How best proceed with overheating i7-4790K?

[REnso1]

I have an i7-4790K in a GA-Z97MX Gaming 5 mobo latest F4 BIOS.

I used a Noctua NH-L12 cooler rated at 95W for cooling the processor rated at 88W in a Lian Li PC V354 with 4 fans, 2 in 2 out, case closed and case open, ambient is 27-30°C.

Temperatures in BIOS and memtest86+ were high so I decided to try stress testing and in Prime95 small FFT cores 1&2 overheated to 100°C using Core Temp.

I tried reseating the heatsink and renewing the NT-H1 TIM and opening the case but it made no difference. I have a photo of the contact pattern here.

When I tested using the OCCT benchmark I was unable to complete a test due to the processor overheating so I underclocked the processor to 3.6 GHz, disabled turbo and manually set vCore to 1.1v.

With an underclocked processor I was able to get a heating and cooling curve using the OCCT auto capture, to enable me to study the problem.

Even when underclocked the processor was reaching high temperatures, rapid fluctuations in temperature with work load suggest a bottleneck in the thermal pathway. When I tested with the intel retail cooler which came with the CPU the cooling was much less effective than the NH-L12 (even when underclocked taking just over a minute of OCCT to reach the 85°C cut off point see below) indicating the NH-L12 was doing a good job of removing heat, which meant the processor was making the heat or the source of the bottleneck.

I have discussed it http://forums.hexus.net/cpus/327593-4790k-overheating-nh-l12.html elsewhere. Advice was to contact Intel due to an absence of information relating to my retailer's testing procedures. I have asked about these but am still waiting for a reply.

So my question is how should I proceed from here? Does this qualify for an RMA? If so is it possible to negotiate this with Intel direct or do I have to go through my retailer?

I have done my best to make sure I am not doing anything wrong and I would be grateful for any pointers to any mistakes I may be making.

[NIaco]

Here's another screenshot.

This time i turned turbo off from BIOS, limited the cpu current limit to 256A and added a negative 0.06v offset to cpu voltage and set my fan to work on max all the time.

This time the temp. rises slower and keeps rising. I stopped the test at 95C to prevent damage. Also, why does Processor Frequency show 0.00GHz?

[MM12]

I had the exact same looking time table in XTU.

If you read my post a few pages back, the problem is internal with the CPU.

No amount of tweaking will fix it. As soon as I de-lid the chip and applied a recommended thermal compound between the CPU die and the IHS, my temperatures drastically improved.It's been a few weeks now and the temperatures have lowered a few more C. What used to be 100C in seconds is now no higher than 65C after 5 minutes.

If you ask some Intel tech support, they will continue to blame the motherboard, and state that the CPU is functioning properly. It's a lie.

Everything about the CPU is great except for the crap thermal compound they use.

To be honest, I had fun taking the lid off and getting this chip to work right.Look up some how to's and you will be fine.Look up how to remove the lid with a vice and WITHOUT a hammer.

[dadi]

That cooler master is a decent cooler, does the job ok for me. There are better ones but is the best I could find in my area for that money.

I recommend a stress test without AVX instruction, intel burn test or OCCT.

[NIaco]

Unconnected i have read your post, but i'm not willing to risk breaking my processor . I'll try to re-seat the stock cooler, if the problem persists, i'll return it. But thanks for the reply

[MM12]

I understand. It's not something you should have to do in the first place, but it is what it is. I was apprehensive at first but I figured I had nothing to lose. In my situation I did not have the ability to return it. I also didn't want to settle for lowering its performance.

Once you have it taken out ready to be returned, if you have a vice lying around, you could even just attempt the process and stop if you felt there was too much pressure being applied. There wont be though, the IHS is held on with what very well could be bubblegum. It's a piece of cake.

But then again if you don't need the i7's performance, get your money back and pick up an i5.

A solution intel could do for this processor in the future would be to simply sell them as a kit, without the IHS attached.

Especially since they're not soldering the die to the IHS, it just seems like the thermal compound is a weak point that will fail sooner or later. Often very soon as we're seeing.Why not leave it to the end user to apply and maintain this layer? It's no different than putting thermal paste on top of it for the cooler.Give us a kit with CPU, IHS and some supplied compound we could use or toss for something better..For the enthusiast line of CPU's, I think it would be appreciated.

[LCCF]

back before i corrected my overheating issues, i looked at the vids on youtube and was tempted to do try de-lidding, but for another reason

I've worked with carbon fiber (CF) and it has the highest heat transfer rate of any element short of diamonds, which is also carbon based. CF transfer heat almost as fast as copper carries electricity. That IHF case that intel uses is made from stainless steel (SS), and Stainless steel has a horrible heat transfer rate - try welding SS and you do not touch it for 15 or so minutes as it's too hot to touch. Aluminum and copper (especially copper) would have been a better choice for heat transfer than SS, but both would have been too soft to support the aftermarket coolers.

That IHS is so simple in shape, it'd been a snap to copy - i'm still curious to try it but my temps are down where i'm happy

anyone out there that knows how to work fiberglass knows how to work CF - i believe there'd be a helluva market for a CF IHS case

[LCCF]

i forget who, but back awhile i recommendedd the noctua nh-l9i cooler for SFF cases with restricted cooler height limitations - just came across one i wish i'd known about

- the CRYORIG C1 - 74mm height while having a 140mm size exchanger with 140mm fan, and spec'd to mount on mini-itx boards. only downside, you'd have to remove the cooler to change or remove the memory boards, not a bigee if you've got a removeable motherboard tray, but like in my case, would have meant 1+ hour to dis-mount everything to get the motherboard out.

Researching cooler comparisons, it would have twice as efficient as my NH-L9I but it does have considerably more heat exchanger air, more cfm, and more heatpipes (the noctua only has two)

there's a neat vid at their web showing it being mounted on an asus h97 mini-itx board

http://www.cryorig.com/c1.php CRYORIG C1 | Research Idea Gear

[NIaco]

I'm back with an update. I removed the stock cooler, cleaned it up, applied some Arctic MX-2 paste (line method) and re-installed the stock cooler. Things have improved quite a bit but still over Intel recommended levels. Before, i had to stop XTU stress test in 5 seconds because it reached 100C, now i can complete a 5 minute stress test with 80C max temp (default BIOS settings, no XTU tweaks)

Here is a screenie:

Then, i set my cooler fan to run at max rpm allways, and re-did the test: temps dropped another 5C

Then, with the fan at max, i tried a slight OC to 4.4Ghz TB on all 4 cores:things got toasty at 93-95C

I'm getting my Hyper 212 EVO on friday, i'll get back with an update then.

[LCCF]

here's a link to a pretty lengthy comparison of most of the major coolers - comparisons on temps at idle, under load, noise etc

it might serve you http://www.relaxedtech.com/reviews/noctua/nh-d15/8 Noctua NH-D15 Benchmark Results | RelaxedTech

[SPark48]

I think you have to take any of those comparison charts with a grain of salt. You can find the same comparison test done by someone else that shows completely different results. Just looking at the results, it's hard to imagine an unbiased 3rd party wrote that up without any incentive from Noctua. Don't get me wrong, the NH-D15 is a fantastic cooler (if they would change that ugly beige/brown motif), I just don't believe it to be the best, or better than many of the other coolers in that list. At least that's how my experience has been.

On another note, Intel forums on a mobile device SUCKS! Especially when the thread is really long, you have to scroll all the way down to get to the end! They need to fix this... Or am I just stupid and doing something wrong?

[LCCF]

ordinarily i'd agree with you on the lack of dis-interested 3rd party evaluations, and especially in this case as Tom's Hardware is a somewhat suspect forum in terms of test comparisons, but in the case of coolers, no matter who has done the comparison, noctua is usually at the top or next to the top. But i didn't post that comparison to push noctua but simply to give Vattulus something to consider

as far as their signature colors, it doesn't matter to me what they look like, pretty or not, as long as they work, but in this case noctua has come out with two new fans with different colors (grey/blue and black/brown iirc) - suspect it won't be long before they offer those fans on their coolers. Kinda neat, their an "industrial grade" fan, with two rpm choices, 2000 & 3000 rpm

- here's a link to the 140mm 3000 rpm fan

http://noctua.at/main.php?show=industrialppc Noctua.at - sound-optimised premium components "Designed in Austria"!

[dadi]

On mobile device this forum is a nightmare, but you can check "request desktop version" in your browser

[NIaco]

Hey guys! i'm happy to report a great success!

A few days ago, my 4790k would get to 100C in 5 seconds with stock cooler, not matter what i tweaked.

Then i changed the paste and re-installed the stock cooler, and my peak temp was 80C, not perfect but better.

Now, i got my Cooler Master Hyper 212 EVO + Arctic MX-2 paste and the temperatures never exceed 62C with no OC, and 71C with 4.4Ghz turbo boost on all 4 cored.

(fist part of the graph is no OC stress test, second part is with 44x turbo boost)

[cmero]

WOW! I got a Gigabyte Z97X-UD5H with a i7-4790K earlier this month for my birthday. I replaced my trusty AMD 1075 6 core which ran circles around this pile of crap. To bad I already passed it on! Anyway, I am running a water cooling system and getting 90 degrees celcius on boot-up right into the bios. THE CPU is seated correctly, I used the best thermal compound available at Micro Center, They even replaced the motherboard once for me in fear it was a temperature sensor. I tried a Stock Intel fan cooler, a upper class fan cooler, then opted for the water cooled system. It was tragic I did not stumble on this 40 page Intel blog until my purchase was complete. I cannot return the CPU. It appears after reading quite a lot of these pages and scanning a few of them Intel has a lot of the I7-4790K chips that are faulty. Now the question is are they going to do something about it? I need not copy and paste benchmark temperature marks of my CPU both at idle and under stress. This wastes time, I need a solution and NOW! So come on Intel ANNIE UP!

[SPark48]

Welcome to the growing club of overheating 4790Ks. Your post begs one to ask the question, what kind of solution are we all looking for from Intel?

Do we want:

1) Solution on how to get 4790K to run in normal range?

2) Replacements for all our our overheating CPUs?

3) Refund or at least partial refund of monies we paid to purchase the 4790K?

I've been sitting here trying to think of how I would respond if I was Intel. For # 1, they have already tried to do that with Ken-Intel's setting change suggestions. This has worked for some, has not worked at all for others, and then there are still others that the setting change will control the overheating, but the CPU still cannot be overclocked or is hit with throttling due to the set settings, rendering the processor essentially "locked" and negating the extra $40-$50 people paid for the 4790K vs the locked 4790.

If # 1 didn't work for you, then you can do # 2 by requesting an RMA. (Zero_Cool, you said you can't return the CPU, but you can still request an RMA, right? It's been less than a year since you have had it?) The problem with this solution is that first you have to PAY the extra $15-$25 to ship the faulty CPU to Intel, which, as I've mentioned multiple times, I just cannot believe. If something Intel made is faulty at no fault to the consumer, why does said consumer have to PAY to get it replaced? Then, even if you get a replacement, there is no guarantee that the replacement won't have the same problems.

That brings us to # 3. This then gets tricky because you have to figure out how much we would be happy with settling on? If the CPU is completely unusable, should you get a full refund? Do they blanket payment to everyone that owns the 4790K, or do you have to file some kind of claim and show proof of your issues? As I have also mentioned before, there is the class action case from October 2014 of Intel paying $15 for manipulating benchmark scores on some Pentium 4 processors from 2000-2001 (nearly FIFTEEN YEARS LATER). Are we going to be happy with some situation like this?

So what do you guys think? Which of the 3 above solutions do we want? Or some other solution that you would prefer?

To answer my own question here, there really is no good answer to the "I need a solution" thing. I'm confident that anyone not finding solace in Ken-Intel's suggested settings will be granted an RMA from Intel if you give them enough grief about it. So for most of us, that probably is the best option. What I still don't like about it is having to pay even more money (shipping costs) to get replacement. I also think # 3 is out of the question as I highly doubt Intel would authorize any refund that is significant enough to make us consumers happy about the situation. If I paid $300 for the processor; am I going to be happy with $20? $40? What if that refund comes 20 years later?

Rock - Intel - Hard place

[Hur_R]

ooooh sampark1980 you are spot on there. I want my money back paying $30 - $40 to overclock when you simply can not. Not to mention the extra money we all have spent getting faster RAMs which again u cant use since using XMP profiles makes the processor go nuts. I mean what's the point then! I want my money back so I could buy something that could keep my system stable or a proper solution cause this clearly isn't working for all. Works for me until I keep the XMP disabled but then again I feel like my system's other hardware is being wasted. These setting doesn't even let processor go over 4.2GHz. I bought it because it was suppose to give me turbo max till 4.4GHz and so i could overclock it too which at this point I can't even think of. Really disappointed in Intel. I never had to review any of their products (particularly processors) online, I wish i would have came across this article before, I never would have bought that thing

[LCCF]

just thought i'd come back and throw something out that i just noticed and supports what ken=intel said just a few posts back.

again, for the record i'm running an ASUS Z97M-PLUS mobo (micro-atx) - Asus just in the past 2 weeks released two new BIOS updates for my mobo. Oddly both show the same release date, with one, # 2501, indicating it "improves stability" and the 2nd one, # 2604 indicating it offers support for the newer Broadwell CPUs. i chose not to install the 2nd one as i assumed it would be only for Broadwell CPUs

anyway, while rendering videos i've been consistently running RealTempGT in the background and notice that my CPU hits a max temp of high 70s and sometimes 80-81C. Usually just once in a 1-2 hour video rendering task.

after 5 days of rendering video files with the 2501 BIOS update, my max temp has never gone over 61C, nearly a 20C drop in temps. Also, when running XTU Benchmark or stress test under the previous BIOS, temps would hit 78-81C. Now, using the same XTU profile as earlier, max temp is 71-73C. That's an approx 9-10C drop in temps just from the BIOS update - add to that the 5-6C drop from un-installing the AI Suite III utilities.

Point is, i've got to wonder how much of this headache i owe to ASUS - ie how sloppy is their evaluation of a BIOS before release. Even now, releasing two bios updates simultaneously, again with one indicated supporting Broadwell CPUs, before installing it I emailed them asking was that BIOS for non- Broadwell CPU use as well, and received an affirmative response. That led me to question, if that's the case, why would they even bother releasing the lower numbered one, 2501, on the same date? That confusing detail was what kept me from installing the Broadwell BIOS - i suspect there's far more confusion within their system than i care to know about

FWIW

[SPark48]

Without a doubt, ASUS shoulders some of the blame on the overheating 4790K's. If nothing else, at least they deserve some finger-pointing for the outrageously high "out-of-the-box" settings that come stock on their motherboards. So then shame on ASUS. But then, oh wait, ASRock motherboards have the same problems with same outrageously high stock settings??? MSI too??? Gigabyte is even worse??? Even some of the less common motherboard brands like EVGA and Supermicro have also reported these same high temperatures with extreme stock settings. This begs one to ask how EVERY single one of the manufacturers did the same thing - they all somehow put outrageously high stock settings onto their Z87 and Z97 boards. Why would they ALL do that? If it was one company that had this problem, you COULD say it was the motherboard manufacturer's problem, but when ALL of them are reporting the same issue and they ALL have one thing in common (they all used processors given to them by Intel to design/create/test their motherboards), then... like the saying goes: the easiest answer is usually the correct answer.

Again, I'm not saying that the motherboard manufacturers have NO blame and that it's all Intel's problem. However, my hypothesis is that the chips that Intel gave out to manufacturers to test on their motherboards were engineered differently than the 4790K's that actually made it to mass production. Since people are having luck with de-lidding, it would make sense to say that the pre-production 4790Ks had the proper heat distribution, thermal compound, and space under the lid to handle the amount of heat the 4790K's produce. Unfortunately, during the mass production phase, Intel made some kind of change that resulted in many of the 4790K's no longer being able to handle the heat.

But then again, Ken said: "Experience has shown that many Z97 boards come with very aggressive default BIOS settings for voltages, voltage offsets and current limits. These aggressive default settings result in a highly overclocked CPU, even though you didn't change anything to initiate the overclock. These settings are not in alignment with the specifications given to the board manufacturers by Intel." If this is so, then Ken, can you explain how every board manufacturer managed to get these settings wrong? I have built numerous Z87, H97, and Z97 computers in the last year from all different manufacturers at all different price points and features. Every single one of them had these same "aggressive default settings". Also, your setting change suggestions have been a huge blessing and it has helped a lot of people. However, there are still many that don't have any reprieve even with your suggested setting changes. Is there a solution/restitution of any kind coming soon from Intel? Or is the official stance "blame the motherboard manufacturers"? Is there anything you can tell us about what is being said internally about the 4790K's heat production? I also noticed that the X99 chipset unlocked i7 processors (5820K, 5930K) are having similar overheating issues. Is this a trend that is going to continue with all unlocked i7 processors that you guys release? Are there any reassurances your company can give to the consumers who have suffered through these 4790K's that would entice us to throw more of our hard-earned money at your company for future releases?

Thank you, Ken, for being the only one from your company brave enough to come on here and be the scapegoat for many angry customers.

[GSing15]

Well from what experience says, my chip has been hitting 95 C, Intel mailed me a used chip that he deemed giving low temperatures for him. When I got it, I installed it, and I STILL hits 95 C. What puzzles me is that Intel told me they got fine temps (assuming ~80C maybe) using a stock cooler using the chip they gave me. I tried it, it hits 95C. I contacted ASRock, my manufacturer, whether it was an issue with Z97Ms by them... They sent me a pic of it running 77C on a water cooler at 4.5gHz with 1.3V on Prime95 on my same motherboard. They told me to use a better cooler and that the motherboards were not a problem.( I also contacted ASRock to ask if they could give me their resulting temps from a stock cooler, which they forwarded to HQ. I still await a response for that) I don't have a better cooler, I'm not overclocking, I just want to run default clocks with a stock cooler, WHICH DOES NOT WORK!!!!

sampark, i used to own an i7 2600 which would have max temps ~80 C. I expected no less when i switched to this chip. I don't care about the excuse that it is clocked higher. (That's intel's' job, not mine.) Either make better coolers, or get manageable temps with your stock cooler. Seriously, I shouldn't have to buy a $60 extra cooler just to get regular temperatures. What was the point of getting a "k" chip if it requires a $60 just to get "in range" temps let alone "low" temps for overclocking. ( I bought the k over the non k version because it had a higher stock clock, i'm not an overclocking enthusiast)

There are a lot of things I'd like to spend $60 on and I would have NEVER bought this chip in the first place if I would have known that I would have to buy an aftermarket cooler.

[KenF_Intel]

Hi Zero_Cool.

I understand your position. You're report of 95C in BIOS is very much out of the norm.

Experience has shown that many Z97 boards come with very aggressive default BIOS settings for voltages, voltage offsets and current limits. These aggressive default settings result in a highly overclocked CPU, even though you didn't change anything to initiate the overclock. These settings are not in alignment with the specifications given to the board manufacturers by Intel.

Would you mind posting a picture of XTU that shows us what your current settings are so we can evaluate if there are changes that might help your system.

It would also help if we could get some additional specs for your system. Especially what memory you are using and whether or not an XMP profile is being used.

Thanks.

[GSing15]

Has anybody resolved their issues with an ASRock Motherboard maybe?

(btw, using another cooler isn't a solution for me)

I hit 95C on a 5 minute stress test with XTU... Anybody with a similar graph?