Starship Computers

Symetric Multi-Processing to the Nth Power

Larry Garfield

Revision History
Revision 1 27 January 2001

Posted in the ED


Table of Contents

The Sources
Analysis
Conclusions

The Sources

In order to clear up many conceptions, misconceptions, knowledge, misknowledge, information, misinformation, and mister information that people have about Star Trek computer systems, I offer the following discussion of Star Trek computers, based on what we know of them from the canonical sources. First, let's list a few of those sources, specifically, several exerpts from the Star Trek Encyclopedia:

[ Isolinear optical chip.Sophisticated information storage and processing device used aboard 24th-century starships. Composed of linear memory crystal material, the isolinear chip came into general use around 2349, and was used for both information storage and data processing. Within larger computing devices, isolinear chips are often housed i n wall-mounted racks, permitting computer access to dozens, even hundreds of chips at once time. ("The Naked Now" [TNG]). Isolinear chips replaced older, less efficient duotronic enhancers aboard Federation starships around 2329. [Description of one episode's events omitted.]]

[ Linear memory crystal.Optically refractive material used in isolinear optical chips, a key part of optical data processing devices. The sentient nanites accidentally developed by Wesley Crusher in 2366 consumed linear memory crystal as nourishment, enabling their reproduction.]

[ Bio-neural gel pack.Component of a bio-neural computer system consisting of a flexible, liquid-tight package containing synthetic neural cells in [a] gelatinous organic medium. The gel packs were designed for convenient management of bio-neural circuitry, and could be easily swapped out as needed. The Starship Voyager initially carried 47 replacement gel packs in its inventory. Bio-neural processors were vulnerable to viral and other infections, which could adversely affect system performance. [Description of the VGR episode where that happened omitted.]]

[ Bio-neural circuitry.Advanced computer technology using synthetic neural cells for data processing. Bio-neural circuitry could organize and process complex information faster and more efficiently than traditional optical processors. The computer systems on the Starship Voyager used bio-neural gel packs to supplement its optical data network. ("Caretaker" [VGR]). In fact, the U.S.S. Voyager was the first Federation starship to have bio-neural fibers incorporated into its systems. ("State of Flux" [BGR]).]

Analysis

The first thing I would like to highlight is that Voyager did NOT use a bio-neural core, but used gel packs to supplement its traditional optical data network. Presumably, gel packs were used in high-processing systems where they would have the most impact. The fact that Voyager carried only 47 spares also supports the argument that they didn't use all that many of them.

The second thing to highlight is that isolinear chips are both processing and storage devices. That is, they are the CPU and RAM and Hard Disk all in one neat little package. The computer cores on the Enterprise-D (we're now switching over to discussing chapter 4 of the TNG Technical Manual, pages 49-53) are subspace-souped up clusters of thousands of isolinear chips spanning several decks, arranged in access to use racks. Each individual chip contains onboard nanoprocessors as well as the ability to store "2.15 kiloquads per chip in standard holographic format" [(TNG TM, page 53)]. When used in the main cores, the core's "subspace flux" increases each chip's processing speed by as much as 335%. (The overclocker's dream.)

380 quaditronic optical subprocessors are spread throughout the ship to aid in processing tasks and increase redundancy, and to act as backup for critical areas of the ship such as the bridge. What "quaditronic" means is known only to the Paramount accounting department (where the real creativity on the Star Trek set takes place). These subprocessors are presumably collections of isolinear chips, just like the main cores. All of the subprocessors and main cores are linked together through the Optical Data Network (ODN).

Alright, so what the heck does that all mean? I'm not sure if any of you are familiar with massive multi-processing, but in modern computers it is possible to design computers that instead of having one or two CPUs have hundreds or thousands of CPUs, all symmetrically multi-processing. Aside from the raw speed and number-crunching power (Quake III Arena must kick serious arse on such a machine), this also brings us into the realm of massive multi-processing. I am not certain of the gory details, but the basic way that we think about a computer changes when you get into such large scale multi-processing. When you have, in essence, hundreds of "binary brains" working together, funky things happen that make the computer much more powerful (and confusing) than simply adding up their combined megahertz would suggest.

Each isolinear chip contains one or more (the Tech Manual is unclear) nanoprocessors, which are simply processors that are really small. Presumably they are also several orders of magnitude faster than any chip Intel has ever produced. If there are multiple in each, then each chip is its own multiprocessing system. Now hook a few dozen of them together and have them work symmetrically, and you have a massively multi-processing bank. That's a subprocessor. Now hook a few thousand of those banks together, and overclock them (the "subspace flux" generated by the FTL elements in the main core) by over 300%. You now have a main computer core, massively multi-massively multi-processing. (No, that's not a typo.) Now connect 380 of these banks and three of these giant cores, using fiber optic cable several zillion times faster than the fiber trunks we have carrying data across the Pacific ocean. That's a Galaxy-class computer. (And you thought the Internet was complicated.)

So where does bio-goo fit into all of this? The Encyclopedia says that gel packs "supplement [the] optical data network." Presumably this means they are used in the subprocessors that are distributed throughout the ship, or possibly in place of the subprocessors, to handle highly complex tasks which are better done using a mini-nervous system rather than "traditional" isolinear/holographic processing.

It is never stated explicitly whether or not gel packs are capable of data storage, a key ability of isolinear chips. If they are not, then they can not be used to completely replace isolinear chips at all, as the computer would have no storage mechanism. We are told that the holodecks on an Intrepid-class starship use only isolinear systems (one of the early episodes, second or third I believe), so that may in fact be the case, although we cannot be certain. Although it is not suggested in the manuals, I would hypothesize that, just as modern RISC chips are fantastic at some tasks while CISC chips are far better for other tasks, gel packs are not suited for every situation, but work best in certain types of applications.

Conclusions

Where does that leave ship designers, then? I would submit, based upon the preceding evidence the following guidelines:

As a side note, I recently read the book "The Computers of Star Trek" by Lois Gresh & Robert Weinberg. My advise would be to save your money. Although I have met Mr. Weinberg and go to school with his son, he missed the mark completely with this book. It is page after page of strawman arguments, that make me genuinely question if the authors actually read the Tech Manual as they claim. Some of the statements regarding modern security algorithms are interesting, but the rest of the book is continued statements of why Trek computers suck, and are backwards, and it doesn't matter because within 5 years we'll all have so many nanochips implanted in our bodies that we will have become Borg anyway. Don't bother, it's not worth your time.