CPU Overclocking Considerations

Statement:

1. Overclocking has a certain risk, will lead to hardware scrap

2. Overclocking hardware will reduce the short life

Overclocking can improve PC performance without spending a penny, and it is also a means for computer enthusiasts to improve their knowledge of hardware. Overclocking is not like a game, give you a button to let you click on the end of things, overclocking need to have a systematic knowledge of the hardware, no patience and no basic knowledge of the people can not be completed.

To overclocking a friend of the BIOS to have a certain understanding, at least to know what is written inside, or reference instructions.

Basic knowledge:

CPU frequency = Double frequency x FSB (FSB)

For example 2.0G (CPU Frequency) =200mhz (FSB) x10 (octave), here you can put 1g=1000mhz.

Explain the formula:

FSB is the front-end bus, you can understand this: the FSB is like every country (computer components) to the city (CPU) road, the frequency can understand the number of road (FSB), so there is a conclusion: increase the number of roads or increase the width of the road can be achieved to increase the volume of traffic

So:250mhz x 8 = 2G, or: 100MHz x = 2G

But: Obviously the width of the road is better than the number of bars, the PC is the same, generally will not use the latter formula, lower FSB will greatly affect performance.

However, FSB on some AMD CPU is inaccurate, but the theory is the same, outside the forest to note that most of the current CPU frequency is locked, can not be lowered or raised.

Some people say: understand, add FSB is finished, this is why not go up super unsuccessful reasons. But in fact, in many cases, when the FSB is raised, the entire system is widened, but memory, AGP, and PCI systems cannot withstand such a load, which can cause overclocking to fail.

How to solve this problem:

At this point a good BIOS system played a role, the Intel original motherboard has almost no overclocking settings to provide the option; The Via series of KT266, 333, 400, 600, 880, k8t800 without locking PCI and AGP (except for K8t800pro) caused a certain amount of trouble with overclocking, the NF series of motherboards are almost completely locked.

The most important part is memory:

When you use the formula 250MHz x 8 = 2G, your DDR400 memory becomes DDR500, and if the memory is unbearable, it will cause overclocking failure.

How to solve?

The BIOS overclocking settings will have a frequency setting, that is, fsb:ram, and the ratio is set, such as 5:4,3:2,4:3 and so on. Note that when using this ratio there will be a certain loss of performance, when a small amount of overclocking, good memory can withstand, then do not use the frequency division. So good memory is much more expensive than a motley memory.

Voltage:

When everything is in place, there is a limit to be achieved, at which point it is necessary to increase the voltage (which is also the root of the shortened life, the effect of small pressure is very small). Any part of the overclocking will be limited by the voltage, voltage is effective but also dangerous, million can not be greedy, when you smell the paste smell all late.

Heat dissipation problem:

This problem is often the first reason for failure, so please make sure your cooling system is sufficient before overclocking. Or the replacement of a stronger cooling system, the current water-cooled equipment is not very high prices.

Stable operation is the key:

There's no stability. Everything is White said, heard the overclocking madman to a certain CPU over 5G, but to kill will not believe that it is stable operation. General stability refers to working continuously for 24 hours.

Small knowledge: Why let PCI/AGP bus Ultra specification run will cause instability?

The main reason for the instability caused by PCI bus Hyper-specification is that it forces the parts with very strict tolerances to run at different frequencies. PCI specifications are usually specified under 33MHz. Sometimes it's under 33.3MHz and I believe it's close to the real specs. The main victim of high PCI speed is the hard disk controller. Some controller cards have a higher allowable bias than other cards, and can run at increased speeds without apparent damage.

However, onboard controllers (especially SATA controllers) on most motherboards are extremely sensitive to high PCI speeds and can have damage and data loss if the PCI bus runs at 35MHz. Most can cope with 34MHz, which is actually less than 1MHz in size (depending on how the motherboard is rounded to 34MHz ...). For example, most motherboards may report a 34MHz PCI rate under any FSB between 134 and 137MHz. The actual range is from 33.5MHz to 34.25MHz and may vary significantly based on changes in the motherboard clock frequency.

In higher FSB and higher crossovers, the range may be larger. Sound cards and other integrated peripherals are also compromised when PCI bus Hyper-specification runs. ATI graphics cards have a much smaller tolerance for high AGP speeds than Nvidia cards (directly related to PCI speed). Remember that most Realtek LAN cards (based on PCI and the expansion slots) are set to operate safely at any frequency from 30 to 40MHz