[HN Gopher] Graphene Interconnects Aim to Give Moore's Law New Life
___________________________________________________________________
Graphene Interconnects Aim to Give Moore's Law New Life
Author : jnord
Score : 59 points
Date : 2024-12-11 13:22 UTC (3 days ago)
(HTM) web link (spectrum.ieee.org)
(TXT) w3m dump (spectrum.ieee.org)
| gigatexal wrote:
| Awesome! Let's hope Intel -- for their sake --- can make this
| happen.
|
| But I'm already thinking about light CPUs that use light instead
| of electricity for computation. Of course I don't fully know how
| it works but it seems to be lower power and the next iteration of
| computation I guess before we get to room temp quantum computers.
| Vecr wrote:
| Quantum computers are never* going to be good at a whole lot of
| tasks that classical computers are already used for.
|
| * Some people have weird ideas.
| freehorse wrote:
| Well, right now, a magical way to resurrect moore's law is no
| more or less crazy than a magical way to scale quantum
| computation.
| ben_w wrote:
| The problem with light is that it's quite a lot bigger than the
| features on current chips.
|
| If you get enough other benefits from going up from 2nm
| features to e.g. 200nm UV-C photons, then you may still choose
| to do so.
| graemep wrote:
| Do optical gates not switch a lot faster? I do not know
| whether it would be enough to offset the bigger size.
| ben_w wrote:
| The switches themselves do (IIRC by a factor of about 1e4),
| but if you have to space them farther apart then the
| combined whole may not benefit from this.
|
| If you have a system clock running at 3 GHz, the speed of
| light limits your causal distance to just under 10cm per
| clock cycle. CPUs are already close to that for size and
| frequency, but let's say you're taking a 1 cm by 1cm
| silicon chip on a 2nm feature size process and replacing it
| like-for-like with a photonic chip with light that limits
| it to 200nm features -- now it's 1m by 1m and can't go
| faster than 300 MHz, likely a lot less.
|
| This doesn't mean it is useless -- for example, there's a
| hope that it will reduce energy use, which is directly
| useful all by itself, but also means it may be sensible to
| move to a fully 3D structure which silicon can't really
| manage because of the thermal issues. Going from 2D to 3D
| helps a lot, might allow that 1m by 1m by 200nm (*2
| thickness for insulation) sheet to be compacted to a 7.4mm
| cube, which then doesn't need to be slowed all the way down
| to 300 MHz due to causality.
| bgnn wrote:
| Very interesting. What about memory? Current problems are
| mainly memory bottleneck related. How can one solve that
| in photonic chips?
| bgnn wrote:
| Intereting question. The answer to the second part: we have
| much faster switching transistors (GaAs, SiGe, InP, now
| GaN) already but they cannot be miniaturized easily and the
| production technology isn't as simple as CMOS. One can
| build computers with them, but due to physical size and
| large distances it wouldn't be performing good compared to
| a CMOS chip. So the answer is: size matters. Large devices
| cannot be used for building complex fast computers.
| awestroke wrote:
| Let's hope the fab that pioneers this is not owned by Intel
| propter_hoc wrote:
| Photonic computation is never going to make sense as an
| alternative to electrical computation.
|
| Among other reasons, you can create an electronic transistor in
| silicon by using an electrical signal to open and close a gate.
|
| You can't really do this with light, light beams just pass
| through each other. And the kind of light-carrying media that
| can be affected by the presence of a control beam respond much
| slower and less effectively than doped silicon responds to
| voltage.
| bgnn wrote:
| This! And optical waveguides are big, and they need to be
| spaced apart to avoid interference. Speed of light is
| limiting for such large circuits to be fast.
|
| Nothing beats CMOS transistors in density.
| devonsolomon wrote:
| Graphene seems to be like a hotshot actor who lands a million
| auditions but somehow never makes it past a walk-on role in a
| toothpaste commercial.
| XorNot wrote:
| It's because it's just about impossible to handle: the number
| one thing a sheet of graphene wants to do is stick another
| sheet of graphene on top of it and become...regular graphite.
| dmead wrote:
| Well someone needs to tell graphene so stop fucking it's
| coworkers and get back to set
| tliltocatl wrote:
| The kids these days are so spoiled. Silicon doping was
| discovered like when? And how long did it take to make a
| practical transistor? Seriously through, it's not every new
| discovered phenomena owes you something.
| devonsolomon wrote:
| True. Guess I'm disheartened by years of clickbait.
| api wrote:
| It's okay. Next year we will defeat and reverse aging with
| one simple trick so you can wait longer, at least according
| to the latest health science click bait.
| bryanrasmussen wrote:
| I will not rest until I have you immortal, flying your
| fusion powered car, using augmented reality VR controls,
| to your very own immersive shopping experience with AI
| assistant android sexbots catering to your every whim and
| I will not REM enhanced super-sleep until that happens!
|
| I'll give you fifteen minutes to call me back.
|
| /Jerry Maquire out
| BiteCode_dev wrote:
| 20 years.
|
| And we have been able to produce graphene around 2004 I
| believe, so we are going soon to cross that threshold.
| dtgriscom wrote:
| I've been watching technology for the last fifty+ years, and
| I had the same (admittedly unfair) reaction as the OP.
| tliltocatl wrote:
| Lol I'm obviously joking, I'm probably younger than both OP
| and 70% of people out here. But my point that the nature
| doesn't owe us anything still stands. University press
| releases are really to blame for building up unrealistic
| expectations, but then you can't expect them to honestly
| tell you "we spend millions on things with zero practical
| applicability just because it's awesome".
| api wrote:
| It takes a long time to go from lab bench and physics papers to
| practical use to mass produced and generally available
| practical use.
|
| Graphene has incredible properties as a structural material too
| but so far producing it at that scale and making it behave
| properly in things like composites has been very hard. But the
| physics says once we get it to work we have composites many
| times stronger than steel or materials like Kevlar.
| brightball wrote:
| The reason is that it's very difficult to get a consistent
| product from mining, from what I have heard.
| gaze wrote:
| There's a few reasons for this. There's a few ways to make
| graphene. You can use CVD or you can use mechanical
| exfoliation. Mechanical exfoliation requires scotch tape and
| scales to maybe a flake per hour per grad student. CVD is quite
| scalable but makes shitty graphene. A lot of graphene
| breakthroughs (superconductivity for instance) needs
| mechanically exfoliated graphene.
|
| Secondly, process fab is VERY conservative. There's numerous
| amazing ferroelectrics that you can grow tons of that would
| absolutely spank NAND flash. However, they're not silicon fab,
| so nobody makes them.
| dehrmann wrote:
| But scotch tape is nearly as cheap as grad students.
| tbrownaw wrote:
| > _There's numerous amazing ferroelectrics that you can grow
| tons of that would absolutely spank NAND flash. However,
| they're not silicon fab, so nobody makes them._
|
| So why doesn't somebody new start making them and put all the
| current flash producers out of business?
| phkahler wrote:
| Moore's law is over. Nothing is going to restore that regular
| cadence of device shrink and performance increases. Each
| innovation is now a single tiny step in the endgame of scaling.
| api wrote:
| We are still many clicks from physical limits for computation,
| so it depends on how much money we want to spend.
| kibwen wrote:
| _" Rock's law or Moore's second law, named for Arthur Rock or
| Gordon Moore, says that the cost of a semiconductor chip
| fabrication plant doubles every four years. As of 2015, the
| price had reached about 14 billion US dollars._"
|
| https://en.wikipedia.org/wiki/Moore%27s_second_law
|
| It seems likely that we're relatively close to the point
| where it will no longer be economical to push the limits
| here. It's unlikely that even the entire world working
| together would want to spend more than $1T for a single fab,
| which Rock's law suggests is less than 20 years away.
| aurareturn wrote:
| It seems likely that we're relatively close to the point
| where it will no longer be economical to push the limits
| here. It's unlikely that even the entire world working
| together would want to spend more than $1T for a single
| fab, which Rock's law suggests is less than 20 years away.
|
| Given that Apple at the start of 2019 was worth $600
| billion, and now $3.7 trillion 5 short years later, I think
| a $1 trillion plant in 2045 is not so farfetched. This is
| especially true if compute requirements for AI continues to
| grow.
|
| Twenty years is a long time. I don't think people in 2025
| could have predicted out needs for chips back in 2005.
| fieldcny wrote:
| Your are conflating asset price inflation and cost
| inflation, they are not the same. Apple could lose $2T in
| market cap next week, the cost of the fab would not be
| discounted in the same way.
| dehrmann wrote:
| That still means cheaper transistors, right?
| Mistletoe wrote:
| It feels like the next era and maybe for the rest of humanity's
| existence is the Age of the Plateau. I wonder how they will
| handle it? We lived in such a special time in all of human
| existence.
| smartbit wrote:
| > _The sacrificial film is placed on top of the transistor chip,
| and a source of carbon is deposited on top. Then, using a
| pressure of roughly 410 to 550 kilopascals, the carbon is forced
| through the sacrificial metal, and recombines into clean
| multilayer graphene underneath. The sacrificial metal is then
| simply removed, leaving the graphene on-chip for patterning._
|
| Incredible
___________________________________________________________________
(page generated 2024-12-14 23:01 UTC)