You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
12FDX has been coming "real soon now" for many years, but -- like many others, I suspect -- I never saw the point of it...
22FDX is a great niche process for some applications like high-speed low-power mixed-signal/analog, having very fast low-capacitance low-power NMOS and PMOS transistors...
Because moving from N2 to A16 needs a much bigger investment in IP and layout/tools than moving from N3 to N2, and is pretty much a one-way transition (difficult to reverse)?
N2 is pretty much N3 with a different transistor dropped in underneath, the libraries and IP and power grid/routing...
Agreed, but there's a big difference between a complete Intel-style flop and a TSMC-style pipecleaner... ;-)
I'm confident that A16 will get into mass production, probably with a relatively small number of designs but some of these (Apple?) will be big volume -- basically, customers with big...
Which is precisely why TSMC is introducing BSPD as a "second-step" 2nm process (now called A16) after introducing GAA in 2nm (which is pretty much 3nm metal with new transistors underneath) -- do one big new thing at a time to reduce risk, and so it's not a total disaster if things get delayed...
I'm sure that AI will improve health care costs/results, but it won't make the US any better compared to other countries.
Your last point is the key -- the US medical system is not cost effective, it's one of the worst in the world in that regard. AI and robotics won't change this so long as it...
Funny, many other countries manage to provide better care for everybody -- regardless of responsibility -- for a lot less money than the USA does... ;-)
Probably because the vast majority of people who fall ill and need care are not to blame for their condition, and it's cheaper to treat...
Today without AI the US has by far the most expensive medical care in the world and is way down the list for quality of care.
With AI in the future -- which will be applied everywhere, not just the USA -- the US will *still* have the most expensive medical care in the world and be way down the...
Maybe not at first for 18A if Intel get there before A16 and can ramp it up fast. But even if they manage this (which I doubt, but let's see...) I suspect A16 will soon overtake it because -- well, Nvidia... ;-)
Did you actually read what you posted?
"A16 is best suited for HPC products with complex signal routes and dense power delivery networks, as these can benefit the most from backside power delivery".
What they don't say is that there's also quite a big cost premium for A16 -- and yes, I've...
Which is exactly why I said the real competition for volume CPU production -- what really matters -- with BSPD is 18A for Intel vs. A16 for AMD in 2026...
Small-volume sampling earlier than this makes headlines and a "product launch", but this is exactly what Intel did in 10nm and then failed...
So the billion-dollar question is -- will Intel actually deliver both the 18A process and the products designed into it *in volume* in 2025?
Not the kind of "hey' we've got this fabulous new process (but it doesn't yield and we can only make a few chips)" launch they famously did in 10nm.
If...
All true, but since the area gain for BSPD is mainly because of removing the topside power grid and shortening each row of cells (and therefore interconnect length, which is where the power/speed improvement mainly comes from) it's obvious that the area saving depends on how dense the grid was...
Of course they don't mention the power/yield penalty of doing this in a Samsung 5nm process, which is likely to be very high -- though I guess IBM don't care... ;-)
Cost or power per core or GOP/GFLOP has never been much of a concern for them, it's all about pushing speed per core and big-iron...
It would, but apart from having a competitive silicon technology to TSMC in PPA -- which they may well succeed at -- they have several other massive hills to climb, not just capex.
The first is cost competitiveness against a much bigger competitor pushing far more different products through the...
I've seen numbers from TSMC showing area saving (and speed increase/power decrease) with BSPD (N2 vs. A16), and they strongly depend on the power grid density -- which in turn depends in the die power density (W/mm2) and chip design. The area savings can be up to >15% in the best case, but also...
It's still nothing to do with "going risk based" or "not rewarding bad health habits" or "using new technology" or "improving doctor's practices" -- as you mention, the single fundamental problem with the US health system is that it focuses on profit not the health of citizens, it's capitalism...
There's a similar system here in the UK -- no drug/doctor advertising is allowed, and NICE (National Institute for Clinical Excellence) evaluates all treatments and drugs, decides which ones are worth using, and then negotiates a (usually massively discounted) price with the manufacturer on...
You're still diverting from the real issue -- regardless of who you think should pay (the state or individuals) or in what proportions or whose responsibility it is, the US has a fundamentally broken system which prioritizes on the wrong outcomes, money and corporate profits not the health of...
