Organic Computers
Dreaming about future technology is often rooted in current trends, failures, or fantasy. It wasn’t long ago the thought of nano-bots was mere science fiction. Today, there are so many things we can do. The primary question anymore is how do we pay for it? How can we make a living? How can we produce and package it for the masses?
A current trend in computing is to move toward the Cloud, sharing massive resources with the "masses". Today David predicts the evolution of a Cloud on a massive scale, costing much less to operate than current technology.
He writes:
I’m more skeptical of biological computing as a total replacement for traditional semiconductor-based computing (and not just because I’m an EE by training ;). Solid-state electronics (the term for any electronics involving nonmoving parts, from vacuum tubes to resistors to transistors to the new memristors), by being solid, are far easier to maintain than any wetware ever could be.
When you do not need to use the hardware, you just stop giving it power, and you don’t have to worry about it for years (you just need to clear off any dust that may be creating a short before you turn it on next time). Wetware needs food, heat, waste disposal, and an uncontaminated environment, whether or not you’re using it. There are environmental and health safety issues that would surround the operation of a wetware system, as there would be no way to “seal” the operating device as an iPhone is sealed, and a dropped and broken wetware system is much more of a hazard than a dropped and broken iPhone.
Where wetware could shine is in the mainframe world. Stop giggling and hear me out: Bacteria are self-assembling, requiring minimal energy and food input to multiply into millions of cells. The expense to just grow the bacteria can be measured in cents per million cells. Of course, growing them into the proper alignment to produce things such as NAND gates (recently demonstrated by scientists at Imperial College London: http://zeenews.india.com/news/technology/biological-computers-come-closer-to-reality_737401.html ), but a modern semiconductor fabrication plant now costs on the order of $1 BILLION to construct (http://en.wikipedia.org/wiki/Semiconductor_fabrication_plant ).
If wetware can be scaled reliably (this is a big if), then a large facility operating under regulations similar to wastewater treatment plants could be constructed to farm out processing and storage for companies on the internet. A massive, biological data center for cloud computing.
The operating characteristics of such a wetware datacenter would probably be vastly different from a standard datacenter – storage and retrieval of data may very well outpace processing of data (as the storage could simply be done by triggering RNA in the bacteria to alter its internal concentration of K+ or Na- ions, and retrieval simply be the measuring of said concentration for an individual bacterium, while processing of said data must traverse several bacteria, with a chemical transition time between bacterial “gates” (on the order of seconds) rather than an electrical transition time (on the order of nanoseconds – a billion times faster). This wetware datacenter can make up for its slow processing time by easily scaling to literally trillions of parallel execution units per high-level operation.
In order for such a wetware datacenter with the characteristics I describe above to be used efficiently, I believe a new programming language in the vein of Java (code run on a virtual machine that automatically doles out the processing and storage to the underlying system) would have to be designed such that the data storage density and latency, as well as the data processing density and latency, must be accounted for by the virtual machine automatically based on the needs of the high-level constructs written by the programmer.
We are taking baby-steps in this direction with GPGPU programming and the abstracted cloud computing (like Microsoft Azure and Google App Engine, unlike Amazon EC2 and Rackspace Cloud), already, but the decision on which processing configuration and on which storage configuration still lies with the developer, and is more rigid (if a more easily parallelizable set of inputs are provided to a high-performance serial processing unit, or vice versa, non-optimal performance will result).
Conversely, quantum computing is more optimal for certain types of problems where there is a large amount of computation on a very small amount of data (cryptography, traveling salesman, etc), and an abstracted programming language that can configure its operation from semiconductor to wetware for certain use-cases should also be able to do the opposite.
Which researchers accomplish this, and what company(ies) they work for, and which companies buy which other companies, are trivial details to this more important advancement of science.
Tomorrow we move on to other topics. However, if you have ideas you’d like to share, feel free to send them to btaylor@sswug.org. I’ll be dropping them in as footnotes in future newsletters. I still have a few I think you’ll enjoy reading.
Cheers,
Ben
$$SWYNK$$
Featured White Paper(s)
All-At-Once Operations
Written by Itzik Ben-Gan with SolidQ
SQL supports a concept called all-at-onc… (read more)
Featured Script
AutomaticMechanismsCanBeSurprising.sql
Helper script for AutomaticMechanismsCanBeSurprising article… (read more)