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!
Seems interesting that foundries are finally buying on this. They are realizing that IoT will need less consuming IP ("For example, the smartwatch typically needs recharging every day. We think we can increase battery life of smart watches by a factor of four using our technology" Sanjay Jha, CEO Globalfoundries) and they want to try FDSOI on newer nodes.
My question would be: How long before they offer FDSOI as soon as the bulk is ready?
Anyway, GF wants to dedicate Fab1 to this technology, probably because of the geographic proximity to wafer suppliers.
Another question that would be interesting to follow: Will ST be able to enjoy this more and more appealing technology they are licensing?
I would love to hear your opinion.
This is talking about 22FDSOI vs 28bulk & 20bulk. 16FFC rocks. Chances are there for 22FD to knock the 16FFC. Maybe some analog or RF performance? Again it is limited by IP. Unless there is a focused spin-off IP company it is hard to find FD design experts on the market.
Same thing for 22FD to 28,20bulk, the thin performance gain from body bias has to be realized by engineers who know how to do it. If ST has the only talent pool in the world for FD, it actually difficult for them to reap the profit from it.
Just my humble opinion.
I was, of course, talking node to node in asking when will FDSOI take over vs. FinFet.
It may have been because I was first-handed ST communication, as I was working there 2 years ago, but they were all saying that if you know how to design in bulk, you can do it in FDSOI.
When you factor in the process difficulties in FinFet vs. FDSOI (supposed to be as easy as bulk) as well as more and more foundries adopting it, I was wondering how much time is there before we see the first significant circuit in FDSOI?
I would say at least a year since very few IP providers are interested in this technology but just by listening to newer communication on FDSOI, foundries are ready (2 years after the technology has been brought up by ST).
re: IP for FDSOI -- Cadence & Synopsys are now on board, as are other many other smaller suppliers. ARM's on board. For Samsung 28FDSOI, you get all the ST IP but you get it from Synopsys with Samsung part numbers and Samsung/Synopsys support. More IP announcements imminent. PDKs available for downloading now. Samsung's done their qualifications and are currently running silicon in MPW. First products ramping to high volumes next year. ST fabbed the SFARDS (bitcoin) 28nm FDSOI chip, which came out this spring. GF's 22nm FDSOI is quite close to ready -- finishing up details in the next few months, ramping to hi-volume in 12-18 months. Big design firms (Synapse, Verisilicon) with FDSOI expertise have already taped out a significant number of designs, more in the pipeline. It's happening. Read my last 10 ASN posts for details. Presentations from the major IP providers (and the foundries and many more) on their FDSOI offerings at from Leti Days last month now posted here. That said, I agree the foundries who are doing both, when they say FDSOI and FinFET will co-exist -- FinFET for high perf digital. FDSOI for everything else. Products in the pipeline from Sony, Freescale, ST, Cisco & others who aren't public yet.
