- Wednesday, 25 November 2009
- INQ Mobile
- RSS
Whoever gossips to you will gossip of you - Spanish proverb
TSMC lands Fujitsu deal
TSMC IS ON A ROLL, it seems. After having successfully overcome its 40nm yield issues and secured some Nvidia business, it is flexing its muscle and prodding its partners into action.
After Qualcomm's CDMA Technology man, Jim Clifford, gushed pleasantries about its partnership with hulking chip fabber TSMC, it's time for Fujitsu to hop on board the 28nm train.
Fujitsu and TSMC jointly announced their engagement to produce 28nm logic and develop a further "enhanced 28nm high-performance process" together. Fujitsu has, so far, used TSMC on the latest three full nodes (90nm, 65nm and 45nm) and this sounds like a natural evolution of the partnership. Fujitsu, on the other hand, will help TSMC in improving packaging with its expertise in extra-low K copper interconnects.
It's true, Fujitsu is as big as they get and manufactures just about anything electronic. From radios to set-top boxes, handsets, notebooks, desktops to servers and big iron, it has its fingers in a lot of pies. There's nothing unusual about committing to a fabbing process ahead of the game, but at this point it does look like TSMC is sounding off to its partners to make some noise.
One question remains unanswered, though. Where does the "enhanced 28nm high-performance process" come from?
What we do know is that it costs quite a lot of time and money to develop a new process node single-handedly, and TSMC supposedly already has one developed. It could be that Fujitsu is bankrolling the 28HP (28nm High Performance) process jointly with TSMC, which would definitely make the pain a bit easier to endure. But alas, Fujitsu and TSMC haven't yet returned calls or emails. It is pretty late in the day in Japan and Taiwan. µ
Share this:
ShareBusiness Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
http://www.semiconductor.net/article/200711-IBM_Alliance_Ready_With_28_nm_Eval_Kits.php
http://www.semiconductor.net/article/203006-Citing_High_k_Costs_TSMC_Plans_Dual_Track_28_nm_Solutions_in_2010.php
fujitsu is i guess is pushing TSMC for...
http://www.xbitlabs.com/news/cpu/display/20090515100227_Fujitsu_Discloses_Further_Details_About_Venus_Chip_Eight_Core_Processor_that_Consumes_35W.html
because of ...
http://en.wikipedia.org/wiki/SPARC_Enterprise#Processor
and ...
http://en.wikipedia.org/wiki/Itanium#Future_processors
Also this could fuel war of not just nano-meter but also sub process. So After GHz war, then nm war, it will be SOI/SiON (LPT) or High-K (HP) war. Could it be thought that AMDs improving 45nm yields (Steppings) were movement from SOI/SiON to high-k !?!
This could also mean AMD will stop preaching its glorious SOI. High-K is the way or atleast the better one of two 32/28nm fab processes or start with crapy steppings (LPT) of 28/32nm and move to performance steppings (HP) as AMD most visibly does or we can say when AMD enshowers 32nm blessings upon thy self, thee shalt asketh... behold is it crapth SOI 32nm !?!
poly oxide gate=SOI/SiON (LPT)
Quotes from prev post URLs
===========================
“With a poly oxide gate, they have to jack the voltage up to get the same performance, and that makes the leakage that much worse.”
"TSMC positions its low-power 28 nm offering as lower cost than the IBM high-k offering. IBM’s analysis is that while the gate-first high-k process adds one mask, by not using stressors IBM has a net reduction of three masks"
Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC, Hsinchu, Taiwan) said it will offer both silicon oxynitride (SiON) and high-k/metal gate solutions at the 28 nm node, with early multi-wafer shuttles starting late this year in time for volume manufacturing beginning early in 2010.
The 28LPT (low-power turbo) process, based on oxynitrides, will enter early manufacturing in the first quarter of 2010, and the 28HP (high-performance) process, with a high-k/metal gate stack, will be ready in the first half of 2010.
another thorough read now possibly reveals it not two processes SOI/SiON (LPT) or High-K (HP)... its...
SOI/SiON DualGate-LPT
quick, cheap chemical vapor deposition (CVD) process
SiON process with a 90% yield may compare with an estimated 70% yield for a high-k/metal gate
SOI/SiON TripleGate-HP
same as above with third gate/transistor to increase performance than High-K.
High-K
low leakage, high performance
high capacitance, requires high active power
thiner but expensive EOT
slow, expensive atomic layer deposition (ALD)
--------------------------------------
“Our 28 nm high-k/metal gate is not at a very mature stage now,” Wei said. “Our plan is that by the second half of 2009, the technology will reach its maturity stage. Some customers cannot wait; they want to do it early, so we try to accommodate their needs.”
meaning the customer is fujitsu pushing TSMC as xbitlabs.com article suggests. because Tukwila-ITANIUM is already too much delayed.