• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

TSMC Plans to Put a Trillion Transistors on a Single Package by 2030

AleksandarK

News Editor
Staff member
Joined
Aug 19, 2017
Messages
3,191 (1.11/day)
During the recent IEDM conference, TSMC previewed its process roadmap for delivering next-generation chip packages packing over one trillion transistors by 2030. This aligns with similar long-term visions from Intel. Such enormous transistor counts will come through advanced 3D packaging of multiple chipsets. But TSMC also aims to push monolithic chip complexity higher, ultimately enabling 200 billion transistor designs on a single die. This requires steady enhancement of TSMC's planned N2, N2P, N1.4, and N1 nodes, which are slated to arrive between now and the end of the decade. While multi-chipset architectures are currently gaining favor, TSMC asserts both packaging density and raw transistor density must scale up in tandem. Some perspective on the magnitude of TSMC's goals include NVIDIA's 80 billion transistor GH100 GPU—among today's largest chips, excluding wafer-scale designs from Cerebras.

Yet TSMC's roadmap calls for more than doubling that, first with over 100 billion transistor monolithic designs, then eventually 200 billion. Of course, yields become more challenging as die sizes grow, which is where advanced packaging of smaller chiplets becomes crucial. Multi-chip module offerings like AMD's MI300X and Intel's Ponte Vecchio already integrate dozens of tiles, with PVC having 47 tiles. TSMC envisions this expansion to chip packages housing more than a trillion transistors via its CoWoS, InFO, 3D stacking, and many other technologies. While the scaling cadence has recently slowed, TSMC remains confident in achieving both packaging and process breakthroughs to meet future density demands. The foundry's continuous investment ensures progress in unlocking next-generation semiconductor capabilities. But physics ultimately dictates timelines, no matter how aggressive the roadmap.



View at TechPowerUp Main Site | Source
 
Only thing that came to mind

1703690634932.jpeg
 
12x the number of transistors of the H100
6-7x the mi300x

Even for TSMC, this sounds absurd; the transition from 5nm to 3nm yielded less than a 50% improvement in transistor density. but I hope to see 3D chips from now on.
 
Мore transistors than number of combine of all stars in Milky way and other closest galaxies..
 
That is what they do when things are bad and they dont want to tell investors the truth.
 
Hahahaha....actually, both of these will probably come close to being somewhat accurate, because as the nodes & technology advance, so does the costs of building the fabs and the equipment to make the chips :)
Maybe that's what Gordon Moore had in mind when he uttered Moore's Second Law.
 
Be prepared for 5000$ Video Cards for end user gamers. That would be the price due to 1% successful yields...
 
Be prepared for 5000$ Video Cards for end user gamers. That would be the price due to 1% successful yields...

I no longer believe the lies that these companies spew; what I do see, are patterns, and the patterns tell a different story to their lies. They have deliberately slowed down production to keep prices inflated in the past, the RTX4000 series being a more recent one.

That being said; perhaps, some of you have noticed, that the crypto market is gearing up for another bull run, and though ETH / Bitcoin is no longer economically viable for GPU mining, there will be a lot of altcoins that will be.

If that moron in his leather jacket doesn't ramp up production soon, we are going to see shortages that will make the year 2020 rush look like a picnic in comparison. Watch this space, remember, if you have an RTX2000+ you have the equivalent of a gold bar in value during said bull run, that is, until the market experiences its inevitable crash.
 
Мore transistors than number of combine of all stars in Milky way and other closest galaxies..
Interesting point about that, when I was a kid (pre-hubble space telescope) they estimated the stars in the Milky Way at 1-2 billion

Apparently now the latest estimation is 100-200 billion stars, I'm guessing due to better telescopes? Maybe this will get revised over the next decade with JWST?
 
they estimated the stars in the Milky Way at 1-2 billion
I do not think. Never since we know what a galaxy is and that we live in one has there been such a low estimate of the number of stars in the Milky Way.
 
Gotta be careful with long term predictions. Otherwise you end up really wrong, like Intel's promise of 10GHz by 2005 running at < 1 volt.
Yup, that actually happened, and that statement was only 5 years prior in 2000.
 
Back
Top