Electronics Resurgence

In his seminal 1965 paper, microelectronics pioneer Gordon Moore famously predicted that the transistor count of integrated circuits would double every two years while transistor cost decreased. The physics and economics of transistor scaling has made it difficult to maintain this trajectory, pushing Moore's Law towards an inflection point that could mark a slowdown in technological progress. In a keynote session from DARPA’s 60th anniversary symposium, D60, Dr. Bill Chappell, director of the agency’s Microsystems Technology Office, introduces the Electronics Resurgence Initiative (ERI), a DARPA-led partnership to continue the pace of technological advancement by moving away from chip generalization and abstraction, central tenets of Moore's Law, and to enable an audacious era of innovation based on chip specialization and uniqueness.

Distinguished panelists Lynn Conway, Robert Kahn, and Mark Papermaster describe their efforts to both further and benefit from Moore's Law, including DARPA investments leading to new circuit design and fabrication methodologies, the Internet, and the iPad. MTO program manager Andreas Olofsson discusses how the design thrust of ERI is dramatically lowering the barriers to modern System-on-Chip design through two new electronic design automation research programs.

Moderator & Keynote
Dr. Bill Chappell – DARPA, MTO

Panelists
Ms. Lynn Conway – University of Michigan, Ann Arbor
Dr. Robert Kahn – Corporation for National Research Initiatives
Mr. Andreas Olofsson – DARPA, MTO
Mr. Mark Papermaster – Advanced Micro Devices

D60 took place Sept. 5-7, 2018, at Gaylord National Harbor, Oxon Hill, Maryland.

[metanumia]

I really wish there was some sort of grand strategy at the highest levels of Western government/democracy with a focus on expanding, improving, and gaining dominance in future electronics engineering, design, and manufacturing technologies. There appears to be a lack of coherent leadership, strategy, and vision right now, not referring to the good folks at DARPA, I'm talking about politicians and elected officials, they don't appear to understand or even listen to the ideas coming out of organizations and think-tanks like DARPA, RAND, CSIS, etc... Maybe DARPA could start innovating a small drone with a computer-brain-interface that can attach to politicians' heads and make them more intelligent and strategic.

[jaycraswell]

I was a little disturbed by the comments about security - or lack of it being a part of computer design. Saying security was on someones mind in the early years is not the same as doing something about it. If the computer users of the early Darpa era really found security to be such a pain might not part of the project been to make it work without being a royal pain to operate? As I see it there is no component of a modern computer that hasn't been corrupted by Black Hat types and whats worse is no one seems to be doing anything useful about this problem. Get your bit coin ready to give away. Ransom ware is coming to your town.



I think the panel was interesting and its nice to see Ms Conway get some credit for her work. One thing that I think is missing from this that is never addressed is Wirth's law which states that software is getting slower more rapidly than hardware is becoming faster. Take a look at the speed of computers and how far we have come from the days of the 8080 or the Cray 1 Supercomputer. It staggers the mind how fast machines have become and how new abominations of software are regularly being released to run on operating systems bloated with mountains of abstraction layers, obscuficated code contests and so on. Its wasting all of this effort to cram more transistors into each new cpu. Its a great sales model to make each "new and improved" version of an operating system so slow as to be unusable to promote the sales of these higher speed machines but frankly I think people who are suckered into this are probably buying Ink Jet printers that are low cost except for the ink that runs out on a weekly basis.

[MageofLime]

This is an interesting video, but I think it should be indexed to the shift in value of the dollar over time as well. That $500M for system development now is equivalent to $76.94M in 1971 so that exponential is there, but it's exaggerated significantly by the effects of inflation.