These days the question I am asked the most often is whether to buy a double, triple, or a quadruple speed CD-ROM drive. Until recently, NEC was the only guy in town in the triple and quadruple speed drive business, and the answer was rather easy because of NEC's high prices - about $1,200 for the quad-speed drive. Now that Plextor has come out with a quad-speed demon for about half the price ($600 typical street price), and NEC has also come down from the clouds, the question is meaningful and needs some serious thought, especially since many users expect that these drives will make their applications run two, three, or four times faster.
Which Speed Really Matters?
The first generation CD-ROM drives ruled the waves from 1985 to 1992. Their access (seek) time kept improving, although there was a very wide performance range - from the 300 msec hare to the 1500 msec tortoise. Transfer time to pump the data from the CD-ROM to the processor was standard: 150 KB/sec that complied with the transfer rate of CD audio. It was an easy time to decide which drive to buy: the one with the lowest seek time and the best price tag.
Interestingly, the price was not always inversely proportional to the seek time. IBM believed that its label would sell its glacial drive for well beyond $1,000, while Texel (the former name of Plextor) sold its turbo for one-third of that price. No wonder that IBM had to lay off a few thousand people, and no wonder that Plextor has a booming business.
For the predominantly text-oriented databases, the average time needed to get from the outermost to the innermost data area of the disc was the decisive factor. Records that matched the user's query were scattered on the disc, and they had to be sought out and transferred to the screen. Shorter seek time produced better response time. The standard transfer rate of 150 KByte/sec was more than adequate. Even assuming the extreme case of a whole screenful of text (80*25 characters), this amounts to less than a sector of information (2K), and the CD-ROM drives could shoot over 75 sectors each second (if they were adjacent as in the case of full-text databases). Our own capability to scan the text displayed on the screen is far lower than that, so transfer rate was not a bottleneck.
Multispin, the Misnomer
All this simplicity vanished when NEC introduced the first double speed CD-ROM drive in 1992 under the trademarked name "Multispin." This was a misnomer from day one because every CD-ROM drive spins at different speeds ranging between 250-400 rpm (rotations per minute) depending on whether data is being read from an inside or an outside track, but it sounded good, and the spin doctors could get away with it at the Patent and Trademark Office.
Double speed is a better name because these drives can spin the disc at twice the maximum of the earlier ones. The first multimedia applications necessitated this increase because the video clips had been forced into the Procrustes bed of the 150 KB/sec transfer rate. What is plenty of bandwidth for text is as miserable for video as the eye of a needle for a camel to pass through.
You Can Run but You Can't Hide from
Broadcast quality video requires the display of 27 Megabytes each second. If you want to check the algorithm, it is as follows: 30 pictures (frames) must be shown each second for the illusion of smooth motion. A single picture consists of 307,200 (640*480) picture elements (pixels). Each pixel needs 3 bytes for true color quality. Therefore, thirty of these pictures for each second take 27 Mbytes. In addition, CD-audio quality sound requires 3 Mbytes per second. An entire CD-ROM disc could hold at best 650 Mbytes, barely enough for a 20 second video clip to start with. The 150 Kbyte/sec pipe is like using a straw to blow water on a burning oil tanker.
To push through 30 frames per second, each frame had to be reduced to 5 KB, which is a 200-to-1 compression ratio. …