s also “more Moyer,” where more than one reference to Moore’s Law is made (whether or not relevant), and then there’s “more than Moyer”, where, in addition to Moore’s Law, some other law is invoked, like, say, Boyle’s Law. Hmmm… Maybe we can stop short of drinking games and just have a scorecard handed out at each presentation for ratings on the Moore, more Moore, more than Moore, Moyer, more Moyer, and more than Moyer scales. All of which would be completely unpronounceable after too many drinks anyway. I’ll get back to you on that.

Link: ITRS 2008 Update

Leave a Reply

featured blogs
Feb 26, 2021
OMG! Three 32-bit processor cores each running at 300 MHz, each with its own floating-point unit (FPU), and each with more memory than you than throw a stick at!...
Feb 26, 2021
In the SPECTRE 20.1 base release, we released Spectre® XDP-HB as part of the new Spectre X-RF simulation technology. Spectre XDP-HB uses a highly distributed multi-machine multi-core simulation... [[ Click on the title to access the full blog on the Cadence Community si...
Feb 24, 2021
mmWave applications are all the rage. Why? Simply put, the 5G tidal wave is coming. Also, ADAS systems use 24 GHz for SRR applications and 77 GHz for LRR applications. Obviously, the world needs mmWave tech! Traditional mmWave technology spans the 30 – 300 GHz frequency...

featured video

Designing your own Processor with ASIP Designer

Sponsored by Synopsys

Designing your own processor is time-consuming and resource intensive, and it used to be limited to a few experts. But Synopsys’ ASIP Designer tool allows you to design your own specialized processor within your deadline and budget. Watch this video to learn more.

Click here for more information

featured paper

Using the DS28E18, The Basics

Sponsored by Maxim Integrated

This application note goes over the basics of using the DS28E18 1-Wire® to I2C/SPI Bridge with Command Sequencer and discusses the steps to get it up and running quickly. It then shows how to use the device with two different devices. The first device is an I2C humidity/temperature sensor and the second one is an SPI temperature sensor device. It concludes with detailed logs of each command.

Click here to download the whitepaper

Featured Chalk Talk

Rail Data Connectivity

Sponsored by Mouser Electronics and TE Connectivity

The rail industry is undergoing a technological revolution right now, and Ethernet connectivity is at the heart of it. But, finding the right interconnect solutions for high-reliability applications such as rail isn’t easy. In this episode of Chalk Talk, Amelia Dalton chats with Egbert Stellinga from TE Connectivity about TE’s portfolio of interconnect solutions for rail and other reliability-critical applications.

Click here for more information about TE Connectivity EN50155 Managed Ethernet Switches