Watson in the Lab: Quoth the Stakeholder Forevermore
April 7, 2016
I read “Lawrence Livermore and IBM Collaborate to Build New Brain-Inspired Supercomputer.” The article reports that one of the US national labs and Big Blue are going to work together to do something with IBM’s neurosynaptic computer chip. I know. I know. IBM is not really into making chips anymore. I think it paid another company lots of money to take the fab business off IBM’s big blue hands.
Never mind, quoth the stakeholder.
The write up reports that the True North “platform”
will process the equivalent of 16 million neurons and 4 billion synapses and consume the energy equivalent of a hearing aid battery – a mere 2.5 watts of power.
I like the reference to nuclear weapons in the article. I used to work at Halliburton Nuclear in my salad days, and there are lots of calculations to perform when doing the nuclear stuff. Calculations are, in my experience, a lot better than doing lab experiments the Marie Curie muddled forward. Big computer capability is a useful capability.
According to the write up:
The [neuromorphic] technology represents a fundamental departure from computer design that has been prevalent for the past 70 years, and could be a powerful complement in the development of next-generation supercomputers able to perform at exascale speeds, 50 times (or two orders of magnitude) faster than today’s most advanced petaflop (quadrillion floating point operations per second) systems. Like the human brain, neurosynaptic systems require significantly less electrical power and volume.
This is not exactly a free ride. The write up points out:
Under terms of the $1 million contract, LLNL will receive a 16-chip TrueNorth system representing a total of 16 million neurons and 4 billion synapses. LLNL also will receive an end-to-end ecosystem to create and program energy-efficient machines that mimic the brain’s abilities for perception, action and cognition. The ecosystem consists of a simulator; a programming language; an integrated programming environment; a library of algorithms as well as applications; firmware; tools for composing neural networks for deep learning; a teaching curriculum; and cloud enablement.
One question: Who is paying whom? Is Livermore ponying up $1 million to get its informed hands on the “platform” or is IBM paying Livermore to take the chip and do a demonstration project.
The ambiguity in the write up is delicious. Another minor point is the cost of the support environment for the new platform. I understand the modest power draw, but perhaps there are other bits and pieces which gobble the Watts.
I recall a visit to Bell Labs.* During that visit, I saw a demo of what was then called holographic memory. The idea was that gizmos allowed data to be written to a holographic structure. The memory device was in a temperature controlled room and sat in a glass protected container. The room was mostly empty. After the demo, I asked one of the Bell wizards about the tidiness of the demo. He laughed and took me to a side door. Behind that door was a room filled with massive amounts of equipment. The point was that the demo looked sleek and lean. The gear required to pull off the demo was huge.
I recall that the scientist said, “The holographic part was easy. Making the system small is the challenge.”
Perhaps the neuromorphic chip has similar support equipment requirements.
I will let you know if I find out who is paying for the collaboration. I just love IBM. Watson, do you know who is paying for the collaboration?
——
* Bell Labs was one of the companies behind my ASIS Eagleton Award in the 1980s.
Stephen E Arnold, April 7, 2016
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
