Can overclocking really break hardware?
Solution 1:
Is overclocking safe with modern hardware?
Overclocking is safe, so long as you keep the processor's temperature and voltage within the manufacturer's specification. This is different for each CPU (or family of CPUs), so you will want to find the datasheet for your particular processor before changing anything.
Could an overclock and/or an associated overvoltage cause physical damage to hardware, when talking about 20% increments?
Software can physically damage hardware, and you do happen to set the CPU's clock speed and voltage in software. Given a mild overclock, however, there will be no immediate physical damage to the processor - but it should be noted that high clock speeds, temperatures, and voltages are all known to accelerate transistor aging.
This isn't something you can completely mitigate - even running the processor at the default clock speed and voltage will wear down the transistors over time. This is all on a timescale measured in years, and the effects can be mitigated by compensating the CPU's speed and voltage over time.
In general, with modern hardware, what temperature ranges be safely reached by components? I'm lucky that my CPU fan keeps the processor 40°C cold when strongly stressed (not overclocked). Can I expect a physical damage at 90°C in a chipset? Or a video card?
Again, check with the manufacturer's rated specifications. I've seen some GPUs that are rated up to 100C, while also encountering CPUs that can only withstand up to 68C. Indeed, the chip will work past it's rated temperature/voltage, but the lifespan will be severely impacted. In general, the cooler, the better.
So, what does this all mean? For those users who want to achieve a 20% overclock, if you're aware of the risks and are comfortable modifying your processor's clock speed and voltage, go for it. While it's impossible to quantify how much you would reduce the processor's lifespan, it's probably safe to assume that you wouldn't shorten it past it's useable life in your system (especially if Moore's law keeps going).
Solution 2:
Can overclocking really break hardware?
Yes it can as you are making your hardware run at speeds for which it is not designed.
Is overclocking safe with modern hardware?
Yes and No, it depends on how much you want to push your hardware, honestly answer to this question is subjective. Most modern processors these days come with builtin safety feature and shutdowns the computer to prevent damage when they reach or cross the max junction temperature. With proper safety measures like better cooling and carefully overclocking your hardware you can reap the benefits of overclocking. In real life scenario over a period of time, I have hardly noticed any major beneficial gains.
Could an overclock and/or an associated overvoltage cause physical damage to hardware, when talking about 20% increments?
Yes it can, read the last paragraph of this answer.
In general, with modern hardware, what temperature ranges be safely reached by components? I'm lucky that my CPU fan keeps the processor 40°C cold when strongly stressed (not overclocked). Can I expect a physical damage at 90°C in a chipset? Or a video card?
It depends on whether your hardware is designed to operate at temperature of 90C. It all depends on the operating temperature for which the hardware is designed. Industrial grade and military grade hardware are designed for higher operating temperature. So this means if your hardware is designed to operate at 100C then even if you reach 90C I don't think it will get damaged. From my experience the cooler the computer hardware runs better it is for us.
Overclocking Risks:
- Overheating of hardware
- Shorter lifespan of the hardware
- Damage to other hardware components. eg. Overclocking CPU can sometimes damage motherboard, RAM etc.
- Unstable performance.
- Void warranty. :P
- Increased cooling needs due increased power consumption
Additional info what other things happen when overclocking is done:
From Benefits And Risks : Overclocking Guide Part 1:
Speed - Integrated circuits have a finite lifespan: each operation deteriorates the circuit an infinitesimal amount, so that doubling the number of cycles per second could cut the circuit's life in half. This alone is not usually enough to "break" a component before it becomes outdated, but speed also contributes to heat.
Heat - Circuits deteriorate more quickly as temperatures rise. Heat is also an enemy of stability, so that low temperatures must be maintained to reach a component's highest stable speed.
Voltage - Increased voltage allows for greater signal strength, which can have a tremendous effect on how far a component can be pushed. But increased voltage also increases circuit deterioration, and is the leading cause of early failure. Increased voltage also increases heat, requiring additional cooling improvements.
Solution 3:
When a manufacturer sells a part rated to work at a particular voltage or MHz, that means it has tested a statistically significant sample of a manufacturing run of the parts, and found that it works according to the MTBF, etc. they have specified and designed for.
Because they don't test every part of a run, you run the chance of getting a part that will work better than specified/designed, or worse than specified/designed.
Hardware wears out overtime, even if only very slightly. You are stressing the hardware more when you overclock it. It likely shortens the lifetime of the component even if you only overclock it a bit. However, you may luck out and end up with a part that can actually handle it for the expected duration you will use it for, due to variances in manufacturing. There is no way to reliably tell except by trying it.
@r.tanner.f is correct though. The safest thing to do is not to overclock.
Solution 4:
Yes, overclocking can cause hardware damage. Damage from overclocking is generally due to electrical overstress resulting from the circuitry being subjected to excessive temperatures or voltages. Wikipedia explains this in more detail:
Electrical overstress
Most stress-related semiconductor failures are electrothermal in nature microscopically; locally increased temperatures can lead to immediate failure by melting or vaporising metallisation layers, melting the semiconductor or by changing structures. Diffusion and electromigration tend to be accelerated by high temperatures, shortening the lifetime of the device; damage to junctions not leading to immediate failure may manifest as altered current-voltage characteristics of the junctions.
In particular, electromigration can occur when a processor is operated at higher than normal voltages for extended periods of time, especially if it is not adequately cooled. This typically manifests itself as instability (system crashes, benchmark failures) that progressively worsens, eventually causing instability at even stock clock speeds. For chips made on a 14nm process, 1.35V is generally considered the maximum safe voltage for long-term operation, though slightly higher voltages are usually okay for brief periods of time.
Note that controlled overclocking with proper cooling can increase performance without causing an unacceptable reduction in hardware lifespan (what constitutes "unacceptable" depends on the user). Since overclocking involves pushing hardware beyond its tested limits, its performance and reliability cannot be guaranteed; as such, damage caused by overclocking is generally not covered under warranty.