1. What's a processor?
   
   If you know the basics, skip this part, but I want to make sure all bases are covered. The processor is the mind of your computer. It does your computation, it rallies the "soldiers" as it were, your video card, your peripherals, your sound... The speed and architecture of your processor basically defines how your machine is going to run. It's a chip inside your machine, plugged into your motherboard.
   
    
2. What do I need to know about processors?
   
   The most important thing to think of is: what do I want my machine to do? The biggest mistake many people make is spending too much money on a machine, because they think faster is always better. Well, it is, but is it always worth it? If you want to surf the web, or do spreadsheets and word processing, you'll be perfectly happy with a "middle of the road" processor, around 400 to 500 mhz or less. If you play lots of 3D games or other demanding applications, and have the money to spare, then go all out.
   
   One important thing to remember is that the top of the line processors are ALWAYS overpriced. After a top of the line processor is released, it usually drops $100 a week in price for about a month.
   
   That being said, if you're buying a processor, there are two things to consider. If you are upgrading an existing system, you have to buy a new processor that fits your current architecture. Processors come in "slot" and "socket" architecture, and their bus speed also determines which motherboard they can go in. If this lingo confuses you, never fear, there's a vocabulary section below.
   
   If you are building a new system, you simply need to make sure that the motherboard and processor you purchase are compatible.
   
    
3. Vocabulary?
   
   Mhz- Megahertz. One hertz means one processor cycle per second, so 500mhz means 500,000,000 processor cycles per second. It's simply a measurement of speed.
   
   Bus Speed- The bus is the part of the processor that talks to the motherboard. The most important part of the bus for our purposes is the speed. This determines how fast the motherboard is able to speak to the processor, in mhz. Each cycle, information is sent between the processor and the motherboard. You'll want to make sure that the motherboard you buy accepts the bus speed of your processor. The bus speeds that I will reference in this FAQ are only FSB, or Front Side Bus speeds. This is the speed that the processor talks to the motherboard. Sometimes people also talk about the speed that the processor talks to the cache, but I reference that as cache speed.
   
   Slot/Socket Architecture- Your processor is shaped like a flat square. If this square is laid flat while plugged into the motherboard, it's a Socket style chip. Pins come out of the bottom and plug into a square socket on the motherboard. If it stands perpendicular to the motherboard, it's a Slot style chip. The bus interface is on the side, and you plug it into a slot on the motherboard.
   
   L1/L2 Cache- Your computer is kind of like a human, in that it has long term and short term memory. Your short term memory is your RAM. While a processor is crunching numbers, it needs to store things in temporary memory, like RAM. The problem is that RAM is (relatively speaking) terribly slow. L1 and L2 cache refer to smaller chunks of very fast memory used by the processor that is either mounted very close to, or (more recently) directly on the processor itself. Most of the time people talk about L2 cache, as it varies the most. Half speed cache means it runs at half the speed of the processor, and full speed means it runs as fast as the processor. On die means that the cache is on the same piece of silicon as the processor, and is faster than off die cache. L1 and L2 simply stand for Level 1 and Level 2 cache. L1 cache is ALWAYS on the chip, L2 cache is sometimes on the chip, sometimes on the motherboard.
   
   ____ Core- Each time a new processor design comes out, it is given a name. Lots of times this code name is used to discuss the processor, because it will span several different processors. For example, the Dechutes core by Intel was used on both the Pentium II and the Celeron.
   
   .xx micron- Processors are etched into wafers of silicon, and the smaller linewidth they can achieve inside the processor, the smaller it will be. Smaller generally means it takes less power, and goes faster. The micron linewidth is often listed with processors.
   
   PPGA - Another name for Celeron's socket 370 interface. Stands for Plastic Pin Grid Array.
   
   FSB - Front Side Bus. Simply referring to the bus speed of the chip.
   
    
4. Intel Older Chips?
   
   Intel made the 8086,286,386 and 486 before making the Pentium family of chips. These aren't particularly interesting except in a historical context. The Pentium isn't being made anymore, and the Pentium II was just taken out of production at the end of 1999. However, a vast number of people are still using these chips, and they are still quite useful, and available for sale.
   
   The Pentium processor came in speeds of 60mhz to 200mhz. They upgraded Pentium to Pentium MMX, and released processors from 133mhz to 233mhz. The original Pentiums running at 60Mhz and 66Mhz used a Socket 4, and then Pentiums 75Mhz through 133Mhz used Socket 5. You really won't find these around too much anymore. The Pentiums and Pentium MMX 150Mhz and above use the now standard Socket 7. Socket 7 is also what you use for Cyrix, AMD, and other non Intel processors up to 233mhz. Intel stopped making Pentium chips at 233Mhz, and that is the highest processor speed that the Socket 7 supports. There is a newer standard called Super7, which supports the non-Intel processors that are being made. So all AMD K6/K6-2, Cyrix, etc. chips that are >233Mhz use the Super7, which is just a Socket 7 that can take higher speed processors.
   
   Intel also released the Pentium Pro around this time, which was for more industrial use. It was released in speeds of 133mhz to 200mhz, came with L2 caches from 256k to 1meg. Astonishingly, Pentium Pros still sell for more than most Celeron and even Pentium III's, because there is a demand by people who bought dual PPro motherboards, and now want to fill the second slot, but the chip is no longer produced. They required a new bus, the Socket 8. An interesting fact is that the Socket 8 has the same pins as the Slot 1, so you can plug Pentium Pros into Slot 1 motherboards with a special adapter. Intel has a pattern of releasing chips in "standard" and "industrial" versions. In the Pentium II and III family, the industrial version is the Pentium Xeon.
    
5. Celeron?
   
   The Celeron is an interesting, and widely misunderstood processor. First, it was basically a PII with no L2 cache. This is the "Dechutes" core. Dechutes was released in 266mhz and 300mhz versions, both running at 66mhz on Slot 1 architecture, and at .25 micron process technology. This lack of L2 cache is what caused the Celeron to have a bad name, and to be considered 'cheap'.
   
   In mid 98, Intel released the Mendocino core Celeron, or the Celeron A. This processor has a 128k on-die full speed L2 cache, which makes it run just about exactly as well as a PII processor at 66mhz FSB with 512k external L2 cache. These are released in 300A(to distinguish from the Dechutes model), 333, 350, 400, 433, 466, 500, and 533 mhz models. These processors are also at .25 microns. The Mendocino Celerons are available both in Slot 1 and the newer Socket 370 format. (Not compatible with the Socket 7 architecture, because it has an on die L2 cache).
   
   Intel is now releasing the Coppermine-128 core Celerons. This core is only .18 microns as opposed to .25 of Mendocino. It's named Coppermine-128 because of the 128k on-die full speed cache. Every Celeron released after 533 is going to be Coppermine-128. Strangely, they will continue at 66mhz bus speed.
   
   One thing to note about the Celerons is that they are great for overclocking, but there is no thermal plate or cover on the processor, so you must cool well if you are planning on doing this. Many people have had success with taking a Celeron 300A, putting it on a board with 100mhz FSB, and thereby increasing the processor speed to 450Mhz. But, before overclocking you should be aware of the many possible hazards of increasing the clock speed of your CPU or system bus. This FAQ isn't responsible to any damage due to overclocking. Please see the following overclocking section for more detail about overclocking.
    
6. Pentium II
   
   The first PII's of interest are those of the Klamath core. (233,266,300mhz) This was what was originally released, and are still available, but were quickly outdated by the newer Dechutes core. The Klamath core is based on .35 micron process technology, and runs at 2.2 Volts, which makes it hotter than the smaller Dechutes core. Like all of the PII family, it has an external half speed 512k L2 cache. The Klamath core runs at 66Mhz FSB, on the Slot 1 architecture.
   
   Intel later released the Dechutes core Pentium II. (300,333,350,400,450mhz) The Dechutes core is of a smaller, .25 micron process technology, and runs at 2 Volts. It also has a 512k external L2 cache, and runs on the Slot 1 architecture. Pentium II processors at 350, 400, and 450Mhz run at 100Mhz FSB instead of 66Mhz, like the earlier PII's. The last PII made was the 450, after which the PIII was released.
   
   The Pentium II Xeon was mostly for high end workstations. It has an external full speed L2 cache available in 512k, 1Mb, and 2Mb sizes, and is available in speeds of 400Mhz and 450Mhz. It uses the same Dechutes core as the PII and the Celeron. It also has some added features like an on core thermal diode that will shut down if the core overheats. The Xeon can also run in greater than dual processor mode(IE Quad processor), while the PII does not. The Xeons run on the Intel Slot 2 100mhz FSB architecture, and are not compatible with Slot 1.
   
    
7. Pentium III
   
   Pentium III was first released with the Katmai core. (500-600mhz) This was an improvement on the Pentium II, but very similar. Still .25 microns, but with an external 512k L2 cache.
   
   Later, the Coppermine core was released. This was a major improvement on the Pentium III. The core was shrunken to .18 microns, and a full speed 256k L2 cache was added. All of the Pentium IIIs from 600mhz on are Coppermine.
   
   The Pentium III is by far the most confusing chip that Intel has released, and it's probably best if I give a rundown of how the chips were released.
   
   NOTE: This is the most important thing to know about the Pentium III. There are two variables involved.
   
   The first is the FSB speed. It is either 100mhz or 133mhz. This speed is almost always listed with the processor, but if it isn't, then you can tell by a B after the speed of the chip. For example, PIII 600B means that the chip has 133mhz FSB. Intel has further confused the issue by generally only putting a B after a chip speed if the chip is released in both 100mhz and 133mhz speeds. This is not exclusive of the second variable, E, described below.
   
   The second is bus interface. It is either FC-PGA or Slot 1. FC-PGA stands for Flip Chip Pin Grid Array, and is a socket interface for the Pentium III. This is almost always listed with the processor, but if it isn't, you can tell by the E after the chip speed. For example, PIII 550E means that the chip is FC-PGA, and not Slot 1. This is not exclusive. IE, you can have a PIII 550EB that is FC-PGA at 133mhz. You need to be sure to purchase a motherboard and processor that are compatible, because you can't plug a FC-PGA processor into a Slot 1 motherboard, and vice versa. (Although adapters are available).
   
   The reason why Intel flip flops between formats is because the Slot 1 supports off chip L2 cache, but is more expensive, and the FC-PGA and Socket 370 are cheaper, but do not support off chip L2 caches. It looks like in the long run that the Slot 1 architecture is doomed for every format but the Xeon, but only time will tell.
   
   For a quick reference, here are the processors released under each bus speed.
   
   100mhz FSB - 500, 550, 600, 650, 700, 750, 800, 850
   
   133mhz FSB - 533, 600, 667, 733, 800, 866, 933, 1000
   
   Again, remember if the chip is released under both bus speeds, look for the B to tell if it is 133mhz, and look for an E to tell if it is FC-PGA.
   
   Pentium III Xeons were released in two types: Cascades and Tanner cores. These still require Slot 2 architecture, and are not compatible with Slot 1. Any Pentium III Xeon with more than 256k of L2 cache is a Tanner core. These have 100mhz FSB speeds, and have L2 cache that is full speed and available in sizes of 512k, 1Meg and 2Megs. Any Xeon with a FSB of 133mhz is a Cascades core. These all have 256k on-die L2 cache. The Cascades models hit the market well before the Tanner ones do, because it takes longer for them to engineer L2 cache that is full speed for the Tanner models. At this time, the only Tanner models available are below 600mhz.
   
    
8. K6/K6-2/K6-3
   
   The first notable family of chips that AMD released was the K6. It was released in speeds of 166-300, with bus speeds of 66 and 100mhz. The 66mhz chips ran on Socket 7 motherboards and the 100mhz ones ran on Super 7 motherboards. L2 cache was handled by the motherboard. None of the K6 chips had L2 cache on the chip, Socket and Super 7 chips always use L2 motherboard caches.
   
   K6-2 chips were produced in speeds from 300mhz up to 550mhz, and again with bus speeds of 66mhz and 100mhz. Again, these chips were made to run on Socket and Super 7 motherboards, and the L2 cache was located on the motherboard itself.
   
   K6-3 chips were an improvement over the K6-2 in that they were equipped with a 256k on-die full speed cache. They were only offered in speeds of 400 and 450, however. The interesting part about these chips is that they still use the Socket 7 motherboards, but the L2 cache on the motherboard is used as a L3 (level 3) cache, since there is L2 cache on the chip itself. The K6-3 were manufactured using a .25 micron technology. They work on any motherboard that supports their voltage, which is 2.3-2.5 volts, and the BIOS has to recognize K6-3.
   
   One last thing to note about the K6 family of processors is that they introduced 3dNow! technology, which is a set of instructions built into the processor that enhance 3D performance, and offered improved performance in 3D applications.
   
    
9. Classic Athlon
   
   AMD later released the K7, or Athlon processor. This was a major leap for them. AMD ditched the Socket 7 interface in favor of a new interface that they dubbed "Slot A". The Slot A FSB runs at 200mhz. They also have a 512k L2 cache. Older AMD Slot A processors have external cache usually running anywhere from .33 of the processor speed to .4 of the processor speed. They also released Thunderbird Slot A processors which had on-die cache on in their core. The Slot A Athlon is/was available in speeds from 500mhz up to 1000mhz, (1 Ghz). Athlon uses Enhanced 3DNow! instructions as opposed to the older 3Dnow! set.
   
    
10. Possible hazards of overclocking
    
    Before overclocking you should be aware of the many possible hazards of increasing the clock speed of your CPU or system bus.
    • Frying Your Processor Through overheating or improper jumper placement, there is a danger that you may short out or melt your processor beyond repair.
    • Frying Your Motherboard This component may also be shorted our or partially melted due to overheating or improper jumper placement.
    • Frying Expansion Bus / Cards When you change your system bus speed, your AGP, PCI, and ISA bus speeds also change. This, although extremely rarely, may result in the damage of your expansion cards.

At what clock speeds does the AMD Athlon processor operate?

-The AMD Athlon processor currently is available at clock speeds of 750, 700, 650, 600, 550 and 500 MHz. New AMD Athlon processors are now being built using AMD's advanced 0.18-micron process technology.

Is the AMD Athlon processor Super7 platform-compatible?

-No. The high-performance AMD Athlon processor requires a more powerful and scalable system bus than current sixth-generation buses in order to keep pace with emerging, bandwidth-intensive applications. The AMD Athlon processor has a 200-MHz system bus, with plans to move to 266Mhz next year, and it is based on the Alpha EV6 bus protocol licensed from Digital Equipment Corporation. This new bus is designed to enable AMD Athlon processor-based multiprocessing platforms that will provide the scalability and enhanced performance required by the commercial enterprise market.

Is the AMD Athlon processor compatible with Intel's PentiumR III motherboards?

-No. The AMD Athlon processor uses AMD's Slot A module design, which is mechanically compatible with Slot 1 motherboards but uses a different electrical interface. Because Slot A and Slot 1 infrastructures are not electrically compatible, the AMD Athlon processor cannot work with Slot 1 motherboards. Slot A is designed to connect electrically to a 200-MHz system bus based on the Alpha EV6 bus protocol, thus delivering a significant performance advantage over Slot 1 infrastructure.

Can I upgrade my AMD-K6R-2 or AMD-K6-III processor-based system to the AMD Athlon processor?

-No. The AMD Athlon processor is a totally new architecture and system platform demanding a new set of thermal and electrical specifications and enclosure requirements to match its leading-edge performance capabilities.

What is the Socket A?

-Socket A is the internal codename for AMD's next generation socketed interface for the AMD Athlon processor. Future versions of the AMD Athlon processor are planned to be available in socketed versions to enable lower infrastructure costs, lower packaging costs, and smaller form factor enclosures. Socket A will be unique to the high performance AMD Athlon processor, and will not be electrically or mechanically compatible with existing, lower performing socket interfaces

How large are the L1 and L2 caches of the AMD Athlon processor?

-The AMD Athlon processor's on-chip L1 cache is 128KB-four times the size of the Pentium III processor's L1 cache. The AMD Athlon processor's advanced L2 cache design is scalable in size and speed. The current version of the AMD Athlon processor, including the 750MHz AMD Athlon processor, have a 512KB L2 cache. Future versions are planned to offer full-speed, on-die L2 cache sizes scaling to 8MB, enabling a total system cache larger than any x86 processor solution currently available.

Is the AMD Athlon processor compatible with my favorite software?

-Yes. The AMD Athlon processor was designed to be compatible with the MicrosoftR WindowsR operating systems, including Windows 98 and Windows NTR, as well as other leading operating systems, such as Unix, Linux, OS/2 Warp, and Novell NetWare. The AMD Athlon processor was designed to be compatible with the existing installed base of more than 60,000 software packages.

What is the AMD-K6R-2 processor?

-A significant step beyond the popular AMD-K6R processor, the AMD-K6-2 processor is the first to offer AMD's innovative 3DNow!¡éa technology to deliver better overall performance than PentiumR II and a superior 3D experience for WindowsR computing at an affordable price. The AMD-K6-2 processor features superscaler MMX¡éa technology for leading-edge two-dimensional multimedia performance and enables support for a 100-MHz system bus to speed access to L2 cache and main memory by up to 50 percent leading to a significant increase in PC system performance. Currently, the AMD-K6-2 processor is available in clock speeds of 366-MHz, 380-MHz, 400-MHz, 450-MHz, 475-MHz and 500-MHz.

What is AMD's 3DNow! technology?

-3DNow! technology is the first innovation to the x86 processor architecture that significantly enhances floating-point intensive three-dimensional (3D) graphics and multimedia performance for today's mainstream MicrosoftR WindowsR-compatible personal computers. Benefits of 3DNow! technology include leading-edge 3D performance, more realistic and lifelike 3D imaging and graphics, big screen sound and video, and the ultimate Internet experience. For more details, see Inside 3DNow! technology.

Does the AMD-K6R-2 processor with 3DNow! technology replace my 3D graphics accelerator card?

-No, the AMD-K6R-2 processor works hand-in-hand with your high-end graphics accelerator to enable a complete PC that delivers leading-edge 3D performance.

How is 3DNow! technology different from MMX technology?

-MMX was developed primarily to improve integer-based multimedia performance, but it does not effectively address the needs of floating-point intensive 3D applications. The growing trend towards 3D applications requires even more powerful processors with specific enhancements to accelerate 3D performance. 3DNow! moves beyond MMX to deliver the 3D performance needed for today's emerging 3D applications such as games, edutainment, and business productivity titles. Both the AMD-K6R and AMD-K6-2 processors include MMX technology. The AMD-K6-2 is the industry's first processor to implement 3DNow! technology.

Is the AMD-K6R-2 processor compatible with existing software?

-Yes, if it's from AMD, it's compatible. Extensive internal and external testing ensures that the AMD-K6R-2 is compatible with all versions of the MicrosoftR WindowsR operating system, other leading operating systems, and more than 60,000 software applications. The AMD-K6-2 processor has the support of Microsoft and the independent software community.

Is the AMD-K6R-2 processor compatible with the popular Socket 7 platform?

-Yes. Socket 7 represents the vast majority of today's installed base of PCs and has been proven for reliability, upgradeability, and low cost. For even better system performance with the AMD-K6R-2 processor, choose a Super7 motherboard. Before selecting a motherboard you should first check with the manufacturer and review the recommended AMD motherboard support list.

What is the Super7 platform?

-The Super7 initiative supercharges today's Socket 7 platform by adding support for AGP (Accelerated Graphics Port) and an industry-standard, high-performance 100-MHz system bus. The addition of a 100-MHz interface speeds access to level 2 cache and main memory by up to 50 percent, leading to a significant increase in system performance. Super7 motherboards are the best choice for high-performance AMD-K6R-2 processor-based PCs.

What motherboards support the AMD-K6R-2 processor?

-Check the AMD recommended motherboard support list for motherboards AMD has tested with the AMD-K6R-2 processor. While the AMD-K6-2 is Socket 7 compatible, a Super7 motherboard is the best choice for peak system performance.

Can I upgrade my AMD-K6R or AMD-K5 to an AMD-K6-2 processor?

-Before attempting any upgrades, you should contact the manufacturer of your motherboard. You may need a new motherboard to upgrade your system effectively. A new motherboard will ensure your system has the correct voltage and BIOS support for the AMD-K6R-2. In addition, many new motherboards will support the new Super7 platform initiative which provides important new features such as AGP and a 100-MHz system bus for peak performance. AMD tests motherboards on an ongoing basis. Please check out our Tips on Making a Smart Motherboard Choice.

What is the AMD-K6R-III processor?

-The AMD-K6R-III with 3DNow! technology is the ultimate sixth-generation processor for home PC enthusiasts and business power users. The AMD-K6-III processor combines 3DNow! technology with AMD's new TriLevel Cache design to deliver exceptional performance on leading business and consumer software applications. The AMD-K6-III boasts as much as one speed grade advantage in performance on mainstream business applications when compared to a similarly clocked PentiumR III processor. The key to this enhanced performance is AMD's innovative TriLevel Cache design, which provides the largest total system cache for WindowsR compatible desktop PCs - more than four times the size of the other competing system designs. The AMD-K6-III processor is currently available at clock speeds of 400-MHz and 450-MHz.

What is AMD's new TriLevel Cache design? How does it improve performance?

-The new TriLevel Cache design of the AMD-K6R-III processor significantly boosts overall PC performance by providing the largest, fastest, and most flexible total system for WindowsR compatible desktop PCs. Complete details are covered in the TriLevel Cache White Paper.

How does the AMD-K6R-III processor differ from the AMD-K6-2?

-The AMD-K6R-III is the ultimate sixth-generation processor and is designed to compete against the PentiumR III in the Performance PC segment. An AMD-K6-III processor-based system offers as much as one speed grade advantage in performance when compared to a similarly clocked and configured Pentium III system. The AMD-K6-2 continues to offer superior price/performance value against the Pentium II and Celeron processors for mainstream PCs. Both the AMD-K6-III and AMD-K6-2 processors offer AMD's 3DNow! technology.

What is AMD's 3DNow!technology?

-3DNow!technology is the first innovation to the x86 processor architecture that significantly enhances floating-point intensive three-dimensional (3D) graphics and multimedia performance for today's mainstream MicrosoftR WindowsR compatible personal computers. Benefits of 3DNow! technology include leading-edge 3D performance, realistic and lifelike 3D imaging and graphics, big screen sound and video, and the ulimate Internet experience. 3DNow! technology and Intel's Streaming SIMD Extensions (SSE) in the PentiumR III processor include similar technologies. However, 3DNow! technology first became available in the AMD-K6R-2 processor and is also included in the AMD-K6-III and AMD Athlon processors. 3DNow! technology has more than a nine-month time-to-market headstart and enjoys a worldwide installed base of more than 15 million AMD-K6-2 processor-based PCs. This installed base is planned to grow to exceed 30 million by the end of 1999. See complete details 3DNow!Technology .

Does the AMD-K6R-III processor with 3DNow! technology replace my 3D graphics accelerator card?

-No, the AMD-K6R-III processor works hand-in-hand with your high-end graphics accelerator to enable a complete PC that delivers leading-edge 3D performance.

How is 3DNow! technology different from MMX technology?

-MMX technology was developed primarily to improve integer-based multimedia performance, but it does not effectively address the needs of floating-point intensive 3D applications. The growing trend towards 3D applications requires even more powerful processors with specific enhancements to accelerate 3D performance. Go to 3DNow! Technology.

3DNow! technology moves beyond MMX technology to deliver the 3D performance needed for today's emerging 3D applications such as games, edutainment and business productivity titles. All members of the AMD-K6 family of processors include MMX technology. 3DNow! technology was first implemented in the AMD-K6R-2 processor and is supported in the AMD-K6-III processor.

Is the AMD-K6R-III processor compatible with existing software?

-Yes, if it's from AMD, it's compatible. Extensive internal and external testing ensures that the AMD-K6R-III is compatible with the MicrosoftR WindowsR operating system, other leading operating systems, and more than 60,000 software applications. The AMD-K6-III processor has the support of Microsoft and the independent software community.

Is the AMD-K6R-III processor compatible with the popular Socket 7 platform?

-Yes. Socket 7 represents the vast majority of today's installed base of PCs and has been proven for reliability, upgradeability, and low cost. For even better system performance with the AMD-K6R-III processor, choose a Super7 motherboard. Before selecting a motherboard you should first check with the manufacturer and review the recommended AMD motherboard support list.

What is the Super7 platform?

-The Super7 initiative supercharges today's Socket 7 platform by adding support for AGP (Accelerated Graphics Port) and an industry-standard, high-performance 100-MHz system bus. The addition of a 100-MHz interface speeds access to level 2 cache and main memory by up to 50 percent, leading to a 10% increase in overall system performance. Super7 motherboards are the best choice for high-performance AMD-K6R-III processor-based PCs.

Can I upgrade my AMD-K6R or AMD-K6-2 to an AMD-K6-III processor?

-Before attempting any upgrades, you should contact the manufacturer of your motherboard. You may need a new motherboard to upgrade your system effectively. A new motherboard will ensure your system has the correct voltage and BIOS support for the AMD-K6R-III processor. In addition, many new motherboards will support the new Super7 platform initiative which provides important new features such as AGP and a 100-MHz system bus for peak performance. AMD tests motherboards on an ongoing basis. Please check out our Tips on Making a Smart Motherboard Choice.

What AMD-K6R-III processors are available now?

-AMD-K6R-III processor-based systems are available now at a clock speeds of 400-MHz and 450-MHz. AMD-K6-III processor-based PCs are available from leading PC manufactures, retailers, distributors, and resellers around the world.

 

What is overclocking?

There are several ways. You may change the multiplier by manipulating resistors on the Athlon's PCB, by purchasing or performing a modification that adds DIP switches to your Athlon, or by purchasing or making an overclocking card (called a "Golden Fingers" card).

AMD Athlon overclocking FAQ:
Answers to frequently asked questions
 

The resistor manipulation method of overclocking the Athlon was first publicized by Tom Pabst. He published an article detailing how multiplier settings and voltage settings are determined on an Athlon via several sequences of resistors mounted on the PCB. By desoldering and soldering resistors on the PCB to change the sequences, one can use this information to modify the multiplier on the chip and run it at a higher speed. Additionally, other resistor sequences may be changed to bump the voltage up slightly as discussed above.

What are the advantages of the resistor method?

The main advantage is that it gives you complete control over every aspect of the processor that you might want to change, from the multiplier to voltage settings to cache divisor. Additionally, once you're done, if you've done a good job of things, the modification doesn't increase the fragility of the processor in any way; it's just as durable to physical shock as it always was.

What are the disadvantages of the resistor method?

There are several disadvantages to this method. The resistors are very small and require a skilled hand to solder and desolder without damaging the PCB and rendering the CPU useless. As many as twelve resistor locations must be manipulated to change the multiplier value of the processor. If you wish to modify the core voltage of the processor as well, as many as four more resistor locations must be changed.

And changing them once probably won't be enough. Overclocking is hardly an exact science, and, as we discussed before, in the past the preferred method has been to start with the smallest step up and gradually work up to the point where the chip becomes unstable. At this point, one might bump up the voltage to the chip slightly to see if that allowed the unstable speed to work properly. If it didn't, one would simply back the chip down to the last stable setting and be done with it.

This is all well and good when you're simply manipulating jumpers or, on more recent boards, changing BIOS settings. But when each change in multiplier or voltage means carefully soldering and desoldering a series of resistors on a $200+ processor, overclocking starts looking a lot less attractive.

What is the DIP switch method of Athlon overclocking?

The DIP switch method was conceived to eliminate some of the headaches of the resistor manipulation method. It was conceived more or less simultaneously by Mark L. Sorensen at Trinity Micro, and Mike Honn. To some extent, the DIP switch method is an evolution of the resistor manipulation method.

Rather than soldering the resistors into a hard-coded configuration, the resistors are removed from the PCB completely and instead wires are soldered to each side of the resistor junction. The wires lead to a series of DIP switches and resistors. Once the modification is performed, one can choose to place or not place a resistor at each resistor junction by flipping the DIP switches in specific patterns. This modification enables the end user to manipulate the multiplier, voltage and cache divisor at will with the flip of a few DIP switches.

What are the disadvantages of the DIP switch method?

There are disadvantages to this method, as well. Obviously, the warranty is right out the window here; the chip has a rat's nest of wires dangling off it. Another problem is it makes the CPU extremely vulnerable; accidentally brush against the CPU and break a wire off while plugging in that new GeForce card, and you're in a world of hurt.

Trinity Micro has altered the layout of their DIP switch method so the wires are protected inside the original Athlon casing. A small hole is cut out of the casing and the DIP switches are mounted there. This setup eliminates the chance of accidentally snagging a loose wire, but we have no way of knowing how fragile the processor remains to physical shock using this method.

How difficult is the DIP switch method?

From the looks of it, performing this modification would make the aforementioned resistor manipulation look easy. Soldering resistors isn't exactly child's play, but now you're soldering 32 wires to those resistor locations, and then soldering them onto the DIP switches. If you want to manipulate the cache divisor, you'll be doing even more soldering.

Fortunately, however there are individuals who will do it for you. The cost is currently around $100 when you supply the CPU. Shell out the cash and all you need to overclock is a ball point pen to flip the switches.

However, if the CPU doesn't want to overclock very well, you're basically out of luck. Once the modification is done, it's highly doubtful the CPU could be returned to its original state without it being fairly obvious that the modification had been performed. Thus, returning the chip for a hopefully more overclockable one isn't likely to happen. While this method is more flexible than the first one, it is still less than ideal.

What is a Golden Fingers card?

The publication of the first public information on Golden Fingers method can also be credited to Dr. Pabst, and it is almost certainly the best means of Athlon overclocking to date.

Where does a Golden Fingers card plug into the Athlon?

The Athlon is manufactured with an edge connector on the top corner of the PCB. The Golden Fingers device is a card that fits onto this connector and overrides the resistor settings on the PCB. As a result, the user can use DIP switches to manipulate both the multiplier of the CPU and the voltage. The card is unable to manipulate the L2 cache divisor, however.

Because the Athlon is shipped with a plastic casing that covers the PCB (and thus the edge connector) the casing must be removed before the overclocking card can be utilized. Removing the casing involves breaking it away from the pins that hold it on.

So how much is a Golden Fingers card going to run me?

The card is actually cheaper than the original DIP switch modification; we've seen the cards selling for US$40 to $69.

Where can I get a Golden Fingers Athlon overclocking card?

An extremely comprehensive list of Athlon overclocking cards may be found www.anative.com/goldfinger/gfbg.htm.

What disadvantages are there to using a Golden Fingers card?

Installing a Golden Fingers card is more or less reversible, save the fact the Athlon's plastic casing must be removed. Simply unplug the card, and the processor once again runs at the speed at which it was sold.

In theory, a motherboard manufacturer could probably put L2 cache divisor controls in the BIOS, but no manufacturer has yet done so.

Several Athlon overclocking cards do offer the ability to manipulate the cache divisor. We have no experience with the cache speed manipulation abilities of any of the cards, so we can't comment on their effectiveness. Since the edge connector on the Athlon doesn't allow for the manipulation of the cache divisor, soldering would typically be required.

Changing the multiplier on an Athlon seems like an awful lot of trouble. Why shouldn't I just buy an Intel processor?

There are a multitude of reasons to go with an Athlon over a Pentium III. First, there is the fact that, clock speeds being equal, the Athlon will beat the Pentium III in just about any test you can throw at it.

Second, while changing the multiplier on an Athlon isn't much fun, it is at least possible. Information on how to change the multiplier on a bus-locked Intel processor isn't publicly available. With an Intel chip, you're stuck with manipulating bus speed--a less than ideal solution.

Finally, there is the fact that, subjectively speaking, Athlon chips seem more overclockable than the newest Intel chips. AMD designed the core of the Athlon with high clock frequencies in mind, and it appears they succeeded. A number of people have reported taking Athlon 500's up to over 800 MHz with nothing more than a good heat sink/fan combo. The biggest clock frequency increase we've heard of for an overclocked Athlon is a rumored case of someone using a water-based cooler to hit 912 MHz on an Athlon 500.

How important is cooling to Athlon overclocking?

On modern processors, good cooling is extremely important. AMD ships retail Athlons with beefy heat sink/fan combinations, and overclocking only heightens the need for good cooling. The more heat dissipated by the heat sink, the cooler the processor runs and the faster it can run without encountering problems. Of course, a good heat sink/fan combo doesn't help enough if your PC's case doesn't have good ventilation, as well.

So what should I look for in a heat sink?

AMD has a list of recommended Athlon coolers on their web site, and we've heard that a number of retail Athlons have shipped with Alpha coolers.

How do I test my overclocked processor's stability?

Generally, folks run highly CPU-intensive programs over an extended period of time in order to test the seaworthiness of an overclocked processor. A truly stable system should be able to run a really nasty program overnight without overheating, locking up, or crashing. Some common torture tests:

· Load up Quake 2 and let it run its looping game demo for hours on end. If you're trying to pinpoint the source of crash problems, Q2 can be run with a software renderer, so you can take your 3D video card out of the equation if needed.

· Or, more simply, run Prime95 or the SETI@Home client for a good, long time. These number crunching beasts will strain your PC's processor quite well, also.

Keep an eye on the system, especially early on, when running these sorts of tests. If an overclocked processor overheats and begins to get a little woozy, shut down the PC before any real damage is done.

What is the future of Athlon overclocking?

We're glad you asked. :) There are some new challenges on the horizon as the Athlon product line evolves.