Overclocking

When overclocking (English: Overclocking ) is the operation of processors and other computer components ( hardware components ) with a higher clock frequency referred to above normal manufacturer specification, to achieve a higher processing power with the goal. The opposite of this is the underclocking, which is applied mostly as an energy saving measure, or with the aim of a longer life.

• 7.1 Records

Overview

The increase in clock rates of components, such as CPU, graphics card, or memory, resulting in a power increase of the overall system. Depending on the components and cooling as a significant performance gain can be achieved, so that in some cases the performance of the highest clocked version of the manufacturer can be exceeded.

This is possible because the manufacturers due to the development and production costs only a few versions of a microchip design. Furthermore, the microchips produced on a wafer in such a way that the heat is varied at the same clock rate differ in quality.

These products will then cover a manufacturer from all price ranges of the market. However, the demand for more powerful chips is lower. It often happens that a chip that would tolerate higher clock rates, is sold with regard to the generation of heat at the maximum permissible heat load, ultimately for lower clock rates. This results in the relatively high overclocking capabilities with mid-size processors.

Procedure

Small microchips are overclocked by ( a quartz oscillator ) replaces the clock generator.

The clock of a component of a computer can usually be in the BIOS and / or by special software during operation to adjust a setting in the BIOS, however, is always preferable, since the change of the clock is limited in operation to a frequency range and the adjustment of the supply voltage of some components usually not possible. Older motherboards also have often also jumpers for setting the frequency.

Is due to the increased performance of the device, depending on the overclocking result, a higher supply voltage necessary. This is especially important for rapid storage and requires, as the power is converted into heat, according to additional or better cooling elements, in order to avoid harmful for the component temperature. This can be achieved with CPUs with good air coolers, water coolers, compressor coolers, dry ice and liquid nitrogen, the latter often are self-made, eg from car radiators, refrigerators, or copper pipes in order to save costs and tailor the properties of cooling to the system. Incorrect wiring can damage the component irreparable. Dry ice and nitrogen but are used mainly for obtaining overclocking records, in systems that were built exclusively for it and are usually no longer used after the record attempt.

When overclocking PC components such as processors per multiplier - raising procedure is always iterative, that is, that for example, the multiplier of the processor or the Front Side Bus ( FSB) gradually as far as the stability it permits increases.

It is recommended to turn off the power saving features like C1E and Cool'n'Quiet or Enhanced SpeedStep, and to waive the automatic vCore voltage since it is often set too high by the mainboard and to undesirable results, such as may result, for example, SNDS and instead with programs such as RMClock or Crystal CPU ID make the dynamic clock control by the multiplier, as this is the possibility of assigning a certain voltage to a multiplier, which may differ greatly extending downwardly from the specified voltage of the main circuit board.

In addition, you should access the BIOS LLC setting with caution. LLC stands for Load-Line Calibration and ensures that the set in the BIOS voltage is meticulously, precisely maintained even under load. This contradicts the fact intentional V- Droop and can lead to stability problems.

Types of overclocking

Manually

The clock of a computer component usually consists of two variables: the reference clock, with many processors the clock of the Front Side Bus ( FSB), and the multiplier. The reference clock is generated from either the main processor itself or by a separate chip on the motherboard. To achieve the nominal clock of the reference clock is multiplied by the multiplier.

It is therefore generally possible to overclock a processor via the FSB or the multiplier. However, many processor manufacturers lock the multiplier of their processors (on most processors, the multiplier is only down freely adjustable), so that the increase of the reference clock is the most common type of overclocking. For special versions, such as the Black Edition processors from AMD and the Extreme Edition processors from Intel, the multiplier is not locked up, so you can lift it in the BIOS.

• Multiplier 10 x 200 = 2000 MHz reference clock frequency

By raising the reference clock such as the PCI bus (regular 33 MHz), AGP ( r. 66 MHz) or PCI bus are usually also the bus clock of memory, (reigned 100 MHz) with raised. Therefore, this bus clock should be fixed in BIOS, because the corresponding plug-in cards easily survive an increase in this own clock only in the rarest cases. With newer motherboards there are usually ways to separately adjust some bus frequencies from the reference clock, giving the overclocking benefits, as not all components can be equally overclocked the system.

Automatically

In addition to manual overclocking increasingly specialized functions are built directly into the main processors to which the automatic overclocking is possible during operation, when operating conditions permit. These are referred to by AMD as Turbo Core Technology and Intel Turbo Boost Technology as.

Stability

Overclocked systems can be unstable and ( computing ) produce error, for example by increasing the temperature. These are manifested in a system crash, crashing programs or similar undesirable properties. When overclocking a PC system is usually a gradual approach. After each rise of the clock (e.g., FSB ), or increase of the multiplier stability tests are carried out. The stability after overclocking is therefore often tested with the software Prime95. Withstands the overclocked system the more hour Torture Test the CPU and the memory, it is generally regarded as stable. Another very popular program, for example, BOINC, which saturates as all processor cores.

There are many more programs to test the system stability, but basically is as good as any program to check the stability when the program saturates the overclocked component (and possibly also checked for calculation errors ).

Hazards

During the operation of components outside of their specification goes in almost all cases the manufacturer's warranty. Also, the life expectancy of the overclocked components can sometimes fall significantly.

By a high clock, and especially the increased voltage, the electro-migration is promoted, which has been known, for example as a Sudden Northwood Death Syndrome. The power loss increases linearly with the clock and the square of the voltage.

A smaller source of danger is the resulting heat when overclocking. While with older CPU models led to high overclocking or too little cooling or cooling failures damage due to overheating, today there is a technique called throttling ( " throttle "), which prevents the processor gets too hot by simply omitting bars. However, this can lead to low performance and overclocking thus is counterproductive.

For the first time this technique was used in the Intel Pentium 4 Willamette processor, while AMD at this time in their models still refrained (although support most newer chipsets from AMD CPUs motherboard controlled throttling). The precursor of the of throttling was used with the Pentium 3, who had never missed any bars, but simply shuts down. Another variant of the protection has already been used in the late 1980s by Siemens. The self-developed motherboards was installed a thermal control chip, the abregelte the clock frequencies to high thermal stress and has also been used in part to adjust the fan speed.

Cooling

Due to the increased power dissipation is overclocking generally improved cooling and possibly a stronger power supply needed. This contributes to the rising popularity of water cooling in the PC area. At a water cooling but you should note that thus only individual parts, such as the CPU to be cooled. Other components, such as the voltage transformer on the mainboard, which are cooled in a conventional air cooling from the airflow of the CPU fan, can be very warm or even overheat and thus destroyed by the lack of this fan. That's why even high-performance case fan or explicit component coolers are used in addition to said water cooling systems. The variety of processor cooler is however almost not restricted.

If air cooling and water cooling is no longer sufficient, may be resorted to more efficient methods such as compressors, dry ice or liquid nitrogen. Due to the complicated procurement of dry ice and liquid nitrogen, however, these methods are usually used only in individual experiments in which it comes to overclock the CPU as far as possible.

Record hunt

Some overclockers work for profit, mostly for relevant sites that are funded by advertising. The goal is to overclock CPUs as high as possible. There are different categories, both in the cooling (eg no restriction, water cooling, air cooling) as well as the necessary stability (eg BIOS bootable Windows bootable (usually until the appearance of the GUI), Prime -stable ). In addition, a distinction is made according to CPU manufacturer, type of processor and chipset. In addition, there are still record experiments in which it is not about the highest possible clock speed, but the best performance in certain benchmarks, such as 3DMark, Super PI and SiSoft Sandra.

Records

2013, Intel Celeron 347 of 3,066 MHz overclocked to about 8398.07 MHz. That was at the time the world record for Intel processors.

AMD introduced shortly before the official presentation of the bulldozers row in front of a record, an AMD Bulldozer FX overclocked to 8,429.4 MHz. Cooled with liquid helium was doing.