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.

[REnso1]

The http://www.anandtech.com/show/8227/devils-canyon-review-intel-core-i7-4790k-and-i5-4690k Anand tech I7 4790K review article and the post it refers to make interesting reading, provides quite a scientific proof that the die-spreader gap is an issue. To paraphrase Dr Strangelove, we must not allow a die-spreader gap...

I find myself deterred from delidding due to the fact it would void the warranty and while cost is a concern it is not the overriding priority which is to have a working computer with a good CPU, since I currently have a replacement hand picked by my retailer. Delidding risks wrecking that chip and its not appropriate on principle for consumers to take that risk where they have no protection against costs, period. In the worst case scenario it could lead to cracking open a string of CPUs like so many fortune cookies hoping for a lucky break! That could get prohibitively expensive, not worth it for 300MHz turbo.

Modifying the heatsink to increase the pressure on the socket and package is also tempting but I am also well aware of how fragile this socket design is so it is probably best left as it is and run a little below spec.

[JVene1]

I get it, and largely we agree.

I wonder, though, if you know that when the world uses Google to find references to the subject of overheating i7 chips, this thread now appears within the first couple of pages of results. There will be a lot of viewers, many of whom are enthusiasts stark raving mad about the problem.

To them, the problem is worse. Many have seen temperatures hit 100C at stock speeds within seconds of virtually any load on the processor. After that the CPU throttles down to cool off, at a performance barely exceeding my AMD 955 X4. Some more common consumers might be experiencing that same problem on non-K processors and have no idea, because they don't even realize they can monitor CPU temperature and don't realize something is wrong unless the thing crashes.

The supply channel is full of CPU's from 3 generations with this same problem in varying degrees. Many have achieved RMA for particularly bad examples, but everyone is stuck with a CPU that is hotter than it should be, degrading faster than it should over a stupidly simple problem (compared to the engineering marvel the i7 represents).

To the enthusiasts, delidding a processor seems like the only possible solution. The thinking, to such owners, is that there is no solution available from Intel. They can trade an extremely bad example for a mediocre one, but it would be a very rare case where the "K" CPU does what they expect of it as it comes out of the box.

Considering the history of this kind of problem, from XBox to the Pentium 3-600, a real solution is likely months away, and quite possibly never for these particular SKU's. Most "K" purchasers can't accept the concept of running at reduced expectations, and really they shouldn't be in that position. They will either have to accept the situation, upgrading to a Broadwell (assuming that doesn't have the same problem), wait for yet another generation in 2015 (or later), switch to AMD, switch back to their previous hardware or take matters into their own hands.

I do agree with your point that delidding is likely to damage the CPU, but these enthusiasts (numbering between the 100's of thousands to maybe a million) are approaching their wits end, rather obviously considering the video evidence, with many reporting fabulous results.

For anyone where the investment is too precious, they should RMA and try something else.

For those finding threads like this while researching their planned purchase of an I7, the cautionary tale is maybe they should wait. Wait, because the only way to ensure satisfaction is to tread into dangerous territory, possibly destroying the chip in the attempt to get it to work as it should have.

To many of us, it's as if we purchased a car that we drive to work and locally, in start-stop traffic, no more than 40 mph. At those speeds the car is fine. Then, upon entering the highway we discover that although the car accelerates to 75 rather fast, within seconds it drops to 45 and goes no faster. After a while it may return to 75, but then back to 45 after a few seconds.

Would anyone in their right mind simply accept that?

Certainly no one in their right mind would take a hammer to the engine in an attempt to fix the problem themselves, but what if the manufacturer had no solution, and, in particular, the only other car you could possibly choose was basically an old Yugo?

There would be class action lawsuits over such a thing.

Yet, in this case, with the only other alternative being AMD, and no other real solution available - I'm simply grateful this isn't an automobile, or even the automotive industry. In that market we have choices.

There is hardly a real alternative to the i7, except, of course, not owning one. The counterparts from AMD are afflicted with a different problem along similar lines, without a solution of any kind. What i7 can do in 90 watts takes 220 watts in AMD, and even then it's not a match. Even if you could arrange sufficient cooling, the motherboards aren't actually able to provide 220 watts reasonably. It's a worse problem because the solution goes all the way down to the transistors. For the i7, it's a painfully simple problem in what is otherwise an exemplary engineering triumph.

It's actually hard to believe.

[COuim]

I just bought the Core i 4790k and I am very disappointed that Intel take this lightly. I have an AMD FX6300 and FX8350 that heats less. I was suppries to see the temp to rise at 100c.

It's not normal that we have to go to the bios to make some ajustment. It's suppose to work at the default config. it's only a bad engineering issues that Intel are not able to fix on the fly.

don't waste your time to RMA the CPU or the Motherboard. I'm working in computer since 20 years and i never RMA a CPU and the problems comes with other motherboard manufacture even if we update the Bios. The problem it's just like i said (Bad engineering). By the way my board is a Asus Sabertooth MRK2 and have the same result.

[WMoor1]

I have an AMD FX6300 and FX8350 that heats less.

TJMax is lower with both chips, but it's kind of funny after reading this thread that 55C alarmed me with the FX6350. With a Hyper 212 EVO it only gets as high as 48C if I play something demanding for hours like Crysis 3. That's only with a very slight (but stable) overclock (4.4GHz).

This time last week I had assembled a list of parts to buy for a new i7 4790K rig. After a few days of alerts from this thread I've decided to shelve the idea until next summer at the earliest.

I'm unsubscribing from the thread now, so best wishes to you all. I hope Intel step up to the plate and you're all able to resolve your heat problems without voiding your warranties or worse still, damaging your new i7 CPUs.

[lpart]

hello

Sorry for the google translation

I just bought a i7 4790k and I have meet the same problem as you.

use the chiller is however a h100i corsair.

At the beginning I have thought a pump malfunction IOA.

J have therefore disassemble the h100i to test with the intel cooler which gave me even higher temperatures.

I have to buy the most efficient thermal compound and reassemble the h100i and I get anyway is 75-80 ° C when the CPU is at 100% (for 4400 Mhz) load with any change in the bios.

Finally I'm falling on this forum by searching the internet.

J INTEL hope will quickly information about the problems faced by many users

[GMatr]

Greetings from Italy. I've had exactly the same issue with my intel i7 4790K cooled with CM seidon 120V. Under the own intel's stress test it reaches temperatures of 100°C per core almos fixed (at some point I thought I could put up and cook some pasta on the Seidon radiator). In Idle it just goes for 30° (+/-4°). The MOBO is a ga z87 hd3. I try to upgrade the bios, but I doubt that it will be effective.

[COuim]

Sorry for my english to. Yesterday night, did some test. With my Noctua NH-D14 i was getting 95c at full load with 140 watt of the CPU Package. The only way ì had to make this CPU more cooler, it by reduce the CPU Voltage. With Asus, i put the voltage on Offset mode with negative position. On this way i drop the temperature at 78c at full load with 105 watt CPU Package. So the point is. Is these problem are the CPU it self in cause or Motherboard Manufacture architecture that make the CPU (4790k) to overheat ?

[SElwa]

ouimetca

Could you please tell me what is your current CPU Frequency: 4.0GHz or 4.4GHz? Also, I'd like to know exactly what you did with the bios settings with exact number values because I will get my new CPU after 2 days and if it overheats again I'd like to do exactly as you did before I consider RMA

[JVene1]

While I'm not ouimetca, I thought I'd point out that this is also a recognized issue. Some CPU's are worthy of RMA (some of us are of the opinion all in this category are), but in some cases there are motherboards which set automatic values such that voltage is too high.

Unless your motherboard is the exact same model / revision / bios / CPU as ouimetca, the information may be less than specifically applicable to your board. It's much better to proceed with understanding than by wrote duplication.

What you need to do is check the voltage reading, either through various monitoring software or through the bios hardware monitor of your board, and compare that reading to your CPU's stock requirement. Some boards default to high voltages for unknown reasons, and manually setting to the correct stock voltage for your CPU has been corrective in many cases. The correct voltages to expect can be found through Google regarding your processor.

[SElwa]

JVene

Thank you for the info, I already know that. And no, my motherboard is different than ouimetca, I am just in a desperate situation that any kind of guidance would help, what caught my eyes is "105 watt" that is why I was asking.

[BBere2]

Someone have tested the CPU setting max wattage at 88W?

[CBuc]

I just purchased a new i7-4790K on amazon along with an Asus z97M motherboard and have the exact same problems. I can't even start a web browser with the stock CPU cooler. It immediately spikes the temp to over 100°C. Everything is at stock default settings and the idle temp is 40°C but quickly rises the second any application loads and the system fans increase to max speed. I've tried lowering Vcore to 1.190 and that helped a little. I installed a Noctua NH D14 with no improvement so there was obviously something more wrong than just the type of CPU cooler. Even the stock BIOS optimization would cause a reboot. I've researched this problem and have found various threads relating to it and it seems to happen with any motherboard so I've ruled out motherboard and BIOS setting as the source of the problem. I tried reseating the cooler 3 times with AS5 with no improvement. Also the Noctua heatsink is not getting very warm for the amount of heat the BIOS is reporting so the cooling failure has got to be chip related. I went ahead and voided the warranty and de-lided the chip and sure enough, I could see what the problem was. The glue holding the lid on was very thick and there was almost no white TIM compound on the chip. I cleaned it up, reapplied compound to the chip , set the lid back on and reinstalled the CPU cooler. Fired it up and problem solved. Idles at 26°C . I set the BIOS to optimize and it did an OC to 4.6Ghz and runs intel burn test on Max settings between 58°C and 90°C.

[RKimb]

Interesting. But presumably Intel don't expect all its users to do that. Do they?

[MDitz]

I for one will not. Its unacceptable to put a product on the market that doesnt even work. I so wanted to buy this processor and I assume that when I open the box and build it in it works. I always use pretty good cooling solutions, I never use stock. Although it still has to work with the stockcooler, I mean why add the cooler right? Looking forward to a reply from Intel! Searching for another processor now:(

[MCron]

Since this discussion has been marked as answered, and there has been silence from any Intel representative, I feel this thread has extinguished its uses and in order to get a response those that still request a response will need to open a new thread or contact Intel directly by phone or e-mail.

[JVene1]

Richard_Kimber wrote:

Interesting. But presumably Intel don't expect all its users to do that. Do they?

They don't. They expect users to apply for a warranty replacement or RMA, and many times that works to reasonable satisfaction, though occasionally on the 3rd or subsequent attempt.

Delidding voids the warranty (which, presumably, overclocking does too), and depending on skill presents a high chance for permanent damage to the CPU.

Unfortunately the RMA doesn't always go smoothly because the problem isn't always as obvious as a unit hitting 100 C in 30 seconds at stock settings. For enthusiasts, now recognizing the problem affects many of the processors in varying degrees, even an adequate performance is disappointing, hence the discussion of delidding the processor.

For users just wanting to correct a particularly bad example, RMA is most certainly the way to go.

For enthusiasts unwilling to accept mediocre results in temperature control it is fast becoming the norm to assume delidding the processor, but I hasten to note that's a select minority already accustomed to assuming such risks. In my own case history I've been known to take a razor to an AMD XP, before the 64 bit x86 era, in order to enable multi-processing, or wrap a couple of thin wires around CPU pins with tweezers to set the clock rate for an AMD Sempron (to make it work on a board not designed for it) in the epoch when that was done more primitively than it is today.

All previous use of delidding was limited to devoted enthusiasts, but at this point even new ones are motivated to join in because the pace of CPU development has slowed of late, and there really are few alternatives in some viewpoints.

[RKimb]

Earlier in the thread an Intel guy said the issue was being looked at by the engineers and that he would report back when he knew something. These things do not happen quickly in very large corporations.

[YYuri2]

Things may not happen quickly in big corporates, but it never should take them 2-3 months to look at an issues which is affecting many many customers and keep quiet as they do now..

[JBras1]

This is the thing that annoys me the most about it, Intel should not take this that lightly. You can't just keep silent about a problem like this for months in a row without giving people some decent heads up.

I'm a patient guy but I'm getting really pissed with the way things are going now. I have spend a lot of hard earned money on this machine and the least we could expect is some decent response and a solution to this problem.

Them being so silent would suggest a hardwareflaw that can't be fixed with some BIOS-update or anything like that. cb609 his post confirms that. All the people having these problems sending their hardware back as RMA would cost them a lot of money. It would also hurt the image of the company of having so many broken CPU's out there.

[SElwa]

I have a serious problem now guys, please advise asap, the problem is after I returned the CPU for replacement, guess what? The maintenance guy at the retail store is not sure if these temperature readings are normal or not and he is doing further testing, do you know what that mean? It means if he thinks that 100c peak is a normal temperature reading for this CPU at the stock cooling then he won't replace my CPU. I contacted Intel so that if it tells me that that temperature reading is not normal, then the retail store will replace my chip because they need evidence that a 100c peak in temp is not normal under stress test at stock cooling for that CPU, but Intel hasn't responded yet what I need is an evidence for that guy because every review about that CPU he reads on the web the reviewer uses an aftermarket cooler so he is not sure about the stock temperature!!

[lpart]

ARK intel ----> tcase

http://ark.intel.com/fr/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz ARK | Intel® Core™ i7-4790K Processor (8M Cache, up to 4.40 GHz)