In the early 1970s, the change from 16-bit to 32-bit architecture was a major watershed in computing. In the mid-1990s, the change from 32-bit to 64-bit architecture will be equally significant. The meaning of a 16-bit designation is that at any tick of the computer's clock it is addressing or dealing with 16 bits of information, or approximately 2 bytes, given 8 bits to create a byte. The 32-bit machine addresses twice as much at any one time; the 64-bit twice as much again. That doesn't mean that 32-bit technology will become obsolete overnight. Prices for new technologies usually are high for the first year or two after introduction, usually dropping dramatically after that. A number of 32-bit machines may continue to offer very good performance at very attractive prices. The question is, how to select the best 32-bit technology until 64-bit technology prices drop.
While it is generally true that a higher-bit rating means a faster machine, there are a number of other factors which must also be taken into consideration in determining the productivity of a machine clock speed among them. CPUs have an internal clock that determines their speed. flow quickly the clock "ticks" determines the speed of the CPU. So that the CPU and the other components of the computer can talk to each other properly, the motherboard also has a clock chip that runs at the same speed as the CPU's internal clock.
Clock speed is measured in millions of cycles-or ticks-per second. In both 386/33 and 486/33 computers, for example, the clock is "ticking" 33 million times per second. However, even though each is ticking at the same speed, the 386 CPU is doing less with each tick. Typically, the 386 takes two clock ticks to do what the 486 can do in a single clock tick. This explains why the 486s are so much faster and will be the mainstay of personal computing for the next several years. But 486-based PCs and workstations are only now displacing the 386 technology.
Intel, the largest chip maker, is now making or planning at least 35 versions of its 486 chip. competitor Cyrix, Inc., has at least three different chips that it also calls 486s, with more on the way. Advanced Micro Devices, Inc., another competitor, has recently introduced yet another line of 486 chips.
In chip terminology, the letter designations that are also part of each chip's name represent other significant differences. Letters as a part of chip names started with Intel's 386SX. Intel wanted to market a lower cost version of its 386 chip, so it came up with the 386SX, which, although still a 32-bit chip, communicated with the rest of the computer via a 16-bit bus or data path. The SX chips lack the built-in math co-processor that other Intel chips employ. Then, to differentiate the full-featured 'version of the 386, Intel started calling it the 386DX. The 486DX is a 32-bit chip that employs a 32-bit data path.
An SL designation means that the chip uses less power than its non-SL counterpart A 386SL, for example, is a low-power version of the 386. A "c" designation means that the chip uses an internal cache, but this designation is used only for 386 chips The Intel 486 chips have an internal cache but don't have a c in their name.
Most CPU chips operate at a given speed both internally and in their interaction with the rest of the computer. However some Intel chips are "clock-doubled"---the chip's internal speed is twice the speed at which the chip talks to the rest of the computer. It's done to allow a computer to use a faster chip without altering the configuration of the motherboard-thus reducing the cost. This is indicated by the designation DX2. For example, a 486/66 DX2 chip works in a motherboard made for a 486/33. To the computer, the chip appears to be a 486/33-that's the speed at which,. data goes in and out of the chip. Inside the chip, though, everything works at double the speed, 66 million ticks per second.
Intel also makes 486 Overdrive chips. These are also clock-doubled chips, but are installed by the user into a special socket on the motherboard.
Cyrix makes a number of chips to which it has given the designation 486, although they are only vaguely similar to Intel's 486 chips. The most confusing of these is the Cyrix 486DLC. It is built with the same pin arrangement as a 3865X chip, and, like the 386SX, interacts with the rest of the computer via a 16-bit data path. The chip uses a subset of Intel's 486SX microcode, the instructions hardwired into the cpu itself that control its operation. Cyrix justifies calling the chip a 486 because it uses 486 code, but its performance, due primarily to the 16-bit data bus, isn't the equal of a "real" 486 chip.
Cyrix also sells a 486DLC chip, which uses a 386DX pinout. This allows manufacturers to use the chip on modified 386 motherboards, which presumably will allow them to build cheaper machines. The 486DLC runs at 40 MHz, which might lead one to believe it would be faster than an Intel 486/33. It's not, though, largely because Cyrix only uses a 2-kilobyte internal cache as opposed to the 8-kilobyte internal cache in an Intel chip, and because the Cyrix chip lacks an internal math coprocessor.
The bottom line is, if one requires a powerful PC, the designation which represents the greatest potential productivity is one with an Intel chip numbered 486/66 DX2. Such a machine, when configured with 8MB of memory, a 170MB hard disk drive, 14 inch 800 x 600 color monitor, and one diskette drive now costs approximately $2,900. This is approximately $600 more than an 486SX/25 running at less than 40 percent the speed. A cautionary note, if the PC has a chip manufactured by other than Intel, make sure that there have not been compromises in the bus speed, absence of the math co-processor, or minimal internal cache.
For those who want simple word processing applications without windows, the 486SX/25 remains a reasonable choice. It probably will be usable and supported for at least another five years.