Slashdot Mirror
Sun Working to Obsolete Motherboards
perl_camel_jockey writes "Sun is developing a new technology that promises to increase computing power by eliminating the need for physical, soldered chip-to-chip connections on the motherboard. Called 'proximity communications', it portends the ability for chips to talk to one another wirelessly just by being next to each other. Potential applications in computer design abound. Apparently this is part of Sun's Hero program, recipient of a $50 million grant from DARPA's High Productivity Computing Systems program to rejuvenate supercomputing in the US and regain the lead lost to Japan, in particular to NEC's Earth Simulator, ranked as the most powerful supercomputer in the world."
Slashdot Microsystems is working hard on a system to eliminate duplicate postings. They hope to have the system working by early 2008.
...
You'd think that since they ask subscribers to email them if there's any problem with the story, they'd pay attention when we email them...
*sigh*
Buttsex.
I think this has happened before...
Cost of all the chips and stuff that makes up an über-1337 computer - 1000$
Cost of fancy cabinet w/ window to artisticly put all your wireless chips in - 200$
The look on your face as your motehr fires up her old vacuum cleaner, blanketing the area with RF-noise - priceless.
Everything in the world is controlled by a small, evil group to which, unfortunately, no one you know belongs.
To obsolete grammar!
About obsolete motherboards I have my doubts though. The Von Neumann (may be spelled wrong) model has been around for a long time because it has proven to work and it also supplies a lot of companies with revenue. If you only have a single chip, then a lot of companies are going to lose money and they won't like that. In fact there may be a silver lining in this that it will push motherboard manufacturers even further for fear of being wiped out by this type of technology.
Well, unless each chip comes equipped with its own miniaturized nuclear reactor, aren't they still all going to have to have leads running to the powersupply?
I don't mean to be a heckler, but are you really "doing away" with the motherboard or just reducing it to a voltage bus with transmitters and receivers replacing some of the input and output pins?
When things get complex, multiply by the complex conjugate.
I might just be horribly ignorant, so please inform me...security? Even though all it is transferring is system calls and such, I am certain that there are ways to back-engineer what the computer is doing at that time by reading the (very faint, I'm sure) wireless signals. Again, I do not know, so will someone please enlighten me? What exactly is going on, what are the security ramifications?
Honey, I'll warm you something in the microwave!
Noooo...
[Beep] rebooting... grmbl...
Dust & dirt. I would imagine that at such low voltage levels, induced current would require a damn near perfect level of alignment between the chip and the "socket". This is admitted in the article. What they don't admit is that it's going to be nearly impossible to get the damn thing in the socket without letting dust or dirt inbetween the chip and the socket. And a more interesting topic is their consistent mentioning of taking the cache of the chip. That's a nice dream and all, but where the hell are you going to put it then? Hardwired onto the motherboard? That's going to dramatically increase the cost of mobo's (so they are simply shifting who gets to eat the high sticker price on their products). And what if I buy a quad capable mobo, but only put 2 processors on it, I'm effectively wasting 2 sets of cache, rather than simply wasting 2 cpu sockets, and the sockets are a hell of a lot cheaper than the cache. I suppose you could fix this by going back to COAST (cache on a stick, yeah i know you remember that nasty stuff). But that brings in a whole new problem: These days, cache is only fast because it's so close to the cpu. If they move it off the die, it's just going to be put back on in 2 years because we can't access the cache fast enough ever since we moved it off the die. I'm no super computer engineer, but these guys better have an entire family of rabbits they plan on pulling out of their asses or this fucker's gonna flop.
WARNING: Viewing This Sig May Cause Blindness.
Sun should find this project rather easy going - their motherboards ARE already pretty obsolete anyway.
The downside is that kids can no longer tell the difference between their candy bags and your new UBER-1337 computers.
Always going forward, 'cause we can't find reverse.
Why should the moderators read the postings? After all, most are dupes anyway, where they will not learn anything new. :-)
The Tao of math: The numbers you can count are not the real numbers.
Let me visualise this... you need the chips but you don't need them to be on a motherboard... how would a computer look? A plastic bag full of chips? you dig in, take the processor out and throw in a new one to upgrade? groovy
http://www.automatiq.se
It's about increasing chip-to-chip bandwidth by using capacitative coupling instead of (comparatively huge) physical wires. This means the chips would have to be more closely connected, probably slotting together like lego bricks.
This reminds me of an issue with a C=64 I used to have that I had removed the aluminum foil casing from (inside the case). The 6502 processor wirelessly queried the SID chip for reassurance and understanding.
Sincerely,
Ass-embly Language
One of the issues we run into where I work (we make oscillators) is that chips behave differently over their specified temperature operating range. Commercial parts are rated from 0 to 70 Celsius (Industrial : -40/+85, Military : -55/+125). Commercial range is pretty decent for most applications; the average user wouldn't expect their computer innards to be heating up past 150 Farenheit. Anyhow, the simpler chips we make operate differently at different temperatures (the part I'm working on now is rated +- 100 parts per million).
A more complicated way of making these is to make what's called an Oven-Controlled model - you basically create a little oven that responds to the temperature of the chip, keeping it at a certain optimal temperature. These parts are much more stable and accurate; they vary in parts per billion. Dust is a big concern during manufacture; they're pretty sensitive, but once they're sealed, they're more or less set to go.
On a completely seperate note, I have to wonder what kind of issues Sun will be having with crosstalk on their new mobo's.
I don't care if it's in the dictionary or not. "Obsolete" is not a verb, damnit.
The first time I read the headline I thought it meant something like "Sun is working with obsolete motherboards". What would be wrong with "Sun working to make motherboards obsolete"? Whoo, five more characters to read, but it's worth it.
Maybe it's because I'm English, I don't know, but I do know that when I am king I will de-obsolete public flogging for people who debase the language thus.
evil math within Nature's Cubic Creation!
This idea reminds me of the free online novel Prime Intellect which I can strongly recommend.
If you want fast computers developed in the US, buy them from the US market and try to mould your tax incentives so that they simulate, as closely as possible, a net asset tax as described in "A Net Asset Tax Based On The Net Present Value Calculation".
The reason Cray Computer Corporation's gallium arsenide fab went out of business wasn't for lack of funding -- it was for a lack domestic market for the end product, supercomputers, in the wake of the end of the cold war. One could also chalk it up to Cray's fixation on supercomputers since the output of the GaAs fab line could have been altered to serve high speed telecom markets, but if DARPA wants fast supercomputers, there was help available from private capital sources.
Its never a good idea for government to compete with private capital sources in high technology.
Seastead this.
For an example of true 3D chip stacking, see Infineon's SOLID technology. Infineon announced that in 2002. Intel and Sharp have also played around with similar approaches.
The Infineon approach is interesting because it puts a layer of copper between the chips. Getting heat out of the middle of the stack is a major problem with all stacking schemes. Infineon claims to address this, but it's not clear how well. You're probably not going to stack up a pile of 50 watt CPUs with this technology. RAM, maybe. Low-duty-cycle flash memory, no problem. Music players are the obvious application.
Not much seems to have come from that technology since the 2002 announcement. So far, none of these stacking schemes have been useful. They're smaller, but not cheaper.