IBM claims title of world's fastest graphene transistor
By Donald Melanson 
posted Dec 19th 2008 8:51PM
posted Dec 19th 2008 8:51PM
As we've seen, plenty of researchers and companies are betting on graphene as being the big thing that will revolutionize transistors and, hence, all manner of electronics, and it looks like IBM is now claiming one of the biggest breakthroughs to date, not to mention the desirable title of "world's fastest graphene transistor." More specifically, IBM researchers have apparently been the first to demonstrate the operation of graphene field-effect transistors at gigahertz frequencies and, apparently even more importantly, they've also established the scaling behavior of the graphene transistors, which they say could eventually lead to the development of terahertz graphene transistors -- or, in other word's, keep Moore's Law around for quite a bit longer than many expected.
Shouldnt you guys be at a party?
Letting the rest of the world know whats new with gadgets while in a party? impressive!
I thought Memristor's were the big thing to revolutionize all electronics and give Moore's Law a kick in the ass. They seem far more promising than this, this is just a faster version of something we already have. Not to mention they aren't limited by size in the way that transistors are.
memristors with allow simplification in the design of memory which RAM for instance uses transistors (like D flip flops and such...) to retain data
we will still need transistors to process that data (unless they figure something else out ... not likely)
They are already making memristor configurable signal processors. As well as replacing the configuration bits on current processors to give them a huge boost in speed and making them smaller. I mean why couldn't they could replace the ram, hard drive, processor... everything with memristors? It stores data when power is off (around 1TB per cm I've read, 10x more dense our brain.) It works like a transistor but instead of on and off 1 and 0 or dealing with voltage it's a resistance reading. Different resistances could each correspond to 1 bit. That or they could make an analog computer (like our brain) that does both processing and memory in one. I always thought (hoped) this is where computers were headed.
I really could be wrong, but this is what I've gathered so far. Here's a good article.
http://www.eetimes.com/rss/showArticle.jhtml?articleID=212200673&
I'd just like to know how you calculate the memory density of a human brain...
Everybody knows that the human brain can only hold 160GB, tops. And that's with a data doubler.
A Engadget graphene race would be awesome!
Terahertz CPU running Crysis at 2560x1600 in Very High +AA +AF at 100 fps?
Dreams do come true!
Too bad it will take long enough to see Crysis 4 which will kill even a terahertz CPU....
in virtualbox no less... on a mac ... exciting times
i like how ibm is on the verge of a computer that can use light on a processor/ electrons for the transistors/ and memristors for the memory ... beautiFUL
Well, graphene would be incredibly promising if we had a better method of producing it than taking scotch tape to pencil lead. And besides, nanotubes are much cooler.
Mass production of graphene is possible by placing graphite oxide paper in a solution of pure hydrazine.
http://www.physorg.com/news145544727.html
Its a heck of a lot tougher to get these things to self assemble than cutting masks. But if they can solve that problem on a mass scale, we'll have some serious chips.
I guess that's why they're called Instant Bowel Movement!!
Sounds great and all, but I can't help but feel somewhat skeptical about any new transistor technology that scientists discover in regards to it's practicality. For years I've heard about promising new materials to replace our current silicon transistor technology. I've heard everything from nanotubes, to daimonds, to fucking light. It's the same story each and every time. "Oh this will revolutionize electronics as we know it. It''ll be available in X years." Here we are and there hasn't been anything really that has come into fruition (as in something mass produced.) I'm just hoping that this time this will be the real deal.
But this time its different!
Hafnium.
http://www.designnews.com/article/2930-Intel_Rolls_Out_New_Hafnium_Based_Transistor_Technology.php
You should think of the various claims you read as simultaneously all being the real done and none being the real deal. The job of research is to investigate candidate solutions. Economics drives the ultimate result. To date, it's been more cost effective to push silicon further than to transition to other materials. This won't always be true and there will be such a transition. The same thing is already seen within the silicon community. The structure of a modern transistor and candidates for upcoming scaled-down nodes all use features once considered too expensive and research-only propositions. This includes high-K gate dielectrics, silicon-germanium for introducing strain (not talking about SiGe HBTs here, which are a different animal), SOI, metal gate, FinFET and so on. Pouring money at scaled down bulk CMOS is no longer producing results and we're already past the point of being able to improve the performance/power ratio at the gate level. Now, it's either performance or power, not both, or the use of new architectures such as many slower cores. Hence, a field that is more open to new ideas than ever before.
If you're not an expert in the field, the only way to read this kind of development is - well - you see that bit in the article (just after the horrible apostrophe error) where it said 'keep Moore's Law around for quite a bit longer than many expected'?
Every time anyone has ever predicted that Moore's Law (in its vaguest not-what-Moore-actually-said sense of microprocessors continuing to get faster on some kind of exponential trajectory) would end, they've been wrong.
So pretty much the only thing we can rely on is that processors will keep getting faster at a fairly steady rate. It may be this development that does the trick, it may be some other one, it may be a combination of various factors. But some improvement in technology will enable it. (For example, there were issues with the very high clock speeds of a few generations ago, which were then solved by using multiple cores on the same chip.)
If anything suggests that development will end, don't believe it.
This works in reverse (developments that are too good to be true) and in certain other areas too. Take storage capacity, for instance: around the time Blu-ray was new, remember all those people who regurgitated some company spin release or New Scientist article about holographic storage? And how it was going to come out at the same time so why use Blu-ray? Well, we've had Blu-ray for a few years now and where are these promises? Empty.
If something seems too good to be true (1TB on an 12cm optical disc commercially available for reasonable prices two years ago) it probably is. The good things usually take a while. This might be one of them.
The speed of a single transistor doesn't mean much. They had 200+ (two hundred) GHz Silicon transistors in 2001. It remains to be seen if these graphene transistors can be combined into complex chips with better performance than silicon.
Personally I'd put my money on the memristor: it can be built on existing fabs and has the potential to make chips smaller and faster.
This is annoying... Shepard research group has done this BEFORE IBM and we submitted our results for publication about 6 months ago... Should have done a press release.
Personally, I doubt that graphene will ever make it into any of the widely used machines or devices because it's so hard to produce it, but this is cool stuff. Just need to get electron mobility up.
When the world experiences an economic explosion, I anticipate a LOT of this kind of computer technology will be put in everyday technology.
When the world experiences an economic explosion, I anticipate a LOT of this kind of computer technology will be put in everyday technology.
Silicon, GTFO.
This is a very good thing and I always welcome new strides in transistor technology. This will be even a better thing especially when it trickles down to AMD.