The Complete Guide to Computer Aviation
by Steve Smith
PROCESSOR SPEED IS EVERYTHING
There's only one hardware rule: forget everything you've ever heard—go for the computer with the fastest processor speed you can afford (with a couple of exceptions, noted below).
Simulations work your computer's resources harder than just about anything else a computer can do, except maybe real-time fluid dynamics modeling. Unlike crunching numbers, where your computer finishes its work and then sits idling, waiting for your next command, simulations keep running until you turn them off. Simulations can also eat up huge amounts of real estate on your hard disk and can demand the highest graphics levels your video setup can provide…and sit up and beg for more.
Consider what your computer's CPU (Central Processing Unit; the main chip) must do. For a typical flight simulation program, your CPU must calculate your aircraft's exact location, altitude, speed, direction, and attitude (which way it's aimed; not necessarily the same way it's headed). With most simulations, the processor must also calculate these same factors for every other moving object within about a thirty-mile radius, not to mention rendering every stationary object you can see, and juggle all this data in short-term memory (RAM).
The CPU usually also runs the video display—accurately putting up to one million individually colored dots on the screen—as well as handing requests from the keyboard, the joystick, the sound board, and the mouse.
Finally, the CPU must recalculate all of this at least thirty times every second (or up to sixty times per second in the case of those million or so little dots). Whew!
The CPU in DOS-compatible computers is usually one of the Intel company's "80x" series (or clones): the 8088 went into the first IBM PC, the 80286 powered the second-generation IBM AT, Compaq was first PC with the 80386CPU, which has in turn been superseded by the 80486. With chip cloners like AMD and Cyrix having coopted this numerology, Intel decided against "80586" for its next-generation CPU in favor of the trademarkable "Pentium."
The performance increases over the ten-year development of these CPUs have been so accelerated that if the same quantum leaps were applied to air travel, we would sit on airplanes with twenty-five thousand other passengers and fly to Japan in about ten minutes. The fare: fifteen bucks.
These improvements have been not only to the chip's power (what it can do—like add, subtract, multiply, and divide) but also how fast it can do it. The most commonly accepted bench mark for this rate of work is "clock speed": an internal quartz clock determines exactly how fast your CPU will run.
The original 8088 ran at 4.77 megahertz (MHz); that is, the clock's quartz timer oscillates 4,770,000 times per second. During each clock "cycle," the CPU can execute only so many instructions ("Add this value to that value," etc.).
The first 286 chips raised the ante to 6 and then 8 MHz, touching off a horsepower race. By now, 386 chips can run at 40 MHz, and, with various "tricks," the 486 can barrel down the highway at 100 MHz. One begins to understand that ten-minute flight to Tokyo.
So, all other factors being equal, a 20-MHz CPU will beat a 16-MHz CPU every time, never mind if it's a 286 or a 386 or an SX or a DX or an SLC (suffixes that indicate different variations of the chip). The most dramatic exception to this rule is Intel's 486, which, thanks to its internal modifications, can suck the gallium arsenide off any other chip—except one; see below—at the same clock speed. The slowest 486, at 25 MHz, for example, is faster than a 33-MHz 386. The other exception is the Pentium, which at its slowest is faster than all the other chips mentioned above.
Why Speed Matters
Who needs speed, you might well ask, if all you're going to do is putter around the sky in a Piper Cub or a Sopwith Camel? The answer: the right hardware is as much about smoothness as it is about speed. And the more you have of both, the more you're in control.
The explanation of why this is so is slightly more complex.
As with a movie, the appearance of motion on your computer screen is an illusion—each picture is a still life. But when you look at a rapid sequence of images, they look as if they're moving…provided the screen is updated (it's called the "frame rate") more than about eighteen times per second. Below that rate, the pictures (that is, frames) are perceived as individual, flickering images. Above it, the frames meld into one seemingly continuous "motion picture."
Movies run at twenty-four frames per second, TV at thirty. Most fully animated computer programs don't have a fixed frame rate—it may take one twenty-fourth or one-sixtieth or some other fraction of a second (depending on how complex the image is) for the CPU to fetch a new screenful from the data-stream and forward it to the video circuitry, which then "paints" it on the screen. But if the result looks like continuous movement rather than a stuttering series of individual images, it's called "full-motion video."
Okay, so a higher frame rate looks smoother; what's that got to do with control? Exaggerate the numbers and you'll see what I mean. Imagine that instead of thirty new images every second, you saw only one new image every hour. If you're ferrying an Avro Vulcan across the South Atlantic in a straight line at forty thousand feet, this wouldn't be a problem … at first. Every hour looks pretty much like the last. You might notice the clouds had shifted position, but that's about it.
Now imagine it's time to land at Wideawake Airport on tiny Ascension Island. Instead of making control decisions every sixty minutes, you have to start making them every second, and then every split second, and, at the moment of touchdown, you'd want as many fresh updates as your system can provide (although anything over thirty frames per second is overkill; your eye can't process the information any faster).
Trying to make do with a slow frame rate is like steering an ocean liner: you crank in a lot of rudder, and—by and by—the ship starts to come about. But if you cranked in too much rudder, you won't know that for quite a while. By the time you've dialed in some countersteering, you may have overcorrected, but you won't know that for some time either. The faster the action unfolds, the faster the response you need to avoid overcontrolling.
So a really hot CPU not only looks smooth, but it enables you to outfly the other guy in a dogfight. Maybe that doesn't seem important to you now, but crash and burn a few times and you'd sell your soul for a few more megahertz.
What's this going to cost? The price of a "bare-bones" high-end system unit (the chassis that contains the CPU, its support chips, the power supply, etc.; the box itself is sometimes referred to—incorrectly—as the CPU) has been dropping like a rock since 1991. As this book goes to press, you can buy a bare-bones 486 for under $1,000. The next-generation CPU, the Pentium, was initially priced over the moon, but the inevitable price cuts have brought the price of a basic Pentium down to within hailing distance of two grand.
A word of warning: various CPU cloners claim they have replacement chips that will "upgrade" your old 386 to a 486. If this strikes you—as it does me—as too good to be true, think about upgrading your motherboard to a real 486.