Friday, August 30th 2019
Researchers Build a CPU Without Silicon Using Carbon Nanotubes
It is no secret that silicon manufacturing is an expensive and difficult process which requires big investment and a lot of effort to get right. Take Intel's 10 nm for example. It was originally planned to launch in 2015, but because of technical difficulties, it got delayed for 2019. That shows how silicon scaling is getting more difficult than ever, while costs are rising exponentially. Development of newer nodes is expected to cost billions of Dollars more, just for the research alone and that is not even including the costs for the setting up a manufacturing facility. In order to prepare for the moment when the development of ever-decreasing size nodes becomes financially and physically unfeasible, researchers are exploring new technologies that could replace and possibly possess even better electrical properties than silicon. One such material (actually a structure made from it) is Carbon Nanotube or CNT in short.
Researchers from MIT, in collaboration with scientists from Analog Devices, have successfully built a CPU based on RISC-V architecture entirely using CNTs. Called RV16X Nano, this CPU is currently only capable of executing a classic "Hello World" program. CNT is a natural semiconductor, however, when manufactured, it is being made as a metallic nanotube. That is due to the fact that metallic nanotubes are easier to integrate into the manufacturing ecosystem. Its has numerous challenges in production because CNTs tend to position themselves randomly in XYZ axes. Researchers from MIT and Analog Devices solved this problem by making large enough surfaces so that enough random tubes are positioned well.
The CPU is based on the RISC-V architecture, particularly it is designed to handle 32-bit wide instructions in a 16-bit wide memory address design. Being that all stages of the CPU pipeline (Instruction Fetch, Decode, Register Read, Execute and Write Back) are 16-bit in width, the CPU is officially declared as 16-bit. It uses 14,000 logic gates, like AND or NOT gates, to become a fully functional design. Given the careful manipulation of nanotubes, researchers managed to pull off 100% yield, meaning that all of the 14,000 gates worked correctly. Here is the waveform of execution of the Hello World program (while it isn't exactly Crysis, given the position of this technology, it is quite an impressive achievement):
The technology isn't perfect, yet. The chip ran at a very low clock speed of only 10 kHz, which means that your average CPU is an order of magnitude faster than this. With all of the flaws, this demonstration is an important achievement for the technology - a proof of concept. It shows that you are able to manufacture a working example of CPU based on something that doesn't require silicon and possibly is even better than it. We just haven't perfected all of the bits and pieces required to get CNTs at the same level of performance we already have.
Sources:
Nature Magazine, ArsTechnica
Researchers from MIT, in collaboration with scientists from Analog Devices, have successfully built a CPU based on RISC-V architecture entirely using CNTs. Called RV16X Nano, this CPU is currently only capable of executing a classic "Hello World" program. CNT is a natural semiconductor, however, when manufactured, it is being made as a metallic nanotube. That is due to the fact that metallic nanotubes are easier to integrate into the manufacturing ecosystem. Its has numerous challenges in production because CNTs tend to position themselves randomly in XYZ axes. Researchers from MIT and Analog Devices solved this problem by making large enough surfaces so that enough random tubes are positioned well.
41 Comments on Researchers Build a CPU Without Silicon Using Carbon Nanotubes
I highly suggest you check out Coreteks's latest video where he speculates on the next 30 years of chip manfufacturing including CARBON NANOTUBES:
If carbon nanotubes doesn't work out then some other material will. Silicon is definitely not the future. It's getting more and more expensive as the process node shrinks. One day it may be possible to continue to shrink but not economically feasible to do so.
"A new kind of computer chip (array of chips on the wafer pictured above) contains thousands of transistors made with carbon nanotubes, rather than silicon. Although the current prototypes can’t compete with silicon chips for size or speed yet, carbon nanotube-based computing promises to usher in a new era of even faster, more energy-efficient electronics."
www.sciencenews.org/article/chip-carbon-nanotubes-not-silicon-marks-computing-milestone
With 10nm they put all eggs in one basket and failed, greed and hubris got in their way by selecting the cheaper way to do 10nm by default.
You could say karma is a b1tch, this one decision gave AMD the future lead in the cpu business and i am so happy Intel screwed over them self's.
A greedy company like Intel should never ever have any kind of monopoly because they will self destruct by prioritising profit over technical advancements.
I am sure someone will develop a new way of doing cpu's without silicon long before Intel.
The next few years sure look quite interesting, though.
Different computers for different tasks.
CNT is just another candidate for future CPUs like we have today: deterministic, binary. We know silicon's potential is ending. We already have better semiconductors that work (like GaN), but they're more expensive, so won't be used until there's no other way.
Quantum computers will be used for different problems. You won't run Windows on them. Ever.Seriously...
en.wikipedia.org/wiki/Kelvin
A single CNT can be a fantastic thermal conductor... along it rotational axis ("through" the tube) and basically an insulator in perpendicular direction.
A randomly aligned CNT material is a worse thermal conductor than copper.
Making big chunks of aligned CNTs is at this moment under research. It may never be easy and cheap enough to be used in consumer CPUs or cooling solutions.
Also, keep in mind the best thermal conductivity is true for CNTs that are electric conductors - hence not usable as silicon replacement.
CNTs that end up in CPUs (if it ever happens) will have different physical properties and lower thermal conductivity.
If you're around 40 or more, your PC CPU is multiple orders faster than another PC CPU you've likely owned in the past.
Anyway, this is all irrelevant.
Even if these carbon CPUs can only do very simple stuff today, they'll catch up at some point. Maybe in 5, 10 or 50 years - we don't know that. But we know their theoretical limits are far beyond what silicon offers.