We are on the verge of seeing a big leap in affordable processing power from inexpensive computers. Essentially, what was a $10 million Cray of a decade ago is now available for the teenage "gamer" ... just as the $10 million IBM 360 evolved into the PC, but this time it is qualitatively different. Virtual reality is on the horizon, as well as machine "learning" and human-like visual recognition.
http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-TITAN-Preview -GK110-GPU-Boost-20-Overclocking-and-GPGPU And closer to home for some vorticians, this massive level of computer power will be capable of controlling a (formerly) low level task - such as say, a small reactor - if needed. The obvious question is: why and how would you need it? DSPs and PICs and Arduinos are pretty cheap already; and they can control dozens of interlocking parameters without breaking into a sweat. Yes, the knee-jerk reaction is - you do not need a supercomputer to do the work of a DSP.... but don't forget that in the 1960s- many experts at IBM could not envision the need for the PC. An expert of today does not need to be a contrarian to suspect that maybe... just maybe... we will "invent the need". An answer in greater detail for that proposition (the emergent need for the cheap supercomputer) is likewise certainly not obvious. But the point is - like so many things in modern technology - the best application for a new device often blindly emerges (to the surprise of all), almost as an afterthought - following the introduction of the enabling product. This is the reverse of tradition where 'necessity' is the mother of invention. The major paradigm shift we are seeing nowadays in applied science is that invention is no longer "need driven" so much as opportunity driven. As they say in the flicks - "if you build it, they will come". Anyway back to the "moonshine" of a supercomputer controlling a gainful energy process... one immediate but general suggestion for how it would fit-in involves the so-called "Maxwell's demon" ... which is a smart device that can select molecules from the Boltzmann's tail of an energy distribution, and move them non-randomly- thereby deriving net energy from ambient conditions. An Arduino could probably control a few dozen I/O channels - but what if one seeks to control a few million? Yes that shifts the "invention" part of the equation to providing secondary sensor arrays - which are non-existent today but still ... "visual recognition" in the human context requires massive computer power, and this could be the initial use, especially if the computer is self-learning. There is another application for LENR, specifically (and the reason for this post) but it is based on the hypothesis of gain particularly in NiH reactions coming from the high end of the mass distribution for protons. I have not convinced many observers that this hypothesis is accurate (that hydrogen mass is not quantized, except as an ideal value like the Bohr atom) ... so it will be a hard sell to convince a VC or angel funder of the need to develop a supercomputer subsystem for optimizing gain from this hypothesis... but that may happen, quien sabe? We are reaching the tipping-point in the appreciation of the societal harm caused by fossil fuel, economic harm more so than climate change. In fact, the "appreciation of the threat" - may be the "necessity" which is the new mother of a two-tiered invention process, which also is co-driven by the new enabling technology, but at a level which in beyond serendipitous ... equal parts 'perspiration' and 'inspiration' but with the information-processor itself defining the major limitation. Jones
<<attachment: winmail.dat>>
