How to know if delidding a 4770k is needed?

I recently swapped cpu's between my gaming pc/htpc and my "work" pc that really doesnt need a lot of cpu power. The 4770k went to the gaming pc and the i5-4430 went in my work pc.

Right away after doing the swap I noticed while playing a game that the cpu fan was so much louder on the 4770k than the 4430. My first thought is to change the HSF but thats a pain in a tiny ITX box when I have to disassemble a lot of stuff to get the mobo out so I can remove/replace the back plate.

So I instead went into the BIOS and changed the max speed of the CPU fan to 62.5% rather than 100%. That seems to have made things quiet enough and surprisingly the temps didnt go up much.

While this temp fix will work for now, I'd like to know how to tell if I need to delid. Can someone point me in the right direction? I would like to OC someday although its not needed right now. Also, if I got a 4790k, would I have to delid it also? If so, then I have no interest in upgrading anytime soon.

These chips don't necessarily "need" to be delidded at all. During gaming and normal operation they tend to peak out at about 65C. They rarely stay at high temp for very long, either. Not unless you are using synthetic tests like prime95 to send constant heavy computations--for the most part your 4770k will be done with whatever it's thinking about before it gets a chance to get much hotter than that.

The really great thing about delidding (or any temperature gains you can make by modding that actually reduce the amount of heat produced) is that the difference in heat is essentially removed from the equation instead of just being cooled more effectively and then having a corresponding amount of hot air dumped outside the cooler.

The power consumption stays the same... but the heat produced is sharply reduced. That's free energy and it's something we rarely get a chance for.

The 4790k is a little cooler than the 4770k clock for clock but still wants to be delidded for optimal cooling. Is it necessary? Absolutely not. Is it more efficient and comfortable to use? Definitely. Less heat dumped in my room is better. I would not recommend a 4790k upgrade from a 4770k unless you are stuck with the 4770k at a disappointing overclock that is keeping you up at night. The gains you'll see, even in that scenario, aren't worth much discussion. I have a 4.6Ghz 4770k and a 4.8Ghz 4790k and they are basically the same, but the Devil's Canyon is about 10C cooler at same speed as the Haswell. That point quickly becomes moot as the DC chip clocks higher and gets hotter overall.

Bluesun311 said:

These chips don't necessarily "need" to be delidded at all. During gaming and normal operation they tend to peak out at about 65C. They rarely stay at high temp for very long, either. Not unless you are using synthetic tests like prime95 to send constant heavy computations--for the most part your 4770k will be done with whatever it's thinking about before it gets a chance to get much hotter than that.

The really great thing about delidding (or any temperature gains you can make by modding that actually reduce the amount of heat produced) is that the difference in heat is essentially removed from the equation instead of just being cooled more effectively and then having a corresponding amount of hot air dumped outside the cooler.

The power consumption stays the same... but the heat produced is sharply reduced. That's free energy and it's something we rarely get a chance for.

The 4790k is a little cooler than the 4770k clock for clock but still wants to be delidded for optimal cooling. Is it necessary? Absolutely not. Is it more efficient and comfortable to use? Definitely. Less heat dumped in my room is better. I would not recommend a 4790k upgrade from a 4770k unless you are stuck with the 4770k at a disappointing overclock that is keeping you up at night. The gains you'll see, even in that scenario, aren't worth much discussion. I have a 4.6Ghz 4770k and a 4.8Ghz 4790k and they are basically the same, but the Devil's Canyon is about 10C cooler at same speed as the Haswell. That point quickly becomes moot as the DC chip clocks higher and gets hotter overall.

Click to expand...

How does delidding reduce the amount of heat produced by the CPU in the same usage?

It doesn't, it just allows it to be dissipated more quickly.

Bluesun311 said:

The power consumption stays the same... but the heat produced is sharply reduced. That's free energy and it's something we rarely get a chance for.

Click to expand...

Not quite.

Pylor said:

It doesn't, it just allows it to be dissipated more quickly.

Click to expand...

Yup.

Same amount of heat being produced.

Bluesun311 said:

These chips don't necessarily "need" to be delidded at all. During gaming and normal operation they tend to peak out at about 65C. They rarely stay at high temp for very long, either. Not unless you are using synthetic tests like prime95 to send constant heavy computations--for the most part your 4770k will be done with whatever it's thinking about before it gets a chance to get much hotter than that.

Click to expand...

With the same HSF and the same game, would you expect the 4770k to run noticeably hotter than a i5-4430? And in turn this would make the fan louder... Is it the entire Haswell line that has heat issues or just the 4770k?

cardboardbox said:

With the same HSF and the same game, would you expect the 4770k to run noticeably hotter than a i5-4430? And in turn this would make the fan louder... Is it the entire Haswell line that has heat issues or just the 4770k?

Click to expand...

The entire Haswell line has a theoretically worse thermal interface between the integrated heat sink and the CPU die; but this is not the issue you're encountering.

Any core series (i3,i5,i7) will have similar dissipation characteristics.
The difference will be only in the amount of heat each one is capable of producing. more powerful CPU's expect to be able to dissipate more heat as needed.

For literally the exact same load on the CPU you should expect, pretty much, identical heat.
There is no "heat issue".... stock Haswell with the stock HSF operating in its intended capacity works just fine.

Your 4770 is capable of more power than your 4430... so it's likely that you're not taking on the same load. Nonetheless, if you want quieter cooling you can get quieter fans.....
The computer's CPU has a rated TDP. Its HSF is specced to that TDP.
You added a more powerful CPU, with a higher TDP.... it's absolutely reasonable for the cooler to get louder.... because your cooler was not designed for the thermal load you're putting under it.

this has absolutely nothing to do with anything wrong with the CPUs and everything to do with the fact that you increased thermal load.

first i have to ask what type of cooler are you using in your machine.. if stock be sure that you are using the correct stock i7 cooler, the i3 and i5 have one type of stock cooler and the i7 have othe type, for the i3 and i5 they use aluminum core and for the i7 coolers they use copper core.. so the stock i7 cooler do a much better job keeping it at good temperature at stock speeds for testing purposes i tested a stock i5 2500 cooler in a i7 2600 and the temps increased dramatically at the point of when just gaming make the i7 peak at 80C depending on the game and with the stock cooler it peaked at 60C under same conditions..

delid worth always, as a [H] community each degree of temperature less for more performance its a win.. most 4770k tend to cap at 4.2ghz-4.3ghz before break the thermal safe point.. delid allow people to reach higher overclock lests say 4.5ghz-4.6ghz and still have a controllable voltage with good temperatures.. my first 3770k was a crap.. 1.256v to be stable at 4.2ghz reaching 90C under P95, i was almost going to delid then i just swapped it for a new one that required 1.240V for 4.5ghz to be stable and peaking at 80C the hottest core with the others at mid 70s under prime 95.. this i think its a good chip to delid and see how much i can push it before the voltage or temperature become a problem and maybe i will do the process of delid it soon, however i consider it good enough as while im gaming im peaking low 60s with a average in all cores in the 45C if i plan to keep my 3770k at 4.5ghz then i may not even think in delid however i want to see how far I can make it run and that's its generally what people may consider before delid i would be happy if i reach 4.8ghz in a similar range of temperature that what im now at 4.5ghz if the voltage are also safe..

thanks for the help guys. I'm really starting to think the problem is my HSF. Here is a link to it: http://www.newegg.com/Product/Product.aspx?Item=N82E16835233081

It was perfectly fine back on my 2500k so I guess I could be wrong.

I just dug up the stock Intel coolers for both cpu's. They look very similar, both have copper, fins on the i7 cooler are slightly different. I'm going to disassemble my ITX box sometime soon and see if the stock cooler makes a difference.

It is.

Zangmonkey said:

How does delidding reduce the amount of heat produced by the CPU in the same usage?

Click to expand...

Not sure and apparently it's wrong! I'm not a physicist, obviously.

My anecdotal experience was that the chip runs cooler after delidding but uses the same amount of power.

I didn't mean to suggest that one can create energy from nothing... but removing the IHS and taking the heat gap out of the equation is input "energy", too.

Consider the "heat island" effect cities like Atlanta, Georgia experience. These metro areas are often as much as 15 degrees Celsius hotter than surrounding rural communities (the difference is highest in the evening). Why? Because of the density and properties of the materials. They receive the same amount of radiation from the daylight sun, but cool at different rates, such that the city has a negative feedback loop of "heating itself" (dissipating slowly) long past sundown. Planting trees or changing building materials has a big impact on this problem for cities.

Haswells have a "heat island" that is the gap between the IHS and die. It can vary and it effects cooling potential in such a way that removing the gap is a lot like replacing asphalt parking lots with fields of grass. Are you creating energy from nothing? No. Are you making it cooler in your room? Yep. Would using a better cooler be the same? I don't think so, but my logic may be wrong. A better cooler will just dump the heat away from itself faster. A better package means less heat being generated to start with.

Different power supplies with different efficiency ratings produce more or less heat at the same output power right? I could be wrong about that too... It's certainly not a "basic" concept.

Bluesun311 said:

Not sure and apparently it's wrong! I'm not a physicist, obviously.

My anecdotal experience was that the chip runs cooler after delidding but uses the same amount of power.

I didn't mean to suggest that one can create energy from nothing... but removing the IHS and taking the heat gap out of the equation is input "energy", too.

Consider the "heat island" effect cities like Atlanta, Georgia experience. These metro areas are often as much as 15 degrees Celsius hotter than surrounding rural communities (the difference is highest in the evening). Why? Because of the density and properties of the materials. They receive the same amount of radiation from the daylight sun, but cool at different rates, such that the city has a negative feedback loop of "heating itself" (dissipating slowly) long past sundown. Planting trees or changing building materials has a big impact on this problem for cities.

Haswells have a "heat island" that is the gap between the IHS and die. It can vary and it effects cooling potential in such a way that removing the gap is a lot like replacing asphalt parking lots with fields of grass. Are you creating energy from nothing? No. Are you making it cooler in your room? Yep. Would using a better cooler be the same? I don't think so, but my logic may be wrong. A better cooler will just dump the heat away from itself faster. A better package means less heat being generated to start with.

Different power supplies with different efficiency ratings produce more or less heat at the same output power right? I could be wrong about that too... It's certainly not a "basic" concept.

Click to expand...

No.

Removing the IHS and placing a HSF directly on the die simply provides a more efficient interface to move heat away from the chip.

A "de-lidded" 130W processor still puts out the same amount of heat as one with it's "lid".


UHI is caused by heat being trapped by man made development in things like asphalt and concrete as well as additional sources of heat from activity. As the surrounding natural landscape cools overnight, the city stays warmer longer due to this.

efficiency in PSU its not so a difficult concept, a high efficient PSU just waste less energy as heat making it to produce less heat at the same amount of load.. less energy consumed less heat of course. so yes different power supplies with different efficiency ratings produce more or less heat at the same output power.. however in a processor the thing its not the same as the CPU its not producing energy its eating energy a delided CPU will generate the same amount of watts so its the same heat produced as a non-delided (unless the heat its extremely high and its making to leak a lot of voltage consuming more power and of course pulling more watts, watts = heat ) taking the case of my 3770k under p95 small FFT test it draw 80C watts continuously at the same 1.240v no matter what type of cooler I use the 80W will be the same if i delid it, it will produce the same 80W at the same load with the same voltage, but now those 80Watts of energy will be drained fast from the CPU and be transferred fast by the Cooler the heat its exactly the same, are producing less heat the chip because its delided? the big answer no, its producing the same amount of heat but now its just dissipated faster, the same 80 Watts are there consumed and thats its heat generated but now instead all of that heat being trapped under the lid and the internal die it will be just pulled faster.. will be the same 3770k cooler? of course, are producing less heat? nope..

here are my results after I delid my 4770k: http://gyazo.com/10fe437f33ac1da8608f70ae4c75b227

before that, I was getting temps up in high 70's.

You can calculate how much heat to be expected from a lightbulb at a given wattage. But you need to know what kind of lightbulb it is. Incandescent has bad efficiency so you'll get 98% heat and 2% light. You can get a different light type that produces the same amount of light for the same wattage--but puts off less heat. Yes the surrounding materials will cool faster when you turn it off. So, how is that different from when you are trying to calculate how much heat a given processor will produce at a certain wattage? You need to know what type of processor it is... and in the case of delidding if it is or isn't.

I'm willing to give you guys the benefit of the doubt that I'm missing something important but for a not so terribly complex concept a proper explanation has not been given at all.

The power supplies are different in efficiency, so yes making the same amount of power means using different amount of power from the wall and producing different levels of heat. But our 2 processors are identical except one is delidded. They use the same amount of power and produce the same results in the same amount of time. But they don't take the same amount of time to cool. Will that make any difference in room temperature? Yes. Just like if you change your drapes from black to white. You aren't changing the amount of sunlight hitting them--just the materials. Yet your room is cooler. Where is that heat that has been removed from the equation? It just got bounced back into the atmosphere--so the efficiency of the room your measuring has changed.

When you change a processors packing materials, like delidding, you change its efficiency. You change the amount of heat produced per watt.

TroyX said:

here are my results after I delid my 4770k: http://gyazo.com/10fe437f33ac1da8608f70ae4c75b227

before that, I was getting temps up in high 70's.

Click to expand...

wow thats impressive. I dont have complete confidence in my ability to delid so I'm going to see what happens when I put the stock Intel cooler on. I just got a good deal on a GTX 770 off ebay and when I get it I'll tear into my rig and do some testing on the cooler while I'm already in there. If swapping to a couple different coolers doesnt solve my noise issue I'll research some more on delidding.

There's two general ways to do it, with a razor blade or the vice method. The razor will need more finesse, but smashing a cpu edge with a mallet can't be beat. If you do the vice method, watch some youtube videos. Don't be that guy who puts the PCB in the vice and tries to smack the IHS off, you'll probably end up ruining the cpu from flexing..

Delidding is probably about as easy as changing coolers and far more worthwhile. The thing is, the first time you do it is unfamiliar and nerve-wracking no matter how many of those videos you watch. Copying someone else completely is possible, but for the most part it's steady hands and confidence that make it go smoothly.

About my conservation of energy problems: I do think you guys are right... I just can't really figure out why exactly. I can think of a good check, though--the energy cannot simply disappear... so if the 2 identical procs are using the same amount of power, they must also output that energy consumption in some form. If the heat output of one of the processors is less, where is the rest of the energy exiting the system from and in what form? It's certainly not making light and it isn't making sound... seems like my argument about less heat being produced by the delidded processor kind of falls apart there. Still, faster cooling seems beneficial to room temperature somehow... but I'm aware changing from air to water cooling isn't somehow going to reduce the amount of heat your system produces.

Bluesun311 said:

When you change a processors packing materials, like delidding, you change its efficiency. You change the amount of heat produced per watt.

Click to expand...

Nope, the heat generated its the same, just its being transferred to the cooler faster.. you change its efficiency only to transfer heat from a point to other point. in this case from the internal die of the CPU to the Heat Spreader. the thing with the TIM used by intel its exactly that word "Efficiency" its extremely bad to transfer the heat produced by the Processor to the Heat spreader its holding the heat there into the CPU package accumulating heat making it to reach certain temperatures, when you eliminate that accumulation of heat the chip will reach lower temperatures because the heat produced its being expelled fast from the die, its just like for example a CPU cooler a direct contact HeatPipe have the ability to extract the heat faster from the CPU than a cooler with no direct contact making it more efficiently to dissipate the heat however the heat produced by the CPU its the same, when you delid and change the TIM for a high quality one its like change the stock cooler for a aftermarket cooler, the heat produced by the CPU will be the same but now the CPU run at a lower temperature its magic? nope, its just a more efficient cooler that remove faster the heat but the heat produced its the same so you aren't changing the heat produced by the CPU, same with direct die cooling, when you completely remove the heat spreader from the equation, will it make the CPU produce less heat? nope just its being removed fast by the cooler..

Bluesun311 said:

Delidding is probably about as easy as changing coolers and far more worthwhile.

Click to expand...

ok you got me interested again.

I have some thermal paste here but not the Coollaboratory Liquid Ultra. If I'm going to delid do I really need to get the Liquid Ultra? The best one I have right now is Antec Formula 7 Nano Diamond Compound.

I switched to the stock Intel cooler today and my temps while playing a game went from a max of about 70 to 82. Fan was about equally obnoxious as the Xigmatek. Not quite as loud but changing speeds constantly and with a high pitched whine.

So I think 82 degrees is enough evidence that I need to delid. Or sell the cpu and move on to something else. My i5-4430 must not get hot because the cooler is whisper quiet compared to the what I've tried on the 4770k.

No I really don't think I necessarily agree that you "need" to delid but I'll elaborate. First off, getting a better cooler will help immensely. Just because the xigmatek is 12C better than the stock cooler doesn't mean it's not going to make a huge improvement getting a better cooler.

Just this morning I finally put on the Noctua NH-u12s I've had sitting around and it's much much more effective than the Corsair H55 on a hot chip... with only 1 fan on it. At idle, doesn't make much difference, but at load you will see huge gains going with a different heatsink, whether or not you delid, unless all of the problem is with the IHS, and I doubt it highly. A lot of times I feel like the mounting system is about as important as the fin density and all that crap. The Noctua mounting system is fantastic (at least on 115x where all of my experience is) and so I use their heatsinks a lot since I know what to expect. But there are equally as effective coolers out there that aren't 70$.

I make it really hard to resell my shit when I talk about what I do to it, but these chips can get this hot (and way hotter. they can function without making mistakes while throttling from 105C) because you kinda have to stab the die with a knife to hurt them. People that kill chips are either: doing insane professional type overclocking/testing and TRYING to kill them by disabling all the voltage protection and thermal protection or they are delidding with the razor method. Smart people that make awesome threads I've read about killing 2 chips in a row and then finally get it right on the 3rd.

If I can delid a chip, anyone can. My hands were shaking, but the vice hammer method is stupid easy. I love my 25$ vice I got to do the job with, I've used it for several other things since, and you could seriously do the job real quick and then take the damn thing right back to Home Depot if you were hurting for cash. The ? is, how to mount the vice. I happened to already have one of these:

http://www.amazon.com/Keter-17182239-Folding-Work-Table/dp/B001CWX26Y/?tag=toolguyd-20 I think we got it at CostCo for less than that though. Looks like:

and I mounted the vice on the bottom tier sidebar, with the folding cross support "X" right behind the vice as a backsplash. I put some paper and stuff behind the vice to catch the chip if it fell but if you do it carefully when the glue is warm you may be able to just pop the lid off without hitting it so hard when you feel the glue start to give. Elsewhere people say heat doesn't change the glue's properties.

Most important thing is to choose the proper striking block. Even if you don't have the hardest wood available, if you can find a piece of solid wood that has a knot in it--use the knot because it is very dense and hard wood.

As for the CLP, it's not necessary but it helps a LOT, especially when used between the die and the IHS--it scales at higher temperatures is my belief now that I'm getting higher clocks with the DC I got. I prefer CLP to ultra, ultra is a ripoff because it doesn't spread easily and the amount you get is just paltry compared to the CLP syringe--but it does seem to be easier to remove and better performing. I am not a Noctua fanboy but I find NT-h1 to be dirt cheap and fantastic. Lately it seems to be a little watery, though, so I'm thinking about switching to something else for non-conductive TIM. Oddly enough i think that Antec stuff is quite good but I have seen the rumor mill saying diamond compounds are a no-no on the bare die. Look around the net about this stuff to make sure first.

The CLP won't kill your chip even if you repeatedly remove it and reapply, but it will damage the finish on the die after even 1 removal. Interestingly, this does not affect performance in any way, but I suspect at some point repeatedly removing it and applying it a lot wouldn't be smart... depending on how hard you scratch the chip with the CLP particles, you may as well have stabbed it. That's not to scare you off from using it, but just be really gentle with it and make sure you are set to use it and you aren't going to be changing things around again the next day. My personal recommendation is to use liquid electrical tape on the whatever components (capacitors? i thought they were vrms) on the top of the chip next to the die, put the IHS on top of CLP spread thin on your die and then try your best to never take that ihs off again. Some people reseal them. Seriously, you kinda have to stomp on a Haswell to kill it.

If you have high temps at low voltage. Put the chip at stock clock and 1.3v on air. What is your temp during stress. If it is high, delid. you'll get a couple degree reduction in temps.

20c reduction under heavy stress from delidding. Even more sometimes.

bluesun311: I've looked at a lot of videos today about how to delid. My dad has a vice but it wont rotate to vertical so its possible the cpu goes flying off when it finally separates. Also, his vice is just metal, does not have plastic or rubber on the grip. I'm afraid the metal might damage the IHS. But if it does scratch it will be on the top side of the IHS. Do you think this would be an issue?

LoneThrasher: I checked my BIOS options and there's a lot of voltage options. The one labeled CPU Core Voltage was 0.9something. I thought going up to 1.3 was a big change and didnt want to take a chance of damage. Why would increased voltage help my temps?

First off that post is just spam. 1.3v isn't going to cause anything to explode, so it got left alone. At stock, .9 something is probably a good sign about your CPu's headroom. That's low voltage though and in fact 1.3 is what I typically run at but you're right that that won't help with temps. You're CPU can run super super low voltage too, but at full tilt it probably regularly goes up to 1.3v on its own without you even knowing. That's because in Auto voltage mode Haswell's internal voltage regulator will just "guess" what kind of voltage your chip might need based on an algorithm. Typically, to be safe and stable, that guess is WAY off, and WAY high. If you get to know your chip really well you can override the IVR and make your chip run at a capped voltage max instead. You can even do this while allowing it to idle at almost no voltage at all. But I don't know your BIOS so I can't tell you how exactly. As it is, I barely understand how to do it with my own board's BIOS. But, most likely it's exactly the same for AsRock and ASUS. Asrock seems to pay ASUS for their software and rename it, that's my guess anyway.

The most important thing to know about this whole voltage issue is that Manual voltage is best unless you are on a laptop. Especially for synthetic benchmarks. Prime 95 or AIDA64 type stuff will definitely cause the FIVR to overvolt your chip more than you really need in AUTO or Adaptive/Offset type voltage modes from your BIOS. I have seen a simple 3D benchmark like Unigine Valley ask for 1.36volts from my 4770k in adaptive mode.

This is the vice I went and got:
http://www.amazon.com/IRWIN-226303-3-Inch-Clamp-on-Vise/dp/B0001LQY44

Forgot to mention that part.
Was literally like 20 bucks and it's like it was made for this. You want the chip to just fly off really. That or barely separate. What you can do is put some electrical tape on the teeth of the vice to keep it from scratching the IHS but I don't bother. They are already chipped from the retention bracket when you install them anyway. And you can crank the vice down on that IHS good and tight, it wont snap or bend. If you start sanding the IHS you'll see how solid it is. It's nickel plated copper I believe. Running naked doesn't really help much.

Same with the rest of the space wherever you expect the chip might hit something you can put some tape, you could even put some packing material/bubble wrap or something around like that. I put an old shirt over a towel or something down on the floor as well. But you want to be working as low as possible. You really don't need to worry too much about static or anything like that. I mean... anything could go wrong so you want to be prepared for the worst, but if you've ever built anything and know how to drive a nail without taking off your thumbnail you can do this.

Do it. Just be very careful. YouTube has a bunch of guides. I would defiantly upgrade to aftermarket cooler before though.

cardboardbox said:

ok you got me interested again.

I have some thermal paste here but not the Coollaboratory Liquid Ultra. If I'm going to delid do I really need to get the Liquid Ultra? The best one I have right now is Antec Formula 7 Nano Diamond Compound.

Click to expand...

My answer, based on my delid, is yes. I used my preferred paste (MX-4) between the cores and IHS, and during initial testing everything seemed fine. After about a month of use however, I decided to try running Prime again to make sure that things were as they should be, and they weren't. I was spiking to 100c almost instantly on 3 out of 4 cores, so I bit the bullet and bought CLU. After using CLU between the cores and IHS, not only did I notice another reduction in temperature, but a month afterwards I saw the exact same numbers as I did right after application. I won't use CLU between the HSF and IHS, but I will always use CLU between core and IHS and nothing else.

gah, had a longer response typed out but it got lost, token expired or something...

Anyway, thanks for the pep talk and advice, especially bluesun311.

I'm a little concerned about what CLU does to the cpu long term. And I really dont want to have to reapply every 6 or 12 months.

I'm low on free time so not sure I want to do this. I still might just sell the 4770k and use my i5-4430 which is very quiet with the same HSF that is noisy on the 4770k. I'll decide within 2 days so I can order the CLU or CLP in time to use this weekend.

You even have a backup chip... if you walk away it will keep you up at night down the road!

Araxie said:

Nope, the heat generated its the same, just its being transferred to the cooler faster.. you change its efficiency only to transfer heat from a point to other point. in this case from the internal die of the CPU to the Heat Spreader. the thing with the TIM used by intel its exactly that word "Efficiency" its extremely bad to transfer the heat produced by the Processor to the Heat spreader its holding the heat there into the CPU package accumulating heat making it to reach certain temperatures, when you eliminate that accumulation of heat the chip will reach lower temperatures because the heat produced its being expelled fast from the die, its just like for example a CPU cooler a direct contact HeatPipe have the ability to extract the heat faster from the CPU than a cooler with no direct contact making it more efficiently to dissipate the heat however the heat produced by the CPU its the same, when you delid and change the TIM for a high quality one its like change the stock cooler for a aftermarket cooler, the heat produced by the CPU will be the same but now the CPU run at a lower temperature its magic? nope, its just a more efficient cooler that remove faster the heat but the heat produced its the same so you aren't changing the heat produced by the CPU, same with direct die cooling, when you completely remove the heat spreader from the equation, will it make the CPU produce less heat? nope just its being removed fast by the cooler..

Click to expand...

Actually he's right due to the physical properties of semi-conductors and how their internal resistance changes with temperature. What it boils down to is, the cooler they run, the less energy they consume, everything else being equal. So getting better cooling does indeed lead to less heat being produced. That said, the temperature difference we are dealing with here would yield in a negligible decrease in power consumption. From a technical perspective, he is correct. From a practical one, you're not necessarily wrong.

cardboardbox said:

I switched to the stock Intel cooler today and my temps while playing a game went from a max of about 70 to 82. Fan was about equally obnoxious as the Xigmatek. Not quite as loud but changing speeds constantly and with a high pitched whine.

So I think 82 degrees is enough evidence that I need to delid. Or sell the cpu and move on to something else. My i5-4430 must not get hot because the cooler is whisper quiet compared to the what I've tried on the 4770k.

Click to expand...

Are you sure you're not overvolting it?

If not, then you likely have a defective CPU and you should return it.

Zangmonkey said:

Are you sure you're not overvolting it?

If not, then you likely have a defective CPU and you should return it.

Click to expand...

I'm way past the return date by about 7 months. Would Intel replace the cpu under warranty just because it gets hotter than the i5-4430 and makes my HSF roar?

cardboardbox said:

I'm way past the return date by about 7 months. Would Intel replace the cpu under warranty just because it gets hotter than the i5-4430 and makes my HSF roar?

Click to expand...

If it can't run at stock clocks on the stock cooler, they'll replace it. If it can, you're out of luck.

Tsumi said:

If it can't run at stock clocks on the stock cooler, they'll replace it. If it can, you're out of luck.

Click to expand...

it definitely runs, never had any instability. Its just that it makes the cooler blow really hard while the i5-4430, doing the exact same task and using the same HSF, stays very quiet.

cardboardbox said:

it definitely runs, never had any instability. Its just that it makes the cooler blow really hard while the i5-4430, doing the exact same task and using the same HSF, stays very quiet.

Click to expand...

All Intel would care about is whether or not it runs in spec or not. Warranty is for defective parts. NOT "not performing as good as other parts" parts.

You could also (If the mobo supports it) Lower the stock voltage. Typically they stock voltage is set pretty high, and often times un-necessarily high. Lower it a notch or two and your heat / temps / fan speeds will all drop.

My fan on my heat sink is hooked directly to the power supply....i could never stand the sound of the fan always changing speeds....plus all my fans have in built in speed switches for high/med/low....my advice is get a quiet fan that doesn't change speed...but moves a lot of air being quiet

cardboardbox said:

ok you got me interested again.

I have some thermal paste here but not the Coollaboratory Liquid Ultra. If I'm going to delid do I really need to get the Liquid Ultra? The best one I have right now is Antec Formula 7 Nano Diamond Compound.

Click to expand...

Depends if you want to put the IHS back on or use a direct die mount. If i went as far as delidding I'd be direct die cooling as well.