Thanks for the update and your opinion of the lead Intel has carved out for itself.
It was my impression that Common Platform Alliance partners had been pursuing FDSOI as an option alongside FinFET. However, with TSMC not able to deliver FinFETs before the 14-nm node it appears that even ARM CEO Warren East acknowledges that Intel has pulled off another chip manufacturing coup.
The main reason why ARM has not actively involved with IMEC is that ARM is an active member of IBM semiconductor industry alliances and SOI industry consortia. I recommend for readers to read two recent articles published in EDN news and see how deeply ARM was involved with FDSOI (fully depleted silicon on insulator) technology developed by IBM. The first article is entitled, “Fully depleted SOI shows its stuff in CPU design (An ARM cortex MO paper design suggest that FDSOI could be a strong contender at 20 nm)”, 2011-02-10, EDN. But IBM suddenly abandons FDSOI technology and adopts bulk CMOS technology for 28nm node. The second article entitled “CNSE moves on 28 nm with IBM license (IBM licenses low power, 28nm high-k metal gate bulk CMOS technology to the college of Nanoscale Science and Engineering of the University at Albany)”, 2011-03-09, EDN was big disappointment to ARM because ARM was heavily relied on IBM’s FDSOI technology for low power 28 nm. Please read my feedbacks at the end of each article. These events occurred just before Intel’s surprise announcement of its 22nm low power FinFETs slated to be volume manufactured in second half of 2011. This is an enormous achievement for Intel. No one predicted it. The FinFET technology is not new, over ten years old. The technology is researched by a number of semiconductor companies, and the merits for low power application are well recognized, but the remaining difficult key issue was its manufacturability. Intel is the first one that has done it. Intel will manufacture the 22nm low power FinFETS in high volume this year, 2 to 3 years ahead of its rivals, and will manufacture 14nm low power FinFET technology way ahead. IBM and its industry alliance members including ARM are big losers. I doubt IBM will continue to work on low power bulk 28nm. IBM has spent too much and too long on FDSOI and ETSOI.
By the same token should ARM be doing leading-edge device and IC research and presenting at conferences such as IEDM and ISSCC?
It is not a low cost exercise to engage with IMEC in collaborative research. ARM is probably doing the smart thing to take what lessons it can from the learned conferences and technology forums and leave it to the foundries and fabless to pay for the research.
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...