Your computer feels a little slower than it did this time last year; is that change something you can chalk up to an aging processor?

今天的问答环节是由SuperUser提供的，SuperUser是Stack Exchange的一个分支，它是一个由Q&a网站组成的社区驱动分组。

问题

超级用户读者本·辛普森提出了以下问题：

This is a hypothetical question about how a CPU operates. If I purchase two identical CPUs, and use one long term (say one year), will it be identical in speed to the unused CPU? Will the number of clock cycles, latency of requests, etc on the used CPU be less than that of the unused CPU?

A supporting argument may be that mechanical devices degrade over time, While a CPU has no moving parts (other than the external fan), it does have circuits that can be damaged by heat, and voltage spikes. Lets say that after a year of intensive use, the circuits degrade and fewer electr*** can pass since the pathway is narrower, etc.

Is this the nature of how a CPU operates, or is it simply working or broken, with no speed degradation in between?

中央处理器会随着时间的推移而退化吗？还是其他因素在起作用？

答案

超级用户贡献者RedGrittyBrick详细介绍了如何控制CPU的速度：

Is the performance of a CPU affected as it ages? after a year of intensive use, the circuits degrade and fewer electr*** can pass since the pathway is narrower, etc.

No: Crystal Oscillator

The speed of a CPU is determined by a crystal oscillator – so far as I know this is an external part for most CPUs

Picture from TechRepublic article

Crystals undergo slow gradual change of frequency with time, known as aging.

However, I suspect this is not a significant factor.

Drift with age is typically 4 ppm for the first year and 2 ppm per year for the life of the DT-26 crystal.

(from TI concerning an RTC IC but I believe this rate is similar for timing crystals in general)

CPU Semiconductor changes

Breakthrough posted a link to an IEEE article that describes the myriad of ways that semiconductors are affected over time.

It is possible therefore that the maximum clock speed the CPU is capable of will decrease over time. However in most cases this will not cause the CPU’s theoretical maximum possible speed to fall, within a year, below the actual operating speed set by the crystal oscillator. Therefore a CPU that has been stored for a year will run at the same speed as an originally identical CPU that has been used continuously for a year.

CPU Thermal regulation

Many CPUs reduce their speed if their temperature exceeds a pre-set threshold. The main factors that might cause a one-year-old CPU to overheat are not to do with semiconductor degradation within the CPU itself. Therefore these factors have no bearing on the question as formulated.

It is unlikely that a given pair of identical CPUs will diverge in capability within one year sufficiently to trigger thermal issues that require one of them to run itself at a reduced speed. At least, I know of no evidence that this has occurred within one year on a device that is not c***idered a warranty failure due to manufacturing defect.

CPU Energy efficiency

Many computers, especially portable ones, are similarly designed to reduce energy c***umption when idle. Again this is not really relevant to the question as stated.

BlueRaja跳了进来，除了Ben的回答：

In theory, no, a CPU should run at basically the same speed its entire life.

In practice, yes, CPUs get slower over time because of dust build-up on the heatsink, and because the lower-quality thermal paste that prebuilt computers are often shipped with will degrade or evaporate. These effects cause the CPU to overheat, at which point it will throttle its speed to prevent damage.

Cleaning the heat sink and reapplying the thermal paste should make it as good as new, though.

Note: if you’re asking this due to having an old computer slow down, there are other reas*** (usually dying hard-drives or popped capacitors) that old computers will slow down over time.

换句话说，糟糕的电脑维护和廉价的组装方法才是真正的速度节流恶魔，而不是物理芯片上的老化或磨损。常规的清洁和高质量的热糊有助于CPU的高效运行。

有什么要补充的解释吗？在评论中发出声音。想从其他精通技术的Stack Exchange用户那里了解更多答案吗？在这里查看完整的讨论主题。