Being number 11 in the series

"What the up-timers don't know that they know"

Fr Nicholas Smithson SJ and Br Johann OSB

Number 11: On Computing, March 1633

A feature of the up-time world that many down-timers long for is their electronics. From radios to phonographs, from public address systems to telephones, from calculators to laptops, everyone longs for the things that come from tubes, transistors and integrated circuits.

In particular we long for the integrated circuit. By the early 1970s up-timer technology had matured past the thermionic valve (the vacuum tube) and the transistor to embrace ICs. Both general purpose ICs and "application specific" ICs were used in almost every device.

Why were ICs so popular with up-time engineers? Prior to the invention of the integrated circuit making complex devices from tubes and even individual transistors was difficult. Some applications required thousands of transistors to be hand-wired into circuits, with an equally large number of other components like resistors and capacitors. The work was time-consuming, error prone, costly and jeopardized reliability.

Another problem—what engineers called “the tyranny of numbers”—also existed. The sheer number of a system’s interconnected transistors and other devices prevented progress. Their size and weight often precluded their use in many devices. Prior to integrated circuits, commercially available computers filled rooms and required other rooms of equipment to provide cooling and power. If one component failed in those rooms full of parts, the entire system could be compromised.

Jack Kilby's invention in 1958 of the integrated circuit made obsolete the hand-soldering of thousands of components, while allowing for Henry Ford-style mass production. In particular, designers wishing to use digital logic for controls and computation were freed to build devices with thousands and even tens of thousands of circuit elements while decreasing size and cost and increasing reliability.

However, we have no illusions that the wonders of the IC age will come to us soon. The refining of single crystal silicon and germanium, the production of masks, circuit boards, and even standard resistors and capacitors are years away. For the next decade at least (and perhaps longer) we of the USE find ourselves in the tube era, and for most of that era tubes will be fragile, hand built and expensive. In regard to electronics, we find ourselves faced not simply with the tyranny of numbers, but the tyranny of scarcity.

This scarcity hits us hardest in field of computation.

Computationally intensive professions such as engineers, designers, architects and aviators, businesses like banking and retail and wholesale trade long for the ease of electronic calculation. The cheapest up-time giveaway four-function calculators are sold across Europe at prices which would ransom a Count. Slide-rules and mechanical calculators are a poor substitute at best, and down-time produced adding machines use scarce machine tools and materials and in any event as yet fare poorly against a skilled abacus operator. What is needed is a substitute for the transistor, and the integrated circuit; something that can be produced now in adequate numbers and with inexpensive resources.

We offer a possible path to a solution.

In the mid 1960s and into the 1970s, lead by researchers at Bendix corporation, a series of developments resulted in the entirety of computer logic elements being designed and built entirely out of unmoving grooves in a solid substrate through which a fluid was pumped. These "fluidic" devices depended on the "coanda effect" (the tendency for a moving fluid to cling to a surface).

An entire suite of fluidic devices was designed including all the critical elements of computer logic; NAND and NOR gates, flip flops, adders, shift registers and more.

While not the subject of this paper, the authors note that fluidic amplifiers were designed and built which could amplify sound cleanly and with reasonable precision with no electronics and no moving parts.

However, up-time, these technologies were a solution in search of a problem. Jack Kirby had already invented the integrated circuit, transistors were everywhere, small powerful batteries could be purchased in every corner market. What need had they for devices that depended on leaky messy streams of fluids, pumps and other complexities? Fluidics was consigned to odd corners of the up-timers' world in the nozzles of ink-jet printers, the control systems of expensive cars' windshield washers and custom built control systems for airplanes which would not die from an electromagnetic pulse during an atomic bomb attack.

No, the up-timers rightly let fluidics languish in the corners of their minds, eventually forgotten by the mass of up-timers.

This forgetfulness need not be the case in our world. Coanda effect devices are essentially two-dimensional and can be pressed with molds into wet clay. The first four-function electronic calculators used a few-hundred transistors. The equivalent device, a four-function calculator made of fluidic gates pressed into clay, powered by water can be made on a clay tablet less than a yard square. (if properly designed, this square yard of clay could be cut into nine parts and stacked into a 1 foot by 1 foot by six inch stack.) A "keyboard" of small valves would control input and output could be to a printer or a set of spinning wheels.

We choose to name this imaginary device the "Aqualator."

Logic diagrams for such a calculator, drawings of the required fluidic logic elements, pumps, indicators, and valves are available at the Grantville research center for the usual prices.

Looking farther ahead, the first "personal" computer, the Altair based on the Intel 8080 computer chip was approximately as complex as the first commercially available vacuum tube computer, the Univac. Both had between 5000 and 6000 logic elements (tubes or transistors). An Altair level fluidic computer of 6000 devices along with 1K of fluidic random access memory would, with some research and refinement require not more than two square yards of pressed clay. (These could be quartered and stacked into a cube little more than 1 foot on a side.) Expanding the ram to 16K can be done using a parallel system "buss" of piping to a second cubic foot of fired clay. The developers will, of course have the privilege of naming such a device once commercially available, but thinking of the Altair, we propose a similarly astronomical name with an aquatic bent: Aquarius.

The Aqualator calculator should run at speeds approximating those of up-time hand-held calculators. Essentially, "as fast as you push the buttons."

The Aquarius' clock speed is limited by the speed of sound in water, rather than the speed of light as in electronic devices. In order for all parts of the device to be in sync, the maximum clock speed has to be half-or-less of the time for a signal to travel from one end of the device to another. Sound, in water travels at 1500 meters per second, so, since the Aquarius is about a meter from one side to the other (following the channels inside the block) the clock frequency of the Aquarius will be 1500 Hz, or about 1/650 the speed of the Altair. A program which ran to completion on an Altair in 1 hour will require twenty-six days to complete. Switching to glycerin instead of water can increase the speed, and reduce the running time by 25%. An Aquarius running on molten zinc at approximately 800 °F almost doubles the clock speed and halves the run-time. The technical details of filtering and pumping molten zinc without exposure to air we leave to the engineers.

Optimization of code, and the implementation of parallel processing will be far more important to fluidic computer designers and programmers than it was up-time.

No complete logic diagram for a programmable computer came through the Ring of Fire, however all the elements of a computer are represented in various up-time documents. These have been collected into a single source document which is available at the Grantville research center at the usual prices.

Conclusion: Once again, a solution is presented to us near complete by the up-timers, unaware of what they brought us and the riches contained in it.

Respectfully submitted:

Nicholas Smithson, Society of Jesus

Johann, Order of Saint Benedict

References:

Popular Mechanics, July 1967, page 114+ http://bit.ly/cPTomX (Google Books)

Programmable Water, Paulo Blickstein http://www.blikstein.com/paulo/projects/project_water.html

A Description of Several Fluidic Breadboards Built to Investigate Transmission Information Over One Tube, Authors: Larry F. Zimmerman; NAVAL AVIONICS FACILITY INDIANAPOLIS IND August 1968.

AUTOMATION OF MACHINES BY MEANS OF PNEUMATIC LOGIC ELEMENTS (UTOMATIZACE STROJU PNEUMATICKYMI LOGICKYMI CLENY), Authors: Ondrej Brychta; FOREIGN TECHNOLOGY DIV WRIGHT-PATTERSON AFB OH, November 1967

Operation, Application and Production of Fluid Logic Elements and Amplifiers. Authors: R. W. Hatch Jr; FOREIGN TECHNOLOGY DIV WRIGHT-PATTERSON AFB OH September 1967.