Cryptonomicon

If you lay a sheet of white paper on an old gravestone, and sweep the tip of a pencil across it, you get one horizontal line, dark in some places and faint in others, and not very meaningful. If you move downwards on the page by a small distance, a single pencil-line-width, and repeat, an image begins to emerge. The process of working your way down the page in a series of horizontal sweeps is what a nerd would call raster-scanning, or just rastering. With a conventional video monitor--a cathode-ray tube--the electron beam physically rasters down the glass something like sixty to eighty times a second. In the case of a laptop screen like Randy's, there is no physical scanning; the individual pixels are turned on or off directly. But still a scanning process is taking place; what's being scanned and made manifest on the screen is a region of the computer's memory called the screen buffer. The contents of the screen buffer have to be slapped up onto the screen sixty to eighty times every second or else (1) the screen flickers and (2) the images move jerkily.

The way that the computer talks to you is not by controlling the screen directly but rather by manipulating the bits contained in that buffer, secure in the knowledge that other subsystems inside the machine handle the drudge work of pipelining that information onto the actual, physical screen. Sixty to eighty times a second, the video system says shit! time to refresh the screen again, and goes to the beginning of the screen buffer--which is just a particular hunk of memory, remember--and it reads the first few bytes, which dictate what color the pixel in the upper left-hand corner of the screen is supposed to be. This information is sent on down the line to whatever is actually refreshing the screen, whether it's a scanning electron beam or some laptop-style system for directly controlling the pixels. Then the next few bytes are read, typically for the pixel just to the right of that first one, and so on all the way to the right edge of the screen. That draws the first line of the grave-rubbing.

Since the right edge of the screen has now been reached, there are no more pixels off in that direction. It is implicit that the next bytes read from memory will be for the leftmost pixel in the second raster-line down from the top. If this is a cathode-ray tube type of screen, we have a little timing problem here in that the electron beam is currently at the right edge of the screen and now it's being asked to draw a pixel at the left edge. It has to move back. This takes a little while-not long, but much longer than the interval of time between drawing two pixels that are cheek-by-jowl. This pause is called the horizontal retrace interval.Another one will occur at the end of every other line until the rastering has proceeded to the last pixel at the bottom right-hand corner of the screen and completed a single grave-rubbing. But then it's time to begin the process all over again, and so the electron beam (if there is one) has to jump diagonally all the way up to the upper left-hand pixel. This also takes a little while and is called the vertical retrace interval.

These issues all stem from inherent physical limitations of sweeping electron beams through space in a cathode-ray tube, and basically disappear in the case of a laptop screen like the one Tom Howard has set up a few inches in front of Pekka, on the other side of that wall. But the video timing of a laptop screen is still patterned after that of a cathode-ray tube screen anyway. (This is simply because the old technology is universally understood by those who need to understand it, and it works well, and all kinds of electronic and software technology has been built and tested to work within that framework, and why mess with success, especially when your profit margins are so small that they can only be detected by using techniques from quantum mechanics, and any glitches vis-à-vis compatibility with old stuff will send your company straight into the toilet.)