• Skip to main content
  • Skip to search
  • Skip to footer
Cadence Home
  • This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  1. Blogs
  2. Breakfast Bytes
  3. 60 Years of DARPA—61 Actually
Paul McLellan
Paul McLellan

Community Member

Blog Activity
Options
  • Subscribe by email
  • More
  • Cancel
Aerospace
magestic
arpa
space
eri
darpa
chips

60 Years of DARPA—61 Actually

7 Feb 2019 • 5 minute read

 breakfast bytes logosputnikOn 4th August 1957, the Soviet Union launched the first artificial satellite, Sputnik 1 ("sputnik" means "satellite" in Russian). It wasn't very impressive by modern standards, a 23" metal sphere with 4 radio antennas. It seems that President Eisenhower and the US military knew about Russian space plans from spy plane overflights, so they were not surprised. However, the American public was shocked, and their assumption that the Soviet Union was a backwater, and the USA was a technological leader, was instantly revealed to be a myth. The US would go on to launch their own satellite, Explorer 1, in January of the following year, but by then the Russians had already launched Sputnik 2 in November. Worse, perhaps, was the very public (on TV) failure of the Vanguard TV3, which was meant to be the first successful US launch of a satellite in December 1957. It got a few feet off the ground, sank back, and exploded. It was obvious to everyone that the US was behind in the "space race".

The launch of Sputnik led to a new emphasis on what we now call STEM subjects in schools. In 1958 Congress provided low-interest loans for college tuition to anyone studying STEM subjects. The perception that the US was "behind" the USSR became an issue in the 1960 presidential election, and, obviously, in President Kennedy's famous 1961 speech, where he said that the country:

should commit itself to achieving the goal, before this decade is out, of landing a man on the Moon and returning him safely to the Earth

Later this year, on July 20th, it will be the 50th anniversary of the successful achievement of the first part of that goal, when Apollo 11's lunar module landed on the moon. On July 24th, when the mission successfully splashed down in the Pacific, the second part of the goal was achieved. I think I'm safe in predicting that you'll hear a lot about the moon landing this summer. Maybe even a blog post here.

ARPA

But Sputnik had another legacy. Today is the 51st anniversary of the creation of ARPA, the Advanced Projects Research Agency. It came into existence on February 7th, 1958, just 6 months after the launch of Sputnik. Initially, it encompassed space too, but when NASA got created later the same year, all responsibility for space (and most of the funding for ARPA) was transferred. ARPA was redirected to high-risk, high-gain, projects.

ARPA was the forerunner of today's DARPA (the added D is for Defense), although it went back to ARPA for a few years in the middle. ARPA created the Information Processing Techniques Laboratory (IPTL) in 1962 and hired J.C.R. Licklider to run it (see my post, "Lick" Licklider, Unsung Hero of US Computer Science for a lot more detail). IPTL funded the first doctoral programs in Computer Science in the US, which created a flow of expertise to feed industry and further academic programs as computing entered its high growth phase.

You probably know that the original name of the packet-switched network that was the forerunner of the Internet was called ARPANET. It was also part of the space race, in some ways, designed to be a network that could survive a nuclear war by routing around the damage in a way that the contemporary telephone and telegraph networks could not. Work started during the 1960s. The somewhat ignominious start was when:

1969: On Oct. 29, UCLA’s Network Measurement Center, Stanford Research Institute (SRI), University of California-Santa Barbara and University of Utah install nodes. The first message is "LO," which was an attempt by student Charles Kline to "LOGIN" to the SRI computer from the university. However, the message was unable to be completed because the SRI system crashed.

By about 1972, the network looked like in the picture above, from the Computer History Museum in Mountain View (for more about that, see my post, Computer History Museum History). Stanford is the node on the left, halfway down. I first used ARPANET when I was an undergraduate in 1975. There was no IMP (Interface Message Processor, what today we call a router) but there was a TIP (Terminal Interface Processor) at University College London where we had dialup access (or maybe it was a dedicated line, I forget). One of our elective courses was on computer algebra (think Mathworks or Mathematica) and the leading project at the time was REDUCE. The other was CAMAL (Cambridge Algebra), created by our instructor John Fitch for his Ph.D. to do celestial mechanics. We were in his office, using a terminal to run REDUCE in the US. Of course, none of us had the faintest idea just how much the internet would change everything in the decades to come.

DARPA's Electronic Resurgence Initiative

eri logoSo this time last year, why didn't I write about DARPA's anniversary? Because DARPA was doing very little in semiconductors, EDA, and design. However, last year, they launched the Electronic Resurgence Initiative. I attended the first annual ERI summit in San Francisco, and wrote about some of the programs DARPA (and Cadence) are involved with: 

  • Overview: The DARPA Electronic Resurgence Initiative (ERI)
  • Machine learning in EDA: Cadence Is MAGESTIC
  • Open source EDA flows: ERI: OpenROAD
  • Advanced packaging: ERI: CHIPS and Chiplets

The motivation for ERI is that the US Department of Defence has a number of problems with semiconductors. The three big ones are:

1) DoD used to be about 40% of the semiconductor market, but today it is more like 2%. So its influence on the direction of US the semiconductor industry has declined commensurately.

2) Any design the DoD does is dominated by the design cost since production volumes are so low. As Linton Salmon said in late 2017 (see my post, Open-Source IP in Government Electronics):

In DoD, I've yet to see a total production that is more than the samples that we would send out in the commercial world.

So it is important to reduce the cost of design and the cost of IP because that has to be amortized over a commercially insignificant production volume.

3) DoD doesn't have access to a lot of leading-edge technology in semiconductor and packaging since so much of it is now in Asia.

Happy Birthday, DARPA. You'll be eligible for social security next year.

 

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.