RISC-V CPU With No Silicon: Made from Carbon Nanotubes

DrLobotomy

[H]ardness Supreme
Joined
May 19, 2016
Messages
6,561
The price of sand just dropped. Watch out Saudi Arabia!!

(Yes, I know they use special sand from some obscure place)
 

w1retap

[H]ardForum Junkie
Joined
Jul 17, 2006
Messages
12,059
What would be funny is if these processors evolve and make it to the desktop market, then we revert back to using just a small passive heatsink like on a 486 or first gen Pentium. :D
 

Mode13

Gawd
Joined
Jun 11, 2018
Messages
748
They have been telling us the benefits of Carbon over Silicon for a long time. I look forward to them nailing down the process so something can actually come to market.
Must be about 5 years ago there were lots of posts and discussions about carbon nanotubes, I recall too.. Conclusion was typically that the tech was decades into the future. If we're at a point now where a working processor exists, it could snowball quite rapidly into something marketable.

Real question is can these be cost effective.. It's going to take some serious investment money to move us away from silicon in the short term I'm sure.
 

andrewaggb

Limp Gawd
Joined
Oct 6, 2004
Messages
437
well if it's a working processor it's something. but it sounds very simple. Let's see a dram or sram module (also simple but dense) made with the stuff or something similar. It'll be years away anyways. Silicon has had so many years and billions of advancements taht anything else will be unpolished by comparison.
 

DrLobotomy

[H]ardness Supreme
Joined
May 19, 2016
Messages
6,561
RISC isn't even going to run windows much less install and run crysis. And if you find that wonky RISC compatible version of WIndows I doubt Crysis would run too.
 

w1retap

[H]ardForum Junkie
Joined
Jul 17, 2006
Messages
12,059
RISC isn't even going to run windows much less install and run crysis. And if you find that wonky RISC compatible version of WIndows I doubt Crysis would run too.
The future hasn't been written yet. Several past versions of Windows were RISC compatible and worked well. Also, there's already a working group for x86 on RISC-V. There's always the option of a x86 card that has hardware acceleration translation as well. If you want to get technical, new versions of Windows support RISC as well, in an ARM implementation.
 

Zarathustra[H]

Official Forum Curmudgeon
Joined
Oct 29, 2000
Messages
29,470
RISC isn't even going to run windows much less install and run crysis. And if you find that wonky RISC compatible version of WIndows I doubt Crysis would run too.
That's not the important part.

They are using RISC-V because it is free. Researchers working on making processors out of carbon nanotubes would have a difficult time paying for a proprietary instruction set.

Thus, do the proof of concept on something free, and move on from there.

There is nothing that says that you couldn't apply this technology to make a CPU with an x86, ARM or POWER derived instruction set.

This is really just a proof of concept of "can we use carbon nanotubes to make densely packed circuits of transistors as are used in CPU's. The answer to that question is tentatively yes. It remains to be seen if it can be done cost effectively and densely enough to make a consumer or commercial product out of it though, but this is a positive development.

If they ever make it to market though, actual products are likely AT LEAST a decade away.
 
Last edited:

pendragon1

[H]ardForum Junkie
Joined
Oct 7, 2000
Messages
15,896
Came here to reply, but see you have it handled! I have seen this movie so msny times I have most of it memorized...Not sure what that says about me.
yeah its one of my favorites. saw it in theater and bought on vhs at release. I watch it at least a couple times a year and the soundtrack on cd has been worn out for years so I have mp3s on every device. so we might be in the same boat...
 

Mode13

Gawd
Joined
Jun 11, 2018
Messages
748
yeah its one of my favorites. saw it in theater and bought on vhs at release. I watch it at least a couple times a year and the soundtrack on cd has been worn out for years so I have mp3s on every device. so we might be in the same boat...
I wonder how many times you had to watch it before you noticed the transparent shirt ;)
 

pendragon1

[H]ardForum Junkie
Joined
Oct 7, 2000
Messages
15,896
I wonder how many times you had to watch it before you noticed the transparent shirt ;)
"rooftop pool leak" scene one? first time, on the big-ish screen. if I close my eyes I can picture it ;)
 

Mode13

Gawd
Joined
Jun 11, 2018
Messages
748
"rooftop pool leak" scene one? first time, on the big-ish screen. if I close my eyes I can picture it ;)
Haha no not that scene, you're thinking of the white t-shirt chicks. There's a scene in the movie where Angelina Jolie is sitting down at a desk writing on some papers in a completely see through shirt. Either that was cut out from certain versions or it took me nearly 20 years to notice it when I got a DVD copy. Just found it amusing.

What was thread topic again? lol... sorry guys
 

Ready4Dis

Gawd
Joined
Nov 4, 2015
Messages
738
Must be about 5 years ago there were lots of posts and discussions about carbon nanotubes, I recall too.. Conclusion was typically that the tech was decades into the future. If we're at a point now where a working processor exists, it could snowball quite rapidly into something marketable.

Real question is can these be cost effective.. It's going to take some serious investment money to move us away from silicon in the short term I'm sure.
Works is one thing, competitive... Not even close. If it's anything like the shrink to 10nm we have a few years still :)
 

DeathFromBelow

[H]ardness Supreme
Joined
Jul 15, 2005
Messages
7,316
Works is one thing, competitive... Not even close. If it's anything like the shrink to 10nm we have a few years still :)
I imagine a nanotube-based cache stacked on top of a regular silicon chip could be quite a game-changer even if the nanotube manufacturing process lags behind silicon for some time.
 

Ready4Dis

Gawd
Joined
Nov 4, 2015
Messages
738
I imagine a nanotube-based cache stacked on top of a regular silicon chip could be quite a game-changer even if the nanotube manufacturing process lags behind silicon for some time.
It'll get better, just saying probably not overnight. They were still having issues getting the nanotubes to work consistently. Then they have to be able to package it small enough with enough repeatability, bring the frequencies up and figure out how to do it without raising prices 10x, aka mass production. I imagine we are still 5 years away from viability at least. It's great progress though, and if prices keep going up on node shrinks this may be the direction they go to increase efficiencies.
 

GoodBoy

[H]ard|Gawd
Joined
Nov 29, 2004
Messages
1,714
20 years out. 5 years for the fabs to figure out that smaller silicon transistors gain you nothing (They just get hotter and slower)... they revert back to 12nm for chips needing to be fast (cpu's and gpu's), then decide ok lets invest in this carbon nanotube thing, then 15 years to get it from where it is to useful production process.
 
Joined
Jun 1, 2018
Messages
692
20 years out. 5 years for the fabs to figure out that smaller silicon transistors gain you nothing (They just get hotter and slower)... they revert back to 12nm for chips needing to be fast (cpu's and gpu's), then decide ok lets invest in this carbon nanotube thing, then 15 years to get it from where it is to useful production process.
or just make cpus 2wice as big(they are quite small), and we can start moores law all over again for another 50 years. why not
 

Zarathustra[H]

Official Forum Curmudgeon
Joined
Oct 29, 2000
Messages
29,470
It'll get better, just saying probably not overnight. They were still having issues getting the nanotubes to work consistently. Then they have to be able to package it small enough with enough repeatability, bring the frequencies up and figure out how to do it without raising prices 10x, aka mass production. I imagine we are still 5 years away from viability at least. It's great progress though, and if prices keep going up on node shrinks this may be the direction they go to increase efficiencies.
Agree.


This is a very interesting proof of concept, but people tend to over-emphasize the difficulty of original research and creating a functional prototype, and GREATLY underestimate the difficulty in turning that early prototype into a saleable product.

This is the diffirence between Research and Development. Development easily takes 100x more time, money and effort than the original research that made it possible.

We see this in pharma all the time. Some research paper published by some university for a study conducted with $1M in federal grants gets people really interested, and they hope they can buy it tomorrow, when it actually takes billions of dollars of investment to transform that grant based study result into an actual saleable product.
 

SvenBent

2[H]4U
Joined
Sep 13, 2008
Messages
3,166
RISC isn't even going to run windows much less install and run crysis. And if you find that wonky RISC compatible version of WIndows I doubt Crysis would run too.
Consumer CPU's today are RISC with a translation front from X76 cisc
 

Ready4Dis

Gawd
Joined
Nov 4, 2015
Messages
738
Consumer CPU's today are RISC with a translation front from X76 cisc
This is very correct, and I don't see why an ARM couldn't run crysis besides recompiling and making it compatible.

And Windows 10 on ARM is a thing... https://docs.microsoft.com/en-us/windows/arm/

Again, just requires companies to recompile their code for the new target in most cases (but possibly some other adjustments as well). It currently supports x86 emulation in order to support apps that aren't recompiled for native arm. There is nothing wrong with RISC and as mentioned, your decoder translates all of your opcodes to micro ops (basically takes the CISC and makes it RISC, this is a bit over simplified of course).
 

Uvaman2

2[H]4U
Joined
Jan 4, 2016
Messages
3,143
Consumer CPU's today are RISC with a translation front from X76 cisc
And in a decade or so we will rediscover CISC!! and it will be new all over again! CISC, the future of computing will read headlines in 2035. Large silicon and CISC!
 

SvenBent

2[H]4U
Joined
Sep 13, 2008
Messages
3,166
And in a decade or so we will rediscover CISC!! and it will be new all over again! CISC, the future of computing will read headlines in 2035. Large silicon and CISC!
Intel will continue its skylake 14++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ design but add in a cisc to risc convertion to its risc to cisc conveter that it put in front of its 86x cisc to risc conversation.
 
Top