Frying by overclocking a GPU or CPU?

People talk about how it's dangerous to overclock a cpu or gpu in mobile. I've done both and had no problems. From personal experience and even in a Linus tech tips/as fast as possible video, he pretty much states that once a gpu or cpu gets too hot, it'll thermal throttle to protect itself. I've seen this happen with my Intel cpu's if i add too much voltage during an overclock. The speed just clocks down and in XTU, it tells me thermal throttling is happening when i get close to 100c.

So my question is, is it still possible to damage/fry out/melt/destroy a cpu/gpu in today's day and age? Wouldn't it just throttle itself down?

 

It can reduce the lifespan but it won't catch on fire if that's what you're asking

 

Yes it can happen, especially if chips are run on the edge over extended period, it just does not tend to be over a very short period.

 

Well an overclocked GTX 480 is good for frying eggs if that's what you're asking.

 

Yes it's possible.
Usually if a catastrophic overheating is detected (THERMTRIP#), around at 130°C on an Intel 4th generation CPU, the processor will shut down.
If the internal temperature monitoring sensor fails then permanent damage may happen.

 

I literally lol'd

 

Easiest way to kill a CPU or a GPU is jacking the voltage too high. Laptops basically always hit the thermal limit before the electrical limit that would damage them though. May not be the case for cooler running cards like the 970M but with the inferior power circuitry on the 980M, I'd bet it is easier to fry one of those through VRM stress.

Degradation generally occurs over time before failure though. A perfect example is Haswell's memory controller. When the chip has been pushed too far, its not uncommon for the CPU to suddenly go unstable in the memory department and then start to go unstable in the core a short period later.

 

Laptops also lack the means to push ridiculous voltages if IIRC. According to Mr. Fox the M18x R2 (or is it the AW18) has a board voltage limit of 1.5V, so you can't jack Vcore beyond 1.5V. It's still a dangerous voltage, but not quite "instant death" level.

And yes memory controllers seem to be the first to go from what I've read. In fact I think I degraded the IMC in my 4930K the first day I got my chip.

 

Here, this is how you do it. It's 8 yrs old cpu, newer ones have better thermal protection, but my small soldering iron is about 30-40w, less than cpu in my laptop.

 

So pretty much if I'm rendering a 30 min video and I overclock my cpu with 150 amps with the power current in XTU and 80w as the TDP, and it keeps hitting 100c, the worst I have to worry about is it downclocking and it won't damage it within that 30 minute period?

 

Over volting to the limit is what can damage a GPU. So many GTX 580Ms perished on these forums.

 

As long as you didn't touch voltage and the CPU is thermal throttling properly, no.

 

The M18x R2 and AW18 I think can both do it. As far as Ethrem checked; clevos top out under 1.3v. I can't remember if it was 1.25v or if he never tested the limit there, but 1.2v isn't out of reach I'm sure. It's still a low limit for CPU though.

And stop getting bad hardware in your desktop and making it worse D=.

 

I know for sure my P370SM could do 1.25V, but never tested beyond that because well, H asswell runs ass hot, plus 1.25V was the most I needed to max out the turbo bins on my 4900MQ and have it stable.

When I upgrade to Skylake-E next year, I'm either gonna get a pre-binned chip from Silicon Lottery, or just FAQ it and buy a Xeon because they use only top of the grade dies. There's anecdotal evidence to suggest (that's 4 separate links!) that 1P Xeons have their multipliers unlocked, so they should be fully overclockable on a supported board. And last I checked, Asus' consumer boards on the 2011 socket all supported 1P Xeons.

At least the overclockable Xeon part held true from Sandy through Haswell, so I have hope that Skylake Xeons will also be unlocked. Well we'll see whether I still think shelling out an extra $500-1000 for "almost guaranteed golden sample" is worth it 14-18 months down the road. For my own sake I hope not LOL

 

And this is why I want that EM heatsink setup. With CLU. I'm dead sure it'll handle 4.2-4.3GHz on 4 cores with that thing and some cooling mods.

I am also convinced my slave GPU heatsink is a gift from the gods because I can't make the GPU hit 80 degrees in any game. I will be sad to see it go.

(let's be real I'm gonna have a 4800MQ and two 780Ms for like the next three years)

 

So you saying that the p370em has better cpu heatsink than any aother p37..sm-a??
Can i put it in my laptop?

 

Yes. You will sacrifice cooling on your slave GPU, but considering you have 970Ms, you probably couldn't overheat those if you wanted to, so it should not be a problem.

 

I ran my 920xm at 1.45V at over 4ghz for over 6 months 24/7. Still kicking strong. M15x board does start having issues with the high voltage though above 1.4V.

 

how do i know what voltage im at in xtu? my 4810mq is stable between 90 and 110 A. i assume the a is short for amps? i don't know the amps to volts conversion or anything. i just move the sliders around until i get a stable clock. what is a safe range to put the TDP and the power current limit?

 

TDP and current limit can be raised indefinitely as long as your heat is fine. A is indeed short for amps.

As for the voltage, you can get an idea of your voltage by looking at the vcore in HWiNFO64. Basically if you don't touch voltage, the CPU will just pull what it needs on its own. It's always too high though, most Haswell chips can sustain a -80mv undervolt and remain totally stable.

 

High temperature determines this. Turn up, so long as the temperature is not a problem.

 

and if temps become a problem, the worst i have to worry about it is just some thermal throttling?

 

High temps degrade the chip over time but its anybody's guess how long it will take. If you're over 90C for a sustained period, that chip is going to fry fast. Its because of how hot Haswell runs that I save the punishing workloads for my desktop.

 

I said as long as the temperature is ok Processor shall not throttle. Then it is too hot ...

 

Im hoping that running my titan x overclocked won't fry it sooner temps are about 80 while gaming so I think it should be ok

Sent from my Nexus 5 using Tapatalk

 

Meh, you'd have to run 90C 24/7 for a long time, like for folding or something, before it would kill the chip. Gaming for a couple hours at a time, unless you plan on keeping the laptop for five years or more, it really shouldn't matter.

 

I've witnessed Haswell degradation first hand, it doesn't take that long and it doesn't seem to depend on heat. My 4770k went down fast. My 4790k is faring much better but the whole point is that there are so many on die components besides the CPU cores that can cause issues.

 

No problems here with my Ivy Bridge chip.. Runs at 85-90C++ in games for 3-4 hours a day.. In fact with more volts, the performance seems to get better lol..

 

Ivy Bridge will burn at 105C without complaining. Haswell is a lot more sensitive, probably because of the FIVR.

 

i only know the Intel Extreme Tuning Utility way to overclock laptop CPU and get extra 0.2 GHz .. is there any other way?

if there, it's known that changing voltage can lead to unwanted stuff

i have i7 4712hq

 

0.2ghz is nothing, you're safe as long as max temps stay below 90

The HQs don't boost over 45w anyways.

 

would you please share a link for a tutorial to overclock laptops' CPUs? i need to know more about it

made a google search but found nothing except the intel utility and it only overclock 0.2 GHz by increasing processor multipliers up to their max including the cache ratio . "don't know if that's different for other CPUs"

sorry if the comment is stupid but i am new to these stuff and i really want to know more about it

thanks

 

0.2GHz only, further OCing only available in MX series processors.

 

you're welcome! http://forum.notebookreview.com/thr...ly-have-it-run-full-stock-turbo-speed.734696/

 

Not exactly true. The 4800 series should have a +400MHz and 4900 series has a +600MHz allowance. You don't find them selling in the HQ line much though, and the HQ line doesn't listen to bios TDP limits so it is indeed moot.

 

Have you seen a Hq processor that can go over TDP (Watts) and actually holding Their Stock turbo clocks during max load? or even overclocked?

 

That's why I said it was "moot" XD.

Though apparently one user has shown a 4710HQ that listens to power limits. I'm still not convinced it's widespread.

 

whats your graphics score on 3d mark 11 P? any game benchmarks?

 

"If" there is one, then there are not many ... The vast majority that have a BGA processor must reduce the core voltage so that the processor shall be able to run over the base clock freq under maximum load...

 

Nowaday, chips got too much protection and locked down, back then you can OC CPU for 20-30++% performance (PLL mod etc etc).

Stuff can really get fried back then, not anymore.~

 

Anyone done pencil lead volt mods on desktop GPUs back in the day? LOL

 

Don't mean to hijack but a few questions. Firstly my temps when gaming on both CPU and GPU are around 85-90C, is this anything to be worried about when gaming for up to 2 hours? Also how much difference will better thermal paste make?

 

Up to 5-10 degrees... Depends on how bad it is from before. The application of the paste is just as important. 85-90 degrees is no crisis but it's going two get worse the older and dried the pasta is going to be ..

 

Nope.. But that's always intrigued me. About the most involved I got with overclocking back in the day was physically changing jumper settings.

 

I think I tried the pencil mod on the Thunderbird Athlon, but definitely nothing with the GPUs.

 

And depends on if it's fettuccine or spaghetti too...

 

LOL

 

I think he meant paste lol..

 

Are more familiar with liquid ultra .. Someone has tested with toothpaste .. Why not pasta?