Is the focus on Moore's law a good thing or should the be thinking of other things besides shrinking things?
Tiny heat-sucker helps keep Moore's Law going
Adding graphene to copper can improve its heat-conducting properties – and that's good news for electronics developers. One of the barriers to the continuing shrinking of electronics is that as you pack more features into a smaller space, you also pack more heat into that space. That makes heat a hurdle standing in the way of …
-
-
Thursday 13th March 2014 08:17 GMT A Non e-mouse
You mean things like:
- Parallelism (multi cores/NUMA/Transputers/GPGPUs)
- Instruction designs (RISC/CISC/VLIW)
- Super-scaler architectures
- Branch prediction
- Multi-level caching
- Out-of-order execution
- Alternative high(er) level languages
- Compiler design
- Quantum computing
- Neural-net based systems?
-
Thursday 13th March 2014 11:26 GMT Don Jefe
Alternatives to anything are always a good idea to look at. If nothing else it stimulates thinking.
That being said, Moore's Law is often misunderstood, it isn't a technical observation. Moore's Law is a commercial observation and, with that in mind, makes it an entirely different kind of animal. For Moore's General Industry Forecast to change something radical will have to occur.
At present the entire industry is trapped by Moore's Law. Everything about the manufacturing side of the industry has been refined to suit the needs of a particular supply chain, manufacture, market and changing extremely integrated industry operations always takes a very, very long time and the search for good 'industry accepted' alternatives usually doesn't begin in earnest until after you run into a brick wall.
Will it change? Undoubtedly, but odds are change will come from an entirely unexpected direction and take a wholly unexpected form. Sometimes that takes a very, very long time, but when it finally occurs it is a monumental shift and everything except the price is greatly improved. Until then, little baby steps to keep things as they are is all we've got.
-
-
-
Thursday 13th March 2014 07:56 GMT Ken Hagan
Re: conductivity
Only somewhat. Both are better when you have a crystalline structure (as in a metal), but whilst conductivity needs a band of free electrons, heat just needs matter, which is why even insulators conduct heat.
Also, even a 24% improvement in conductivity isn't going to be a game-changer for interconnects if they have very low resistive losses anyway. (I suppose someone who works in a chip foundry will now butt in and say that, in the very thin wires we're talking about, resistive losses *are* a problem. Let's see...)
-
Thursday 13th March 2014 08:32 GMT Charles Manning
What about diamonds then?
Diamonds conduct heat many times better than copper, yet are for all practical purposes perfect electrical insulators.
Heat conduction does not require electron migration, just the ability to transfer the thermal energy (in layman's terms - pass the heat "vibration" from one atom to another). Strong, rigid bonds do this well, hence diamonds perform pretty well.
As a side note, this is also why ice has better thermal conductivity than liquid water.
-
-
-
Thursday 13th March 2014 08:21 GMT Charles Manning
24% bah!
To have any promise (ie. game changing), you need technologies that give hundreds of % increase. 24% is piddling. You get way better than 24% by doing a process shrink.
This is about the same as pure silver which is way cheaper than fancy copper/graphene mixtures.
We already have heat pipes that give us 250x (ie. 25000%) the conduction of copper. That is 200 times as good as this graphene/copper sandwich.
This is only news coz it has graphene in the press release.
-
Thursday 13th March 2014 08:47 GMT Charles 9
Re: 24% bah!
We already have heat pipes that give us 250x (ie. 25000%) the conduction of copper. That is 200 times as good as this graphene/copper sandwich.
So why aren't they actually being used if they're so good? Probably because they're too novel and expensive. The one thing we don't know about is how much this new process costs. I mean, 250x improvement doesn't mean much if it costs 300x as much, but if a 25% improvement only raises the cost by 10%, most would consider that a practical ROI.
PS. I have a thing about these research article. I see so many research articles and not to many "going to market" articles about new tech. I feel the research articles could stand some control so that we mostly see things that actually have a reasonable chance of actually going to market. Otherwise, we're just being teased.
-
Thursday 13th March 2014 08:54 GMT Richard 12
Re: 24% bah!
Those heat pipes are already being used.
The reason this is interesting is scale - heat pipes are physically many orders of magnitude larger, what with all the tubes and flow, while this works on the nanometre interconnects which are already there.
That said, it does rather sound like another, cheaper process might give the same result. Is this simply due to the annealing effect of the application process, or does the graphene itself do the work?
-
Thursday 13th March 2014 11:40 GMT Don Jefe
Re: 24% bah!
Charles is correct. This is a very positive development, but is nothing more than an incremental step in the evolution of a product. It will certainly show some benefit, eventually, but at best anything that occurs will be a short term solution and, without 24% more improvement in very short order equilibrium will be restored.
That's not to say this isn't a really cool thing, but 24% isn't nearly as significant as it sounds. Intel share prices will be kept steady, and that's about it. The Graphene 'industry' really, really wants to get into the IC sector, but, as others noted previously, applications in non IC markets will likely benefit the most.
-
-
-
Thursday 13th March 2014 16:20 GMT mIRCat
You only hear what you want to hear... now with more graphene!
"Researchers at the University of California, Riverside, have found that a single-atom-thick graphene coating"
I may be mistaken, but would it not have been sufficient to simply say "graphene"? Perhaps the copy meant 'a single-atom-thick graphite coating'.
-