Overclocking FAQ is just a collection of basic prompts / techniques for overclocking, and what it is and what it includes one of the basic overview. What extent? Not all chip / components overclocking. Just because someone let Prescott go to 5 GHz, it doesn't mean that you can guarantee to 4 GHz, and so on. Each chip is different in overclocking capabilities. Some are very good, some are garbage, most of them are general. I tried it. Is this good overclocking? Do you feel happy? If you are sure, that is, unless it is only 5% or less overclocking - then you need to continue, unless you have become unstable after overclocking). Otherwise it will continue. If it reaches the boundaries of the chip, it will be powerful. How much is too hot / how much voltage is too hot? As a normal definition for security temperatures, the temperature under full load should be less than 60 c for P4, and the Athlon is 55 C. The lower the better, but you should not be afraid when the temperature is high. Check the part to see if it is well in the specifications. As for the voltage, 1.65 to 1.7 pairs of P4 are good boundary, and Athlon can go down to the air-cooled 1.8 / water 2.0 - Generally. Depending on the heat dissipation, higher / lower voltages may be appropriate. The boundaries on the chip are surprisingly high. For example, the maximum temperature / voltage on the Barton core Athlon XP is 85 C and 2.0 volts. 2 volt most overclocking is sufficient, and 85 c is quite high. Do I need better heat dissipation? Depending on what the current temperature is and what you are planning to do for the system. If the temperature is too high, it may need to be better heat dissipation, or at least you need to re-use the heat sink and finishing the wire. Good wire arrangement can work greatly to the air flow of the chassis. Similarly, the appropriate application of the heat sink is important for temperature. Let the heat sink close to the processor as much as possible. If it is a lot of help or completely, then you may need to be better heat. What is the most common heat dissipation method? The most common method is air cool. It places a fan above the heat sink and then buckled on the CPU. These may be quiet, very noisy or between the two, depending on the fan used. They will be a quite effective radiator, but there is a more effective heat dissipation solution. One of them is water cold, but I will discuss it later. The air-cooled radiator is manufactured by Zalman, ThermalRight, Thermaltake, Swiftech, Alpha, Coolermaster, Vantec, and more. Zalman makes some of the best mute heat dissipation equipment and is known for their "flower radiator" design. They have the most effective mute heat dissipation design 7000CU / ALCU (all aluminum or aluminum copper mixture), it is one of the better performance. ThermalRight is a highly performance heat dissipation device producer that is an unlatable highest performance heat dissipation device when using the appropriate fan. SWIFTECH and Alpha have a king of performance before THERMALRIGHT, and now is still an excellent heat dissipation device, and can be used in a broader application area than the thermalright heat dissipation, because they are usually smaller than the thermalright heat dissipation equipment. Motherboard. Thermaltake produces a large amount of cheap radiator, but I am in the words, they are not worthless. They do not have the same level as the heat sink of other heat dissipation equipment, but they can be used in a cheap chassis. This covers the most popular heat dissipation equipment manufacturers. Take it again. Water is mainly marginalized, but it has always become more mainstreaming. NEC and HP have manufactured water cooling systems that can be purchased in a retail method. Despite this, most of the water is still facing the world of enthusiasts. Several most basic components are included in the water-cooled circuit.
There is at least one tank, usually on the CPU, sometimes on the GPU. There is a pump, sometimes there is a water storage. There is also one to two radiators. The water tank is usually built with copper or (less information) aluminum. Even less, but it is growing is a silver treasure. There are several different types of internal design to the water tank, but I am not prepared to discuss those here. The pump is responsible for driving the water through the way. The most common pump is an EHEIM pump (1046, 1048, 1250), Hydor (L20 / L30) and Danner MAG3. The IWAKI water pump is also popular among high-end groups. The SWIFTECH MCP600 pump is becoming more popular. The two are high-end 12V pumps. The reservoir is useful because it increases the volume of water in the loop and makes filling and deflation (discharging the bubbles back) and maintenance easier. However, it occupies a considerable space in most chassis (small reservoirs are not in the case), and it is relatively easy to leak. The heat sink can be a finished product such as a Swiftech radiator or a Black ICE radiator, or the car heater core is also modified. The heater core is usually good in superior performance and lower prices, but it is more difficult to assemble because they usually do not use the shape that can be quickly and easily used. The fuel tank radiator is a desirable approach to those who have strange dimensional demand because they use very varied shapes and sizes (but usually a rectangle). However, their performance is not as good as the heater. Pipeline systems are also a factor in performance. Usually, high performance, 1/2 'diameter is considered to be the best. However, 3/8 'or even 1/4' diameter equipment is becoming more common, and their performance is approaching 1/2 'diameter loop. It is so much about the water cooling in this section. What is some rare heat dissipation type? Phase transition, frozen water, Peltier effect (thermal converter) and submerged equipment are rare, but the performance is higher. Both Peltier effect heat dissipation and frozen water are based on water-cooled because they use improved water-cooled circuits. Pell is the most common in these types. Pelt Post is a device that is refined and the other side is changed while the current is passed. This can be used between the CPU and the water tank or between the GPU and the water tank. It is rarely the Peltol on the North Bridge, but this is not necessary. The frozen water circuit uses a Peltier or a phase change to make the water in the loop cool, usually replace the heat sink to the CPU / GPU heat dissipation in the loop. It is not very efficient to use Pell Post to do this job because it often needs another water-free circuit to make it cool. Pell Post usually is sandwiched with another water tank with another water tank with a heat sink and water tank or water tank. The phase change method includes placing a cold air head or a cold air member in the A / C unit, or like it is in a water storage pool. Antifreeze in the frozen water is usually added to the water at a ratio of approximately 50/50, because the ice is not good. The pipeline system must be insulated and the water tank is also the case. The phase change includes a compressor and a cooling head connected to the CPU or GPU. Here I am not preparing to discuss it too deeply. Other uncommon methods include dry ice, liquid nitrogen, water-cooled PSU and hard drive, and other similar. It is also considered and tried using the chassis as a heat sink. What is the prefabricated water cooling system? Koolance and Corsair are the only truly consider. Small globalwin products are OK, but it is not cold than any medium and high-end wind. The rest can not. Avoid use them. The latest Thermaltake product may be good. The new kit may be quite good (Kingwin products seem to be like this), but to read several evaluations before purchasing any products, at least one is tested on the platform you will use. What is the danger of overclocking? There are several dangers about overclocking, which obviously should not be ignored.
Over-specifications run any part will shorten its life; but the new chip is in handling this problem, so this is hardly a problem, especially if you upgrade every 6 months or every year. For long-term stability, such as computers that are ready to run more than 2 years or similar working hours, overclocking is not good. Moreover, overclocking may damage the data, so if you don't have back up any important data, the overclock is really not suitable for you unless you can restore data without increasing data, and it will not cause any problems. However, the possible data loss should be considered before starting overclocking. If you only have a computer and need it to do important things, it is not recommended to overclock (especially over high voltage overclocking), because the possibility of partial damage is still (I have lost several parts to overclocking But it is better to lose much less than some people), so it is also necessary to consider. How do I overclock? This is a quite complicated issue, but the foundation is very simple. The easiest way is to increase FSB. This is almost valid on any platform. However, the VIA chipset (KT266 / 333/400 (A) / 600/880 and K8T800 - do not confuse the existing K8T800 PRO) No PCI / AGP lock, so you have to increase the FSB carefully because the super specification is running PCI The bus (33MHz is standard speed) may damage the hard disk data, hinder the correct operation of the peripheral device (especially the ATI AGP graphics card), usually unstable. This will be explained later. NFORCE2 chipset for the XP chip for AMD, NFORCE3 250, VIA K8T800 PRO, and Intel 865/875 chipset all have a locked PCI frequency. Otherwise, many I845-based motherboards will also have PCI / AGP locks. This makes it easy to adjust the FSB because it eliminates some limiting factors, such as peripherals that are sensitive to frequencies. However, the restriction is still existing. In addition to the impact of the chip itself, the RAM and the chipset and the motherboard can limit the FSB that can be obtained. That is the land of multiplier adjustment. On some Athlon XP chips, multiplier is adjustable. These chips are referred to as "non-locked". In addition to the FX series that is completely unlocked, the ATHLON 64 series allows multiplier to adjust to lower frequency. Pentium 4 is locked unless you have obtained engineering samples through some channels. However, almost all motherboards allow multiplier adjustment as long as the CPU supports it. Once the system becomes unstable because the CPU is limited, there are two options. Can be lowered to return to its stable position or improve the CPU voltage (possibly RAM and AGP voltage) to it be stable, or even higher to further overclock. If you increase the CPU voltage or increase the memory voltage, you can also try "relax" memory delay (raising those numbers) until it becomes stable. If all of this is useless, the motherboard may have an alternate scheme for increasing the chipset voltage. If the chip group charges heat, this may be helpful. If there is no help at all, you may need to heat up on the CPU or other components (for MOSFETS - smashing the chip cooling of the power supply - possibly useful and is quite common). If it is still useless, or if it is harder, it is on the limit of the chip or motherboard. If the voltage does not affect stability, then the most likely the motherboard. The voltage regulatory chipset is a possibility, but it is a bit too advanced and needs to be exceeded by normal heat dissipation. Similarly, it may be helpful to the South Bridge and North Bridge, or may improve stability.
I know that in my motherboard, if you don't have a cooling in the South Bridge, you will start the sound card when you run WINAMP / XMMS and UT2004, and you are now in Windows and Linux), regardless of the FSB. So it is not a bad idea, but it may not be necessary. It usually also makes warranty failure (more serious than overclocking - overclocking can usually do not leave traces). The basic overclocking is covered here. More advanced overclocking typically includes adding a heat sink to all parts, the voltage regulating motherboard may even be a power source, add more / better fan or water-cooled or phase change / laminating heat. What should I do if my computer does not show (displaying the BIOS screen while turning on)? It depends on the motherboard you have. The "failure recovery" scheme is used to reset the CMOS, usually complete through the jumper discharge. Find details in the motherboard manual. If the overclock is too high, the BIOS setting remains complete and unkflated, and the newly near-burn-up motherboard has an option to display at a reduced frequency, then you can enter the BIOS and lowered to a stable running clock speed. On some motherboards, this is done by pressing the INSERT key when turning on the computer (usually the PS / 2 keyboard). If the computer is still not displayed before the previous efforts, some will automatically reduce the frequency. Sometimes the computer is not cold-start (display when pressing the power button) but will run after a while, then restart. It will be hot in other occasions, but it is not possible to heat up (restart). Those are unstable signs, but if you feel satisfied with this stability and handle this problem, then it usually does not cause big problems. What limits my overclocking? Usually RAM and CPU are the only important limiting factor, especially in an intrinsic operation due to memory asynchronous operations (see the fsb chapter below). RAM has to run at the same speed as FSB or at its frequency frequency. Memory can run at a high speed than fsb, not just below it. However, there is a choice to run more high-delay / higher memory voltage, it becomes more and more like restriction factors, especially because new platforms (P4 and A64) have a less performance loss from asynchronous operation. The CPU has become the main limit factor. The only way to handle faster CPUs is to increase the voltage, but exceeding the maximum core voltage will shorten the life of the chip (although overclocking is also like this), the sufficient heat dissipation portion solves this problem. With another problem with the use of too high core voltages, in the P4 platform, in the form of SNDS, or Sudden Northwood Death Syndrome, any voltage above 1.7V can cause the processor to quickly And premature scrap, even if it uses phase change heat dissipation. However, the new C core chip, that is, EE chip, and Prescott chips do not have this problem, at least different. Heat heat can also hinder overclocking, because too high temperature can cause unstable. But if the system is stable, then the temperature is usually not too high. Now I have overclocking very much, what should I do? If you want, you run some benchmarks. Let PRIME95 (or you choose to emphasize - completely depending on you) run a sufficient long time (usually 24 hours without failure) is considered to be stable). What is FSB? FSB (or Front Side Bus, Front End Bus) is one of the easiest and most common methods. The FSB is the speed of the CPU and other parts of the system. It also affects the memory clock, which is the speed of memory operation. In general, the higher the fsb and memory clocks, the better. However, in some cases this is not established.
For example, let memory clocks are running fast than FSBs that will not have real help. Similarly, on the ATHLON XP system, the FSB runs at a higher speed and enforces memory and FSB asynchronous (the memory scorer using the memory divider will be discussed later) will compare the performance of performance in lower FSB and synchronous memory. Be serious. FSB involves different methods on Athlon and P4 systems. In Athlon, it is a DDR bus that means that if the actual clock is 200MHz, it is running under 400MHz. On P4, it is "four-core", so if the actual clock is the same 200MHz, it represents 800MHz. This is Intel's market strategy, because of the general user, the higher it is equal to the better. Intel's "four-core" FSB actually has a real advantage, which is to allow P4 chips to run synchronously with memory at a rate of smaller performance loss. The higher the cycle speed of each clock, so that it is more likely to cause the memory cycle to coincide with the CPU cycle, which is equivalent to better performance. Why do the PCI / AGP bus superconstructures can cause unstable? Let the PCI bus superconstruction result in unstable facts because it is forced that the components that have very strictly allowed deviations at different frequencies. The PCI specification is usually specified under 33MHz. Sometimes it specifies in 33.3MHz, I believe that is close to the real specification. The main victim of the high PCI speed is a hard disk controller. Some controller cards have higher allowable deviations than other cards, which can be operated at an increased speed without obvious damage. However, on the onboard controller (especially the SATA controller) on most motherboards is extremely sensitive to high PCI speed, if the PCI bus is running in 35 MHz, it will damage and data loss. Most of them can cope with 34MHz, in fact, the size is less than 1 MHz (depending on how the motherboard is rounded to 34MHz ... For example, most motherboards may report 34MHz PCI speeds between 134 to 137 MHz. Actual The range is from 33.5MHz to 34.25 MHz and may vary more based on changes in the main board clock frequency. The range may be larger in higher FSB and higher divider. Sound cards and other integrated peripherals are also damaged at the PCI bus superconstruction. The ATI graphics card is much smaller than the NVIDIA card with high AGP speed (directly related to PCI speed). Remember, most Realtek LAN cards (based on PCI and occupying the extended slot) are set at any frequency from 30 to 40 MHz. What is multiplier? The multiplication combined with the FSB to determine the clock speed of the chip. For example, the FSB of the multiplier 12 of 12 will provide a clock speed of 2400 MHz. As in the above overclocking chapter, some CPUs are locking multipliers and some, that is, only certain CPUs allow multiplier adjustment. If you have a multiplier adjustment, you can use a higher clock speed on the board restricted on the FSB, or higher FSB is obtained when the chip is limited. What is memory divide? Memory Division determines the ratio of the memory clock speed to the FSB. 2: 1 FSB: RAM Division will get 100MHz RAM clock to 200MHz FSB. The most common use of divideves is to allow the P4C system running at 250FSB to match PC3200 RAM, using 5: 4 divide. In most Intel systems, there are 4: 3 divided division and 3: 2. The ATHLON system cannot utilize memory as the P4 system is used in the use of divided systems, as explained in the fsb section above. Memory dividing should only be used to get stability, not the moment of origin, as it also harms performance even on P4. If the system does not take memory division, it is stable (or if the memory voltage can solve the problem), do not use the crossover.
