System Integration Leads to MegaFunction Gadgets

nanotrends writes "The IEEE Spectrum is running a piece entitled 'Moore's Law Meets Its Match', about the system-on-package (SOP) approach to technology. The (SOP) approach combines Integrated Circuits (ICs) with micrometer-scale thin-film versions of discrete components, and it embeds everything in a new type of package so small that eventually handhelds will become anything from multi-to megafunction devices. This integration is actually developing at a rate faster than Moore's law." From the article: "SOP technology represents a radically different approach to systems. It shrinks bulky circuit boards with their many components and makes them nearly disappear. In effect, SOP sets up a new law for system integration. It holds that as the components shrink and the boards all but disappear, the component density will double every year or so, and the number of system functions in an SOP package will increase in the same proportion."

Comments

Integate CPU + DRAM + Flash

[seanadams.com]

This will be a huge boon for semi-custom embedded apps.

Many embedded processors have some typical complement of flash memory and SDRAM which is about the same for every app, and which takes up half of more of the CPU's pin count. The chips can't be combined on one die, because the yields and economies of scale would go down, and they're different processes.

But combine the dies in a small package and you get the best of both world. Less packaging material and lower pin count == lower cost, easier to design in, and more reliable (at some expense in flexibility). It's not a new idea but it's great to see it catching on.

Re:Integate CPU + DRAM + Flash

[sc0p3]

Custom application definately, but the point of the article is that it combines *discrete* components, resistors, capacitors, inductors and other nontransistor widgets that go into circuit design

CPU, DRAM and Flash are all intensively transistor based devices.

However custom applications for radio transmitter chips (as in TFA) is a awesome example. Combine all RF circuitry into a single chip will make my job multitudes easier. I designed implants for animal testing and the cap's and resistors were 90% of the implant(for this co. www.telemetryresearch.com)

Reliability?

[nonother]

The more functions the more apt it is to crash. For a computer or a PDA to do lots and lots of functions that's great, but if they start putting these into phones or GPS units, or even worse - cars then it's just going to start degrading reliability.

Re:lol

[Anonymous Coward]

You know Moore's Law would be a whole lot better if it were based on some scientific principle rather than some guy's musings about the state of the industry 30 or more years ago.

It would also have more credibility if they didn't change it from 1 year to 18 months.

Moore's law is not something written in stone. I wish people would quit using it.

Law of Accelerating Returns

[smackdotcom]

Interesting that this is the sort of thing that Ray Kurzweil predicted. Kurzweil's generalization of Moore's law is the sort of thing that I suspect is largely correct, though how exactly it manifests itself is something that will be interesting to observe. I recommend familiarizing yourself with his thinking if you have any interest in the future of technology and society, and in particular look into buying his latest book. Disclosure note: I have neither affiliation nor even connection to Kurzweil, I simply find his thinking on the issues of future technology to be both fascinating and compelling.

We're going to see ever more rapid acceleration of technology at an increasing rate that will one day leave Moore's Law in the dust, and the impact on society promises to be phenomenal. Just the notion of ever-more-sophisticated portable gadgetry is already altering society in very interesting ways (and yes, some of those alterations are annoying or inconvenient--oblivious cell phones users and so forth). But with the way these gadgets are going, we're going to rapidly outstrip the imaginations of Star Trek writers in terms of the capability and utility of such ubiquitous and powerful devices. I look forward to having the electronic equivalent of a Swiss Army knife (and yes, I'm sure there's going to have to be some clever work done on improving the user interface on such units--but there are inventive types out there working on that sort of thing). It all promises to be very interesting.

Moore's Law only applies to transistors on an IC

[sidfaiwu]

It looks like the other 90% of devices have been largely neglectic up till now. The fact that the size of devices will shrink at a rate faster than Moore's Law might suggest that the rest of a device is just catching up with ICs after a late start.

Re:Yeah

[LoRdTAW]

I agree with you somewhat but it all comes down to what functions are practicle. I used to have a camera phone but the pictures it took were crap compared to my 4MP canon cam. Now I just have a simple phone that I have a few custom ringers I made on it as well as a wall paper, phone book and clock. It has gps but I prefer gps on a larger screen.

My thing is if they got camera/multimedia phones right like:
-High quality camera with at leat 2MP or 1600x1200
-removable flash card that stores the pictures
-external screen that acts like a viewfinder, great for taking pictures of yourself with someone
-simple usb connectivity
-an mp3 player that also works off the flash card that would work on a bluetooth stereo headset that auto pauses the music on a call. I could ride my bike with my phone in my bag listening to some tunes and take a call easily without fumbling with the phone or wires all over.
-maybe a video service BUT it has to be based on open standars and can to connect to any video server even one I might host. Also simple video recording to the flash card that can record longer then 15 seconds.
-good 3 + day standby time.

Hell there might be such a phone now but im not that on top of those things.

Well, it looks like we're finally on track

[ScrewMaster]

for Vernor Vinge's Singularity.

Re:Yeah

[daniel422]

Actually smaller is worse. It's cheaper, yes, becasue it's less silicon area, but the smaller you get, the more issues you have in creating the chip, and coupling issues with high-speed lines in closer proximity, the less reliable it tends to be.
The semiconductor industry is also continuously updating silicon to justify keeping prices the same level or higher. Since silicon cost goes down once its been created, that's the only way to increase ASP on silicon -- add more features in the same space.
Thus we see cell phones that continue to get smaller and smaller, with more features, yet less reliability.
Hopefully the continued integration process we see in cell phone ICs will provide more reliability by having many systems (now integrated) be made in one fab at one time, thus decreasing fab process differences.

The near-term interface may be a touch screen.

[VernonNemitz]

Cell phones don't have to be as small as they are; the hand-set size of ancient rotary-dial phones was that size for a reason.
Well, if that size was used as a grip behind the body of the unit (with various hardware inside it, of course), then the face of the unit could be a fairly decent-sized touch-screen.
It can even be a decently low-power screen, once companies like this one and this one and this one finish their R&D in things like full-color and size-scaling.

I'd also like to mention that There was a buzz-phrase a number of years ago, "wafer scale integration", and I posted my own thoughts about it
here, in Nov 2003.
While they might not be using silicon as the substrate for this modern version of WSI, I have little doubt that something like what I described is what they are doing. Perhaps I should seek a royalty... :)