Join 3,436 readers in helping fund MetaFilter (Hide)


Breakthrough: The World’s First Carbon Nanotube Computer
September 28, 2013 5:58 AM   Subscribe

"It’s only got 178 transistors, but it’s an important proof-of-concept that’s poised to keep Moore’s Law right on track. The breakthrough, in which a basic computer was powered by microscopic chains of carbon atoms, means we may have finally found a viable alternative to silicon chips."
posted by marienbad (27 comments total) 11 users marked this as a favorite

 
I'll believe it when I see it, or when one of them is secretly implanted in my brain while I am sleeping.
posted by Literaryhero at 6:11 AM on September 28, 2013 [2 favorites]


..and judging from that picture, it's powered by a fork?
posted by lumpenprole at 6:13 AM on September 28, 2013 [2 favorites]


You know, I'm fairly certain those are just silicon chips on that fork, and I really don't know why the picture is there. However, I have a variety of narratives that spring to mind. That thing behind the fork chips is a hungry, man-devouring beast of a machine that can only be placated by being constantly fed silicon wafers. If the guy isn't careful he might just lose his hand. Or maybe, less sinisterly, it is a prototype xbox.
posted by Literaryhero at 6:37 AM on September 28, 2013 [1 favorite]


"Breakthrough" is a terrible name for this computer; it'll never sell.
posted by chavenet at 6:37 AM on September 28, 2013 [3 favorites]


Hrmmmmm. The 1st static shock or power line spike is going to do what to that computer?
posted by rough ashlar at 6:39 AM on September 28, 2013


I'll believe it when I see it, or when one of them is secretly implanted in my brain while I am sleeping.

See also computers with magnetic RAM or a thousand other things.

Hrmmmmm. The 1st static shock or power line spike is going to do what to that computer?

Well, this is just a POC and so I'm sure has plenty of problems, but I'm pretty sure people who can invent nanotube chips have heard of static electricity.
posted by middleclasstool at 6:49 AM on September 28, 2013 [11 favorites]


Hrmmmmm. The 1st static shock or power line spike is going to do what to that computer?

I'd be curious to know what happens to it if it's jostled slightly in shipping.
posted by Sys Rq at 6:50 AM on September 28, 2013 [1 favorite]


We've secretly replaced Literaryhero's brain with a carbon nanotube computer. Let's see if he notices.
posted by laconic skeuomorph at 7:06 AM on September 28, 2013 [9 favorites]


No ARM instruction set. Fewer transistors than an iPod. Lame.
posted by sysinfo at 8:11 AM on September 28, 2013 [13 favorites]


178 transistors is enough to embed a MIPS microprocessor?
posted by jepler at 8:21 AM on September 28, 2013


I am somewhat alarmed by the fact that the computer is programmed to literally vaporize bits of itself when they go wonky. Could a hardware maker stick a gradual-suicide routine into a device that renders it automatically obsolescent within x number of months?
posted by Strange Interlude at 8:22 AM on September 28, 2013


It can also run MIPS, a commercial instruction set designed back in the 1980s, allowing it to run over 20 different instructions.

Wait, what? Only 178 transistors, and they implemented a 32-bit RISC instruction set? You'd need 128 transistors just for a single 32-bit register (SRAM uses 4 transistors per bit). And MIPS I had thirty-one 32-bit general purpose registers. So yeah, there's something off about how the article is framing this.

(BTW, the "fork" is a pair of wafer tweezers, used to safely handle silicon wafers.)
posted by jcreigh at 8:24 AM on September 28, 2013 [3 favorites]


That's pretty much the design of cellphones, where the temperature-sensitive LiPo battery is placed right next to heat-generating electronics. Over-temperature conditions slowly degrade the battery, which is often designed not to be replaceable by the end user or for which first-party replacements are obscenely priced. But we didn't come here for an iphone thread…
posted by jepler at 8:25 AM on September 28, 2013


Only 178 transistors, and they implemented a 32-bit RISC instruction set?

I believe this is a 2-bit computer (see tiny, shitty figures here).
posted by Behemoth at 8:40 AM on September 28, 2013 [3 favorites]


I believe this is a 2-bit computer (see tiny, shitty figures here).

Ah, thanks, that seems right. From your link:
The CNT computer is a one-instruction-set computer, implementing the SUBNEG (subtract and branch if negative) instruction, inspired by early work in ref. 23. We implement the SUBNEG instruction because it is Turing complete and thus can be used to re-encode and perform any arbitrary instruction from any instruction-set architecture, albeit at the expense of execution time and memory space
So they just emulate MIPS. Very slowly, I would guess.
posted by jcreigh at 8:50 AM on September 28, 2013 [2 favorites]


static shocks can and do kill traditional semiconductors and most cell libraries have some kind of ESD (electrostatic discharge) diode built into the I/O pad cells to ameliorate this. but static can still kill chips, that's why you use a grounding strap when working on this stuff.

anyway the single-instruction computer is a cool idea and makes a lot of sense for this project. it's akin to the idea that if you implement only a NAND gate out of transistors then you can create any other logical function out of combinations of NAND gates.
posted by joeblough at 9:23 AM on September 28, 2013 [1 favorite]


I am somewhat alarmed by the fact that the computer is programmed to literally vaporize bits of itself when they go wonky.

I understood the article to mean that the vaporization happens as part of the manufacturing process, not during the production lifetime of the processor. Did I misunderstand? Do individual nanotubes turn "metallic" after the processor has been shipped and installed?
posted by escape from the potato planet at 9:35 AM on September 28, 2013


Does this mean those silvery static proof bags will be making a comeback?
posted by maryr at 9:43 AM on September 28, 2013


"where the temperature-sensitive LiPo battery is placed right next to heat-generating electronics"

Sneaky bastards. They should have placed the battery in that cavernously empty other portion of the phone.
posted by schwa at 9:51 AM on September 28, 2013 [6 favorites]


The tiny shitty figures linked above are thumbnails linked to big, not-shitty ones.
posted by flabdablet at 10:46 AM on September 28, 2013 [1 favorite]


Strange Interlude: "Could a hardware maker stick a gradual-suicide routine into a device that renders it automatically obsolescent within x number of months?"

I'm pretty sure this explains the two hard drives in the two laptops I bought on the same day just over two years ago and which failed nearly simultaneously just after the warrant ran out. x = 25 in my case.
posted by chavenet at 11:29 AM on September 28, 2013


Dangit, I was hoping for rod-logic.
posted by butterstick at 11:44 AM on September 28, 2013


To everyone worried about vaporizing chips, sIlicon chips also die if they take too much voltage. There are many layers of power regulating circuitry between the chip and the wall socket. We buy fancy power strips in case it isn't enough. These same techniques will work for carbon chips.

We even use the same process on silicon chips, applying voltage to burn out the malfunctioning parts. When Intel makes chips, they print a bunch of high-end chips, but many have defects. They burn out the path to part of the cache, or to one of the cores, and sell it as a low-end chip.
posted by foobaz at 12:45 PM on September 28, 2013


So the cyberpunks were wrong. Our robot overlords won't be silicon-based after all.
posted by double block and bleed at 2:12 PM on September 28, 2013


Our robot overlords won't be silicon-based after all.

Carbon has better semiconductor characteristics in nearly every way than silicon, but the difficulty is working with it. This trick to deal with the poor quality and mislaid nanotubes is a step forward, but it won't lead to anything on the scale of modern processors. The real money is on doing what we do now with silicon crystals with diamond crystals instead.

That will, of course, completely disrupt the world's jewelry trade and the people involved in the research are probably rightly afraid of the deBeers cartel. Article
posted by localroger at 2:52 PM on September 28, 2013


Owing to substantial fundamental imperfections inherent in CNTs, however...

I think someone should re-think that initialism.
posted by Western Infidels at 3:21 PM on September 29, 2013 [1 favorite]


Wow... half of this thread is people who have never worked in the computer industry (hur, hur, what happens if they drop it in A MUD PUDDLE, huh?), and the rest is people with a close tie to current R&D.

I'll represent the middle ground: I stopped working in the chip-manufacture industry five years ago. And I regularly drop my cellphone in the mud.

THIS IS COOL!
posted by IAmBroom at 8:03 PM on September 29, 2013


« Older Earlier today, the leader of Golden Dawn and three...  |  Looking for a unique wine? The... Newer »


This thread has been archived and is closed to new comments