What kills a CPU/GPU the quickest? Temperature or clock speed?

[guymarshall]

I was wanting to overclock an FX 8350 to 5.5 GHz on a H110i GTX water cooler, and was wondering that, if the temperatures are fine, whether the processor will have its lifetime reduced due to high clockspeeds.

 

[Nuwan Fernando]

It might since you need to add more voltage to make it stable.

 

[Titanforall]

I was wanting to overclock an FX 8350 to 5.5 GHz on a H110i GTX water cooler, and was wondering that, if the temperatures are fine, whether the processor will have its lifetime reduced due to high clockspeeds.

Yes definately overclocking the CPU or GPU reduces lifetime of the components even by Increasing voltages you get a stable OC but the components become weak gradually overtime more faster .Temperatures doesnot kill CPU or GPU quickest but the higher clocks overclocked kills them sometimes in a second.

 

[guymarshall]

I was wanting to overclock an FX 8350 to 5.5 GHz on a H110i GTX water cooler, and was wondering that, if the temperatures are fine, whether the processor will have its lifetime reduced due to high clockspeeds.

Yes definately overclocking the CPU or GPU reduces lifetime of the components even by Increasing voltages you get a stable OC but the components become weak gradually overtime more faster .Temperatures doesnot kill CPU or GPU quickest but the higher clocks overclocked kills them sometimes in a second.

Ok thanks for the quick reply, so if I use the AMD catalyst control centre to overclock my processor using a slider, will 5.5 ghz damage it any more than 5 ghz with the same voltage?

 

[13thmonkey]

depends on voltage.

 

[guymarshall]

depends on voltage.

So in theory, any clock speed (if cooled to the right temperature that isn't too high) will work, but increasing the voltage to get to the clock speeds damages the processor?

 

[13thmonkey]

That's my view of the world, at the same temp and voltage a 4.5 or a 5.5GHz clock speed has the same life span. In real like you can't get to 5.5 without increasing the voltage.

As to what is 'too high', no one really knows. They'll have opinions but that's it.

 

[guymarshall]

That's my view of the world, at the same temp and voltage a 4.5 or a 5.5GHz clock speed has the same life span. In real like you can't get to 5.5 without increasing the voltage.

As to what is 'too high', no one really knows. They'll have opinions but that's it.

I currently have a strix 970 and am upgrading to an r9 390, so should I run it at 5 ghz or up to 5.5 with increased voltage and lower lifespan? I want to play games at ultra 1080p 60fps and not 1440p or 4k.

 

[13thmonkey]

consider what happens when it starts to die, BSODs causing possible disk write corruptions, it could take out the mobo, so you'll be looking for a new Mobo and CPU, probably for a dead socket.

That choice is yours to make, not mine, what's the risk worth to you, do you have the money to replace it tomorrow, do you have a good backup mechanism in place etc.

 

[guymarshall]

consider what happens when it starts to die, BSODs causing possible disk write corruptions, it could take out the mobo, so you'll be looking for a new Mobo and CPU, probably for a dead socket.

That choice is yours to make, not mine, what's the risk worth to you, do you have the money to replace it tomorrow, do you have a good backup mechanism in place etc.

Yeah I backup to external hard drives, and have multiple backups online of important things. Will enabling the clock speed to go up to 5ghz under load and drop to 1.4ghz when idling do any damage or will it increase the life span slightly? The voltage will also drop when the clock speed does.

 

[13thmonkey]

changing frequency not a problem, changing voltage (inside it's allowable voltage envelope) not a problem.

you are also stressing the Mobo VRMs too though, I had a good asus board go poof! and start smoking 5s into an OC that should theoretically have been safe.

 

[guymarshall]

changing frequency not a problem, changing voltage (inside it's allowable voltage envelope) not a problem.

you are also stressing the Mobo VRMs too though, I had a good asus board go poof! and start smoking 5s into an OC that should theoretically have been safe.

So does the constantly variable voltage damage the VRMs more than a constant high voltage?

 

[4Ryan6]

The fastest CPU or GPU killer is applying too much voltage.

Temperature and clock speed are longevity issues, because the CPU will protect itself temperature wise by dropping the clock speed if the CPU overheats, and if that is insufficient to drop the temperature to an acceptable level it will throttle back by other means, and if all that doesn't drop the temperature it will simply shut down your machine.

So as far as temperature and clock speed the CPU is well capable of defending itself against the most clueless overclocking idiot out there!

The GPU however does not have the same protection that the CPU does and it can be very much affected by temperature and clock rate, the greatest Graphics card crippler, is usually overclocking it without good thermal protection of the memory chips and GPU die, which is usually a factory out the door lousy quality control of the application of the cooling hardware on the graphics card.

There are always possibilities that too much voltage can be applied to a CPU or GPU and unfortunately that can be an instant killer.

Most GPU auto overclocking programs protect against overvolting into the danger zone, but you can always get around that!

Unfortunately most CPU overclocking motherboards allow voltages that can kill a CPU, which I would assume they design and build high end overclocking motherboards for the educated individual that knows how to use them, so that's why it is important to learn what you're doing, before you do it, and learn what your CPUs voltage requirements actually are.

My 2 Cents! Ryan

 

[guymarshall]

The fastest CPU or GPU killer is applying too much voltage.

Temperature and clock speed are longevity issues, because the CPU will protect itself temperature wise by dropping the clock speed if the CPU overheats, and if that is insufficient to drop the temperature to an acceptable level it will throttle back by other means, and if all that doesn't drop the temperature it will simply shut down your machine.

So as far as temperature and clock speed the CPU is well capable of defending itself against the most clueless overclocking idiot out there!

The GPU however does not have the same protection that the CPU does and it can be very much affected by temperature and clock rate, the greatest Graphics card crippler, is usually overclocking it without good thermal protection of the memory chips and GPU die, which is usually a factory out the door lousy quality control of the application of the cooling hardware on the graphics card.

There are always possibilities that too much voltage can be applied to a CPU or GPU and unfortunately that can be an instant killer.

Most GPU auto overclocking programs protect against overvolting into the danger zone, but you can always get around that!

Unfortunately most CPU overclocking motherboards allow voltages that can kill a CPU, which I would assume they design and build high end overclocking motherboards for the educated individual that knows how to use them, so that's why it is important to learn what you're doing, before you do it, and learn what your CPUs voltage requirements actually are.

My 2 Cents! Ryan

Thanks a lot ryan! I have overclocked my strix 970 to +200 on the core and +400 on the memory without changing the voltage, and temperatures have been really good. AMD Catalyst control centre uses 1.4 volts when overclocking and it jumps down to 0.8 when at 1.4ghz. So should I be ok to overclock it to 5.0 ghz with it going down when not in use? I have a beefy PSU and motherboard that can cope with high voltage and power draw.

 

[4Ryan6]

Thanks a lot ryan! I have overclocked my strix 970 to +200 on the core and +400 on the memory without changing the voltage, and temperatures have been really good. AMD Catalyst control centre uses 1.4 volts when overclocking and it jumps down to 0.8 when at 1.4ghz. So should I be ok to overclock it to 5.0 ghz with it going down when not in use? I have a beefy PSU and motherboard that can cope with high voltage and power draw.

That is totally up to you!

If you had seen what I have seen, you would never overclock a graphics card at all until you inspected under the hood to see how well the air cooled heat sink or water block was installed on the card.

My EVGA GTX Hydro Copper Titan, a $1300.00 graphics card, only had the GPU die 3/4ths covered with thermal compound, one corner had absolutely nothing on it, and 2 of the memory chips were only partially covered with thermal padding, Thankfully I did not do any heavy overclocking before hand.

We all assume what we get from the factory is absolutely what it should be and unfortunately with a graphics card you do not know until you pull the heat sink or in my case the water block and inspect under it.

This is not my first encounter either my previous setup was 2 EVGA 580GTX that I switched from air cooling to water cooling, upon removing the stock heat sinks one was installed perfectly, but the other looked like it was installed by a monkey, way too much thermal compound on the GPU die, and a couple of partially covered memory chips.

You would have thought I would have learned regarding EVGA, but previously I had seen just as bad from MSI, so the best rule of thumb is check for yourself before you do any serious overclocking of a graphics card.

 

[guymarshall]

Thanks a lot ryan! I have overclocked my strix 970 to +200 on the core and +400 on the memory without changing the voltage, and temperatures have been really good. AMD Catalyst control centre uses 1.4 volts when overclocking and it jumps down to 0.8 when at 1.4ghz. So should I be ok to overclock it to 5.0 ghz with it going down when not in use? I have a beefy PSU and motherboard that can cope with high voltage and power draw.

That is totally up to you!

If you had seen what I have seen, you would never overclock a graphics card at all until you inspected under the hood to see how well the air cooled heat sink or water block was installed on the card.

My EVGA GTX Hydro Copper Titan, a $1300.00 graphics card, only had the GPU die 3/4ths covered with thermal compound, one corner had absolutely nothing on it, and 2 of the memory chips were only partially covered with thermal padding, Thankfully I did not do any heavy overclocking before hand.

We all assume what we get from the factory is absolutely what it should be and unfortunately with a graphics card you do not know until you pull the heat sink or in my case the water block and inspect under it.

This is not my first encounter either my previous setup was 2 EVGA 580GTX that I switched from air cooling to water cooling, upon removing the stock heat sinks one was installed perfectly, but the other looked like it was installed by a monkey, way too much thermal compound on the GPU die, and a couple of partially covered memory chips.

You would have thought I would have learned regarding EVGA, but previously I had seen just as bad from MSI, so the best rule of thumb is check for yourself before you do any serious overclocking of a graphics card.

Ok cheers a lot mate, really helped me out! Yeah I will not overclock it without looking at the memory, as the normal temperature I stated is the GPU, something bad could be happening. Thanks for the advice though!

 

[4Ryan6]

Ok cheers a lot mate, really helped me out! Yeah I will not overclock it without looking at the memory, as the normal temperature I stated is the GPU, something bad could be happening. Thanks for the advice though!

In most situations Guy, you can get a powerful light and look down through the card to see if all the memory chips and voltage regulators are properly covered, but unfortunately you have to pull the heat sink to inspect the thermal compound on the GPU die.

If your memory chips and voltage regulators appear properly covered with the thermal padding you are probably OK?

 

[13thmonkey]

To be 100% clear, it is your risk to OC, your consequences to deal with. You seem to be wanting to to say that x/y/z will be safe. We cannot. We will not (well those of us that don't understand might). We can give you warning signs, but education will be more critical than that.

 

[4Ryan6]

To be 100% clear, it is your risk to OC, your consequences to deal with. You seem to be wanting to to say that x/y/z will be safe. We cannot. We will not (well those of us that don't understand might). We can give you warning signs, but education will be more critical than that.

Absolutely!

This is something you do not want to learn the hard way.

 

[13thmonkey]

To be 100% clear, it is your risk to OC, your consequences to deal with. You seem to be wanting to to say that x/y/z will be safe. We cannot. We will not (well those of us that don't understand might). We can give you warning signs, but education will be more critical than that.

Absolutely!

This is something you do not want to learn the hard way.

Learned once, no warning <5s and holy smokin VRMs batman!

 

[4Ryan6]

Me too, 13thmonkey, That's why I try to warn them, unfortunately only a few listen, but at least we have the few.

 

[grimsin]

I wanna see you get the 8350 to 5.5ghz! most people dont get close to that.

 

[guymarshall]

To be 100% clear, it is your risk to OC, your consequences to deal with. You seem to be wanting to to say that x/y/z will be safe. We cannot. We will not (well those of us that don't understand might). We can give you warning signs, but education will be more critical than that.

I was just wondering, instead of using AMD Catalyst Control Center to lock the voltage at 1.4 at any clock speed above 4.0GHz, should I change the clock speed in the BIOS to 4.5GHz and leave the voltage on "auto" and see what happens? Every time I tried to manually change the voltage before I had to clear CMOS and start again.

 

[13thmonkey]

In my, limited, experience, auto voltage always overdoes it, it has to so that it is stable on the worst of silicon.

 

[guymarshall]

In my, limited, experience, auto voltage always overdoes it, it has to so that it is stable on the worst of silicon.

Okay, I just found it strange how I got 4.6 GHz with only 1.3 volts... are you sure it is overdoing the voltage?