Fourth Generation Processors

The fourth generation of processors saw rapid growth in the CPUs' power and capabilities, and the introduction of several new technologies as well. It was here that AMD and Cyrix had their first real early successes in not just emulating Intel's designs but coming up with real value leaders of their own, at the top end of the fourth generation scale. Also, it was at this time that the new phenomenon of upgradable processors and standardized motherboard sockets began to become prevalent.

Intel 80486DX

The Intel 80486DX was the first member of the 486 family (which has many more members than the 386 family did). It provided a very significant increase in power over the 80386DX processor, in fact, far more proportionately than the 386 did over the 286. A 486DX processor provides approximately 100 to 150% more performance than a 386DX of the same clock speed. The 80486 brought GUIs to the mainstream on PCs; it is the minimum processor that most people consider "usable" for running an operating system like Microsoft Windows.

Interestingly, the 486 does not provide its performance improvements by widening any of the buses, as had been the case in the previous two generations: it is still a 32-bit processor with 32-bit data and address buses, just like the 386DX. However, internally, the 486 incorporates several significant improvements over the 386:

• Faster Overall Instruction Execution time: The core of the chip can execute instructions in less time than earlier processors.
• Deeper Pipeline: The execution pipeline was increased by one step.
• Primary Cache: The 486 processor was the first to incorporate level 1 cache on the chip, to reduce the number of required accesses to memory.
• Integrated Floating Point Unit: The chip includes an integrated math coprocessor (not on the SX version however). In addition, the coprocessor provides much more performance than the optional 80387 used with 386 chips, in part because it is integrated into the chip.
• Burst Mode: The 486 introduced the use of burst mode to reduce wait time on memory accesses.
• Power Management: SL power management enhancements as an option (instead of a specialized SL chip as was the case for the 386SL).
• Improved Support Architecture: In general, 486 motherboards were more efficient than 386 ones, and began to use secondary cache as well. This improves processor performance significantly.

Interestingly, the 80487SX coprocessor that is intended for use with the SX version of the 80486, is in fact a full-blown 80486DX processor. This is part of how Intel structured its fourth generation family's upgrade path. See the description of the 80486SX for more details on this.

AMD and Cyrix did produce clones of the original Intel 80486 processors, but they were not a big player in the 486 clone market until the higher-speed DX2 and DX4 processors.

The 486DX is considered obsolete, although the chip still has fairly good power for performing a wide variety of light tasks, such as word processing and some older games, and light Internet access. The 486DX-50, which runs on a 50 MHz system bus, provides performance comparable to the 486DX2-66 in many ways, because the latter uses only a 33 MHz system bus. The 486DX-50 was not used in nearly as many systems as the other processor speeds were. It should not be confused with the 486DX2-50, which runs at the same processor clock speed but is clock-doubled relative to the system bus (which runs at 25 MHz).

Note: The 486DX processor normally was purchased as part of a new system only, not as part of an upgrade. Most early 486 systems used a 168-pin socket for the chip, which predates the numbered standardized socket system that Intel created. The 486DX will fit into a Socket 1, Socket 2 or Socket 3 however.

Look here for an explanation of the categories in the processor summary table below, including links to more detailed explanations.

Intel 80486SX

The Intel 80486SX is the same chip as the 80486DX with one exception: the lack of an integrated math coprocessor (floating point unit). Note that this is a different kind of difference between the SX and DX versions than is the case with the 386; the 386SX and 386DX both had no coprocessor and the 386SX had narrower data and address buses than the 386DX. The 486DX and 486SX have the same bus widths. Since it is the same chip except for the floating point processor, the 486SX has the same advantages over the 386 that the 486DX does. Note that the 486SX was made available in slower clock speeds than the 486DX; the SX comes in 16, 20, 25 and 33 MHz versions, while the DX is 25, 33 and 50 MHz.

Intel made some rather interesting (some would say bizarre) decisions about how to market the 486 line. The 486SX was in many ways a marketing gimmick only; it was in fact a 486DX with the math coprocessor disabled! Why did Intel bother to sell a full 486DX and a "crippled" one (the 486SX)? Mainly for market targeting reasons: they wanted to be able to fill the niche of low-end buyers without dropping the price on their "top of the line" 486DX. The 486SX was a popular choice for laptops due to its lower cost and lower power consumption.

The 80486SX uses a "math coprocessor" called the 80487SX (there is no 80487DX at all so don't be confused by that). The 80487SX is, in fact, a fully functional 80486DX chip! What Intel wanted people to think was that (like with its earlier coprocessors) you would put the 80487SX in and it would handle the math functions. In fact, when inserted, the 80487SX shuts down the 80486SX and handles both integer and floating point operations (since it is internally a 80486DX, which does both). This makes no difference from a performance standpoint but is kind of a technical curiosity.

With the large number of very cheap 486DX and faster chips and systems around, the 486SX is really quite obsolete. In addition, it does not come in a 50 MHz version. If you are not doing any floating point operations at all it is comparable to a 486DX of the same speed, but there is usually no need to have to worry about this given that even the 486DX is quite obsolete at this time anyway.

Note: The 486SX processor normally was purchased as part of a new system only, not as part of an upgrade. Most early 486 systems used a 168-pin socket for the chip, which predates the numbered standardized socket system that Intel created. The 486SX will fit into a Socket 1, Socket 2 or Socket 3 however.

Look here for an explanation of the categories in the processor summary table below, including links to more detailed explanations.

Intel 80486DX2 and 80486DX2 OverDrive

The 80486DX2 was the first chip to use "clock doubling" technology, where the processor runs at a faster speed than the memory bus it talks to. This was done to allow the processor speed to be increased without having to deal with the much more difficult task of increasing motherboard speed. Chips that run at faster than memory bus speed improve performance but at a diminishing rate as the multiplier increases, due to the processor waiting for data from memory. This is discussed in detail here.

Intel produced 50 and 66 MHz DX2 chips, intended for use in 25 and 33 MHz system bus systems. These chips have been sold as regular chips intended for use in new systems, which generally come in 168 pin packages to go in the original 168 pin socket used in 486 systems. They have also been made in 169 pin OverDrive versions to go in Socket 1 (the original OverDrive socket). These can be used to upgrade older 486DX or 486SX systems.

AMD and Cyrix not only cloned the 66 MHz DX2 processor, they took Intel one step further with the 80486DX2-80, running at 80 MHz. This uses a 40 MHz system bus, which isn't a speed that is normally used by Intel systems but that became more popular late in the 486 life cycle due to the performance increase it gives over 33 MHz bus systems. In addition, the AMD (enhanced version) and Cyrix chips have several advantages over the Intel chips (they had the benefit of developing them well after Intel):

• Power Management: They have SMM (power management) built in.
• Write-Back Cache: They have write-back capability for the primary cache which provides a small boost in performance.
• Lower Power: They use less power because they run at 3 volts, but will still work in standard 5 volt motherboard sockets without a converter or regulator (they are said to be "5 volt tolerant").

Other than clock speed, the 80486DX2 is virtually identical to the 80486DX. They are obsolete due to the availability of faster, very inexpensive processors such as the 5x86-133 that go in the same sockets, but 486DX2 systems are perfectly viable for many uses, including routine office word processing and spreadsheet work under DOS and Windows 3.x. The 80486DX2-66 is by far the most common version of this chip; a great number of these systems were produced and many are still in use today, especially in small businesses.

Note: The 486DX2 was the first processor to really require a heat sink in order to operate reliably. The increased speed of this chip means that it runs very hot (at least the Intel 5 volt versions).

Look here for an explanation of the categories in the processor summary table below, including links to more detailed explanations.

Intel 80486DX4 and 80486DX4 OverDrive

The 80486DX4 continued the trend started by the 80486DX2 toward faster clock speed processors on slower motherboards. The DX4 uses "clock tripling", where the processor runs at three times the speed of the memory bus (motherboard). Note that despite the name the DX4 does not run at four times the memory bus speed ("DX3" refers to 2.5 times the memory bus, but never became a shipping product for the 486.)

Intel's DX4 runs at two speeds: 75 MHz (for the 25 MHz bus) and 100 MHz (for the 33 MHz), with the 100 being by far the most popular. In order to keep power and heat to a manageable level the voltage of these chips is reduced to 3.3 volts. The AMD and Cyrix versions are very similar to their DX2 cousins, in that they are 5 volt tolerant, meaning they can handle being put in a 5 volt motherboard; the Intel chip is not 5 volt tolerant and cannot be put into a 5 volt board (well, you can physically put them into the board, but you shouldn't do it! :) ). AMD and Cyrix also make a 120 MHz version that runs on a 40 MHz bus; this is relatively rare. The Intel, AMD and Cyrix versions of the DX4 chip are quite similar. One difference is that the AMD and Cyrix support write-back cache while the Intel does not; however the Intel has its level 1 cache doubled to 16 KB. All three chips support power management.

The 486DX4 was most commonly put into systems as OverDrive processors for older, slower systems. Intel did not originally make the DX4 available as a stand-alone chip, presumably since it shipped after the Pentium had already been introduced. However, they did make the chip available stand-alone later on. New 486DX4-100 systems were often made with the AMD version of the chip. The OverDrive version is generally required for older motherboards that don't support 3.3 volt power.

Like the 80486DX2, the 80486DX4 is technically obsolete, but the 100 MHz version especially has good processing power for routine applications. It is also acceptable for some other uses, but is not found in modern systems.

Look here for an explanation of the categories in the processor summary table below, including links to more detailed explanations.

AMD 5x86 (80486DX5)

The 486DX4-100 was the fastest 486 processor made by Intel before they decided to leave the fourth generation and concentrate on the Pentium. AMD took clock multiplying one increment further with what it calls the "5x86" chip. Despite the implication that it is a fifth-generation chip, it is not--it is a high speed, clock-quadrupled 486 processor, that runs in 486 motherboards. It does use a rather advanced 0.35 micron CMOS process (advanced for a fourth-generation chip that is).

The AMD 5x86 was made available in one speed only, 133 MHz, for use in 33 MHz motherboards. The processor runs at four times the system clock and fits into a Socket 3. Since having a clock multiplier of four was not part of the original Socket 3 design, AMD made the 5x86 look for a two times setting from the motherboard and interpret that as four times instead. In other words, to use the 5x86 you want to set the motherboard to the 2x setting. This will actually cause the 5x86 to run at 4x. The chip will actually physically fit into an older 486 socket such as a socket 1 or 2 or the original 168-pin 486 socket, but doing this requires a voltage regulator since the AMD chip runs at 3.3 volts.

AMD calls this chip the "5x86-P75" because it offers performance comparable to low-end fifth-generation chips. In fact, it is comparable in integer performance to a 75 MHz Pentium. It is also called the 80486DX5-133, which is a more accurate reflection of what the chip really is. Architecturally, the chip is virtually identical to the 80486DX4, except that it matches Intel's 16 KB level 1 cache (the AMD 80486DX4 has only 8 KB). Note that this is not the case with Cyrix's 5x86 chip, which is quite different.

The 5x86-133 is the most powerful 486-class chip available (the Cyrix 5x86-120 is actually faster but was discontinued by Cyrix many months before AMD stopped mass-producing the 5x86-133). The 5x86-133 has traditionally been an excellent choice for an economy PC, especially for home-builders, because not only is the chip itself very inexpensive, so are the motherboards that use it. With 200+ MHz Pentium and Pentium-compatible chips coming well down in price, the 5x86-133 is following the rest of the fourth-generation chips to obsolescence, quickly. However, since it is the "king of the hill" for 486 motherboards, it remains a good choice for those who want to upgrade their 486 systems without replacing the motherboard. The chips are still out there, though they are getting much more difficult to find now.

Look here for an explanation of the categories in the processor summary table below, including links to more detailed explanations.

Cyrix 5x86 ("M1sc")

Despite having the same name as AMD's 5x86 processor, the Cyrix 5x86 is a totally different animal. While AMD designed its 5x86 by further increasing the clock on the 486DX4, Cyrix took the opposite approach by modifying its M1 processor core (used for the 6x86 processor) to make a "lite" version to work on 486 motherboards. As such, the Cyrix 5x86 in some ways resembles a Pentium OverDrive (which is a Pentium core modified to work in a 486 motherboard) internally more than it resembles the AMD 5x86. This chip is probably the hardest to classify as either fourth or fifth generation.

The 5x86 employs several architectural features that are normally found only in fifth-generation designs. The pipeline is extended to six stages, and the internal architecture is 64 bits wide. It has a larger (16 KB) primary cache than the 486DX4 chip. It uses branch prediction to improve performance.

The 5x86 was available in two speeds, 100 and 120 MHz. The 5x86-120 is the most powerful chip that will run in a 486 motherboard--it offers performance comparable to a Pentium 90 or 100. The 5x86 is still a clock-tripled design, so it runs in 33 and 40 MHz motherboards. (The 100 MHz version will actually run at 50x2 as well, but normally was run at 33 MHz.) It is a 3 volt design and is intended for a Socket 3 motherboard. It will run in an earlier 486 socket if a voltage regulator is used. I have heard that some motherboards will not run this chip properly so you may need to check with Cyrix if trying to use this chip in an older board. These chips have been discontinued by Cyrix but are still good performers, and for those with a compatible motherboard, as good as you can get. Unfortunately, they are extremely difficult to find now.

Look here for an explanation of the categories in the processor summary table below, including links to more detailed explanations.