IBM Demonstrates a Nanosheet Transistor that Loves 77 Kelvin—Boiling Point of Nitrogen

[btarunr]

IBM, at the 2023 IEEE International Electron Device Meeting (IEDM), demonstrated a concept nanosheet transistor that posts a near 100% performance improvement at the boiling point of nitrogen, of 77 Kelvin (-196 °C). Given how relatively industrialized and scaled out the manufacture, safe transport, storage, and use of liquid nitrogen is, this development potentially unlocks a new class of chips that attain top performance under liquid nitrogen cooling. Think a new generation of AI HPC accelerators that can instantly double their performance under LN2, provided a new kind of cooling solution is developed for data-centers.

Nanosheet transistors are the evolutionary next step to FinFETs, which have been driving semiconductor foundries since 16 nm, which could see their technical limits met at 3 nm. Nanosheets are expected to make their debut with 2 nm-class nodes such as the TSMC N2 and Intel 20A. At an operating temperature of 77 K, IBM's nanosheet device is claimed to offer a near doubling in performance, due to less charge carrier scattering, which results in lower power. Reducing scattering reduces resistance in the wires, letting electrons move through the device more quickly. Combined with lower power, devices can drive a higher current at a given voltage. Cooling also results in greater sensitivity between the device's on and off positions, so it takes lesser power to switch between the two states, resulting in lower power. This lower power means that transistor widths can be lowered, resulting in higher transistor densities, or smaller chips. As of now IBM is wrestling with a technical challenge concerning the transistor's threshold voltage, a voltage which is needed to create a conducting channel between the source and the drain.



View at TechPowerUp Main Site | Source

 

[Wirko]

This means that new family homes and commercial buildings will have connections to water, sewer, electricity, internet, gas, and LN2 mains.

 

[RH92]

IBM heared people saying that XOC was pointless so they took it personally , now everyone will have to XOC

 

[mechtech]

Overclockers dream

 

[bug]

77K - when nitrogen boils, yet we freeze to death

 

[Soul_]

Overclockers dream

Last time someone used this phrase was, AMD's marketing team during launch of FuryX. It didn't age well at all, as we saw 3 weeks later. .

 

[Operandi]

Does anyone know why or have a link to a story as to why IBM sold off their fabs? IBM has always been doing stuff like this and from my casual readings on TSMC at lot of the advantage the have no is due to their close relationship with schools pursuing cutting edge fabrication techniques the same way IBM has always done. I always thought it was odd that IBM could not leverage and basically be the North American version of TSMC.

 

[AnotherReader]

Does anyone know why or have a link to a story as to why IBM sold off their fabs? IBM has always been doing stuff like this and from my casual readings on TSMC at lot of the advantage the have no is due to their close relationship with schools pursuing cutting edge fabrication techniques the same way IBM has always done. I always thought it was odd that IBM could not leverage and basically be the North American version of TSMC.

They sold the fabs, because they didn't have enough volume to justify spending on fabs. In fact, they had to pay Global Foundries $1.5 billion in 2014 to take them off their books. IBM didn't really play as a foundry even though there had been times in the past when it produced other companies' chips like Nvidia's GeForce FX in 2003.

 

[MentalAcetylide]

This means that new family homes and commercial buildings will have connections to water, sewer, electricity, internet, gas, and LN2 mains.

heh, never happen for residential. Gas lines are a challenge themselves, but manageable; however, LN2 lines underground would cost way too much. Plus a major leak could mean a lot of dead people depending on the circumstances since the gas is just barely less dense than air.

 

[Wirko]

They sold the fabs, because they didn't have enough volume to justify spending on fabs. In fact, they had to pay Global Foundries $1.5 billion in 2014 to take them off their books. IBM didn't really play as a foundry even though there had been times in the past when it produced other companies' chips like Nvidia's GeForce FX in 2003.

However IBM has remained very active in basic research. They demonstrated a 2nm nanosheeted prototype chip in May 2021, for example. I guess they (and Imec of Belgium) make a lot of money selling distant future tech to other parties.

 

[AnotherReader]

However IBM has remained very active in basic research. They demonstrated a 2nm nanosheeted prototype chip in May 2021, for example. I guess they (and Imec of Belgium) make a lot of money selling distant future tech to other parties.

Yes, it's easier to monetize research breakthroughs compared to running a fab and upgrading it nearly continuously.

 

[R-T-B]

Plus a major leak could mean a lot of dead people depending on the circumstances since the gas is just barely less dense than air.

I hate to break it to you but a leak wouldn't do much to people. We breathe an atmosphere that is majority Nitrogen. Unless you breathed it in a liquid state raising the density a bit wouldn't change much for pretty much anyone.

 

[Wirko]

I hate to break it to you but a leak wouldn't do much to people. We breathe an atmosphere that is majority Nitrogen. Unless you breathed it in a liquid state raising the density a bit wouldn't change much for pretty much anyone.

Indeed, the atmospheric nitrogen isn't in a hurry to travel up, or down.
I can still see a major leak in an enclosed space being dangerous or even fatal to people, though, if it displaces most of the oxygen.

 

[ymdhis]

I hate to break it to you but a leak wouldn't do much to people. We breathe an atmosphere that is majority Nitrogen. Unless you breathed it in a liquid state raising the density a bit wouldn't change much for pretty much anyone.

Air is a mixture of nitrogen and oxygen, if the ratio of that mixture is changed so that nitrogen is in larger quantities, then it can become lethal quite fast especially in closed areas (ie. underground, which is where the pipes usually are).
I'd also be concerned about what the liquid nitrogen will do to whatever it spills into, such as the ground or other pipes. At the temperatures it has to stay at to remain liquid, it could potentially cause infrastructure damage just by freezing everything in the vicinity.

This is all theoretical because I don't imagine LN2 will become available for commercial housings but let's not be completely dumb and say that just because air contains nitrogen, we can breath in any amount of nitrogen without harm. That's dumb even by American schooling standards.

 

[R-T-B]

we can breath in any amount of nitrogen without harm.

I wasn't claiming that to be clear. I agree with the rest.

 

[Minus Infinity]

Wow so apart from the egregious amounts of power AI already consumes, we will need egregious amounts of extra power for LN cooling systems.

 

[usiname]

Its cheaper to have 2x large chip that work on normal temperatura than keep smaller chip with LN2 cooling for just double the performance. The current chips can overclock with ~30% more under LN2, so not so much better than what we have now.

 

[Beginner Macro Device]

I hate to break it to you but a leak wouldn't do much to people. We breathe an atmosphere that is majority Nitrogen. Unless you breathed it in a liquid state raising the density a bit wouldn't change much for pretty much anyone.

I am not a maths expert by any mean but going up by like 660 thousand times in volume in a relatively short amount of time will lead to massive damage. It's basically an explosive. Not as explosive as nukes but still.

 

[Wirko]

How complex and costly are LN2-making machines? Perhaps datacenters could have their own.

I am not a maths expert by any mean but going up by like 660 thousand times in volume in a relatively short amount of time will lead to massive damage. It's basically an explosive. Not as explosive as nukes but still.

About 660 times. Evaporation can be spectacular but not anything close to explosive.

 

[TumbleGeorge]

Not needs to be highly explosive for to be deadly disaster.

 

[Assimilator]

Does anyone know why or have a link to a story as to why IBM sold off their fabs?

1. The majority of IBM's fab capacity was dedicated to producing PowerPC CPUs for Apple. Because they were the only supplier they were ripping Apple off big-time, and for the same reason PowerPC's performance was rubbish.
2. IBM's fabs were far behind the curve in terms of process technology, since they'd been using their profits from the first point to shore up other areas of their business, instead of reinvesting into improving their fabs. This made their CPUs even worse.
3. Apple got tired of IBM being complete and utter shit, and switched to Intel CPUs instead.
4. IBM was suddenly stuck with outdated fabs that had no customers, so they made a good decision for once and got rid of said fabs. But because those fabs were so utterly garbage, instead of making money from the sale, they had to pay GlobalFoundries to take their junk.

 

[Panther_Seraphin]

How complex and costly are LN2-making machines? Perhaps datacenters could have their own.

Complex? Not a huge amount beyond an extremely low temp AC system. And to seperate Nitrogen from other gases isnt terribly hard either.

Costly? Thats where the rub is Its a lot of mechanical effort to drive that sort of refrigeration cycle and keep up the gas pressures to constantly feed gaseous nitrogen into it.

About 660 times. Evaporation can be spectacular but not anything close to explosive.

Chernobyl would like a word

 

[FeelinFroggy]

LN2 will not be used to cool home PC systems in the future, because in the future there will not be any home PC's. There wont be a need for everyone to have their own PC, they will all be cloud based severs that people will access. Those cloud based servers will probably be cooled by LN2 on an industrial scale. Cloud computing (and gaming) is the future. AI will remove bottlenecks in the broadband to make it all happen.

So there is really no reason for someone to be worried about being asphyxiated from LN2.

 

[R-T-B]

Chernobyl would like a word

That was a little more than a LN2 problem.

Evaporation can be spectacular but not anything close to explosive.

Yeah. They used to expose LN2 to air to do fun tricks at the local science center. It did cool air fog but did not explode at all.

 

[Panther_Seraphin]

That was a little more than a LN2 problem.

True but the initial explosion was actually a steam explosion aka evaporation.


I am curious to see if this helps with things where People say RAM/Cache doesnt benefit from the smaller nodes. Would it be a way of stacking cache so instead of extra die space being taken up with extra cache all they do is stack it with this new tech meaning core sizes remain small.

Would be really interesting for perhaps making x3d almost obsolete?