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.

[LCCF]

on your question of how a locked 4790 running 1600 ram freq could hit the scores i have, when i posted that 976 or 986 benchmark, there were two above me, one at 1011 and the other at 1038 - neither were running asus mb, and when i tried their settings, both gave me considerably higher temps than i'm seeing now.

before selecting the G.Skill ripjaws ram i did, when i researched the clock speed & benefits of higher freq ram, one analysis (that was a little above my pay grade, comprehension wise), the results showed a 3-19% range of actual realized speed benefit. On that basis, because of the cost of the higher clock speed ram, i didn't feel it was worth it and went with 1600 GHz, making sure it was listed on asus's qualified vendor's list, in terms of compatibility

[jhill15]

I would like to describe my situation involving the 4790k and the problem outlined very clearly in this thread, and how I went about to resolve or at least minimize it.

My average computing usage:

Gaming and Statistics software

My setup:

ASUS Z97M-PLUS motherboard with latest BIOS revision

Intel 4790k CPU: Batch# L420B887

G Skill F3-2400C10-4GTX 8 GB Dual Channel Memory Kit

CPU Fan: Zalman CNPS8900 Quiet

Videocard: NVIDIA GTX 960 (purchased less than one week ago)

Windows 8 64 Basic.

Originally, I had the stock cooler and ran with an updated BIOS version at factory defaults. I did not enable XMP profile yet as I was waiting to upgrade to an aftermarket CPU cooler and did not want the memory controller to produce more heat than the stock fan was capable of dissipating.

This is the results of stock operation, everything factory default with latest BIOS revision from ASUS:

It's important to note here that the core is operating idle at 4.4ghz in windows 8, with power settings set to high performance.

As you can see, the default ASUS BIOS settings have been changed by ASUS to those outlined by ken_intel on his post:

Running the XTU Benchmark:

Upon inspection you can see that the processor does not exceed about 69°C, the CPU Core Voltage is static at 1.21V, and the clock speed is controlled by Current Limit Throttling. The problem here however is that the processor never actually operates at 4.4ghz clock frequency. It operated between 4.0ghz and 4.15ghz throughout the test. It is constantly throttled by the 105A Processor Current Limit. I tested increasing the Processor Current Limit in increments, and found that at 256A, the processor starts Power Limit Throttling instead, but yielding the same thermal profile throughout the XTU benchmark and the same clock speed limitation between 4.0ghz and 4.15ghz.

So, if my understanding of the processor is correct, if you put it at load it will never actually see that max turbo value, but rather throttle away between 4.0ghz and 4.15ghz to keep thermals in check. Is this as advertised?

I am using aftermarket cooling with a Zalman CNPS8900 Quiet cooler, seated correctly with Arctic Silver 5 thermal paste. I had some untapped memory speeds that I could likely adjust to improve the performance and could dissipate any additional heat generated by the increased activity of the integrated memory controller, so I enabled the XMP profile and took another look.

.

.

.

Here are my result after enabling the XMP profile:

As you can see, the processor is locked at 44x multiplier across all 4 cores at 1.210V, and the memory is operating as designed at 1199.8 MHz DRAM Frequency, so effectively 2400mhz 10-12-12-31 2T.

After enabling XMP, the cores are locked to 44x multiplier, the Processor Current Limit is, Turbo Boost Power Max/Short Power Max are all set to maximum values. Knowing this, I understood that current limit throttling and power limit throttling would no longer be occurring.

Running the XTU Benchmark:

Upon inspection you can see that indeed there was no current limit throttling, but the processor reached an abhorrent temperature of 97°C at which point I stopped the benchmark. The core operated between 1.21V and 1.23V throughout the duration of the test. At this point I became a bit disappointed because I realized I would not be able to enable the XMP profile and stress the processor lest I want to send it to an early grave. I figured considering the marketing around this processor and how it has been dubbed an overclockers dream, that operating at the specified factory turbo with a reasonable aftermarket cooler would not send the computer instantaneously into dangerous thermal territory, but rather operate somewhere underneath the thermal limitations of the processor. It would appear that I was wrong.

My work-around for the time being:

By enabling the XMP profile to take advantage of my increased memory speed, I realized I could probably adjust the voltage to bring the core temperatures down a bit.

I manipulated the Dynamic CPU Voltage Offset setting in the XTU profile incrementally until it didn't result in a blue screen. I found that decreasing the voltage such that 4.4ghz is running with 1.13V was optimal. Anything less and it would cause blue screen, anything more and it would produce thermal graphs that were very Xtreme indeed.

Here is the setting change:

Running the XTU Benchmark:

</p...

[IMorg]

Hi everyone, new to the board and a little concerned.

I bought a i7 4790k in late september and have only recently got round to actually doing the build, for various business and personal reasons. I noticed in the bios my cpu was running hot considering it is winter and my room is probably around 10oC, the processor was running at 32oC in the BIOS. I've tried to download the XTU software for windows 7 64 bit to run the diognasitics to check my chip isn't a dud, but the software fails to install because it tells me that it doesn't match my platform, even though it does match my OS. How do I check my chip is okay? I'm only really slightly above the average user knowledge on these things and I was hoping this would be a good "plug and play" chip I could stay with for a few years and not worry about upgrading. Really disapointed I have to worry about this crap considering the cost of the chip and intel's reputation.

My batch number is L421B956.

[SPark48]

32C is a wonderful temperature for your machine. Why would you think that's hot? On top of that, that's your idle temperature. What's important is your load temperatures. You need to get software to stress your CPU to 100% and see what your temperatures are. As far as your XTU not installing, you have a bad link. It's crazy that Intel has not fixed that issue yet. My guess is they don't want you to install XTU unless you have some basic knowledge so they purposely made it difficult to do it...

Here is the proper link: https://downloadcenter.intel.com/Detail_Desc.aspx?DwnldID=24075 Intel® Download Center

[IMorg]

I thought that 32oC was hot considering how cold the ambient temperature is in the room at the moment and it was pretty much idling in bios for a few minutes, but I'm no expert so thanks for the reasurance that the chip is okay. I'll download and install XTU tonight and check to see if everything is okay.

[SPark48]

Maybe with the computer completely turned off it may match ambient temperatures, but it is a machine with electrically currents constantly running through it. Wouldn't you expect it to be hotter than the air in your room?

Idle temperatures are not that important, but 32C is a great idle temperature. Check your load temperatures and get back to us if you have over 80C.

[dadi]

Get real temp or hw monitor (this one monitors a lot more things) to watch the temps and intel burn test to stress your cpu; if you get temps above 75 celsius then use xtu to read and tweak your cpu settings, when satisfied with the changes implement them in bios.

[СЧирв]

My processor has overheated to 100 degrees and burned (after 3 minutes of stress test !!!!) (i7 4790k) on cooler Cooler Master 212 evo (on default settings in bios)

[dadi]

Did you clear the CMOS ?

Should be a button on the main board.

But I read that you tried another CPU on that board and it worked, so I think that your 4790k is dead

Talk to the seller, the store that sold you the CPU. It should be covered by warranty.

[СЧирв]

I try everything, i bought new motherboard, clean CMOS, replaced PSU and more-more.

I thinkб my next processor will be AMD, but i do not like AMD (

About Intel support - terrible and slowly, my processor costs $400, but users support worse than china phone for $5

[dadi]

Use the warranty at the place you bought the CPU. The product has at least

1 year warranty. The store should replace it and send it to Intel for

investigation. If 30 days days have passed from purchase date they are not

under obligation to replace it but they still manage the product warranty.

The 30 days term might be different from region to region, or depending on

store policy.

Where did you bought the CPU from ?

[СЧирв]

I bought CPU in Kiev, Ukraine, but i bought from my friend and he cannot have any documents (CPU- BOX version).

But, I have 3 year warranty (i check serial number on this site).

Problem - Intel support cannot contact with me (I mean Ukrainian and Russian support first) - and i do not know, what i need to do

[dadi]

I'm sorry to hear that, without proof of purchase the seller can't help you. If your friend doesn't have the receipt you'll have to talk to Intel service people yourself.

Unfortunately I can't be of any help, I'm just a user here and nothing to do with Intel, just using their product.

[LCCF]

just to wrap up before i leave this thread. In an earlier post i had indicated un-installing the asus performance utility had dropped my temps some, 6-8C at idle and 4-5C at high end under 100% load, dropping temps down to 94-95C. Clearing my CMOS and my RTC dropped them some more, approx 7-9C at top end but still leaving them high, per intel's specs.

I ordered a new Noctua NH-L9i compact low profile cooler to replace the stock intel box cooler. It made a big difference, i think in my case for two reasons: 1) the noctua is 37mm high vs the intel's 57mm, so it gave me .75" more distance between the bottom of the PSU's fan and the cooler's fan (PSU sits directly above the CPU and that large 140mm fan has to be robbing air from the cooler's fan.

I also added in a 4th fan in the case side (1 80mm, 3 60mm) and there is now a steady cool stream of fresh air blowing directly across the entire MB as well as at the cooler. (my cpu resides in a SFF case that is limited in interior space, the silverstone sugo sbo2. But with that 4th fan, air coming out the one exhaust fan opening at the top is actually cool at idle, and nowhere near as warm as it used to be after running at full load for awhile.

With an ambient temp of 21C, my temps are now 39-42C at idle, and running the stress test in XTU, with default settings they peak at 74-78C - if i change core voltage to 1.016, which was the orig default voltage, they peak at 76-78. In the XTU stress test they never went over 78 in three consecutive tests.

Just ran a video file for 1.5 hours with RealTemp 3.70 running in the background, and RealTemp showed it hit 68C max, so i'm a happy camper.

I noticed something about the noctua cooler that ties in with the early complaints about excessive glue under the cpu case cover causing an excessive gap between the cores and the cpu cover. One reviewer of the noctua NH-L9i cooler, had commented that the heat sink was not perfectly flat, and actually had a crown in the center, and in measuring it with a straight edge across the heat sink, iirc he found something to the order of .032 thousandths gap under the straight edge on one side when the edge was held down on the opposite side of the heat sink (think in terms of a "see-saw" - that would make for a peak in the center of approx .016 higher that the edges. You can see evidence of the shallow cone in the image below from the reflections catty-corner to each other - i say "shallow cone" as that's exactly what it is, there's a barely perceptable high point dead center of the heat sink contact area, that's about .016" higher than all four edges (it doesn't matter which direction you place the straight edge on the heat sink, you get the same gap whether the edge is oriented 3:00 to 9;00 or 6:00 to 12:00. I can't capture it with my camera, but as you move the heatsink under a light, you can see "rings" growing in size, out from the center point, but they're perfectly formed circular rings (think like a phonegraph record - probably from the machining in the die that sink was struck in (struck as in "stamped" under high pressure.

i work with metal - that is not an accidental shape, ie not the result of poor quality. That was designed in the tooling they struck that heat sink plate - i suspect they were allowing for the CPU covers that are sitting too high with too great a gap between the cover and the cores, by designing that crown to press the center of the CPU cores down in the center.

That's only my suspicion but it's the only logical explanation i can come up with.

FWiW

Thanks to everyone for their assistance and especially their patience

almost forgot - for those with asus boards that go the route of un-installing their asus AI Suite III, over in the Motherboard chat forum on Tomshardware there's a thread (a sticky) dedicated to ASUS motherboards, with some ASUS reps chiming in to posters' questions. On the subject of un-installing AI Suite III, one of them commented that if you do un-install it, there's a AI Suite III "cleaner" to complete the un-installation, and provides a link to it. I don't think this forum is going to allow the thread link to be posted, but the thread is titled "ASUS Z97 Motherboards - Official Support Thread"

and the link http://www.tomshardware.co.uk/answers/id-2139079/asus-z97-motherboards-official-support-thread.html http://www.tomshardware.co.uk/answers/id-2139079/asus-z97-motherboards-official-support-thread.html

[Hur_R]

have used the settings all the mentioned settings on this discussion on my machine:

core i7 4790K

asus Z-97K

32gb ram corsair vengeance 1866

aslo using Cooler Master Seidon 120V for Cooler.

on idle getting, 40 degree Celsius and stressing getting 90 Celcius plus. Does anybody have any suggestions?

[LCCF]

have you tried un-installing Asus's AI Suite 3 and resetting the CMOS?

it's worked for some and not showed any improvement for others - when i un-installed the AI performance utility, my idle temps dropped from 51-52C down to 44-45C and load temps fell approx 5C (from 99-100C down to 94-95C), so i assume there was a conflict between the asus performance utility and the bios

when i cleared CMOs, temps fell considerably more, bring my load temps down to low 80s -

fwiw, i procrastinated un-installing the Asus performance utility and then clearing CMOS, thinking that it wouldn't or couldn't account for that much of the excessive temps - i felt kind of foolish when it gave me that kick start on the road to decent temps. I've since upgraded my cooler and installed some extra case fans and am now seeing temps i'm happy with

[Hur_R]

I never installed it, even without it, its outrageous I don't understand why these processors won't turn off when we all are apparently exceeding the Tcase way over I mean 74 is what is they have stated on the Intel's website and we all are at 100 C and this computer keeps getting restarted every now and then.

[LCCF]

I agree it's outrageous but it is what it is, and the best right now is to stay focused on correcting

did you clear the CMOS?

[Hur_R]

Yes cleared it, updated it then cleared it you name it . Under-clocking the processor for now till the real solution arrives? to protect from damage?

[MM12]

Without a doubt, I believe I have found the problem.

I just successfully completed delidding my i7-4790K with at first NO CHANGE in results or temperatures, potentially worse.

Of Note: After lid removal, I noticed that it appeared all of the thermal paste from the factory had been squeezed out beyond the CPU die. I have photos I could provide.

I applied Coollabs liquid pro (CLP) to the die, reassembled the system and began intels stress testing where I saw no, or possibly worse temperatures.

Suspiciously, and consistent with prior to delidding, my water cooler and radiator was not warm after this few minutes of testing and half an hour or so of idling.

With the XTU open so I could monitor the temperatures, I went to remove my water cooler from the CPU and to my surprise while loosening the screws, the idle temperatures plummeted from low 40s to high 20/low 30s.

I ran the XTU stress test again and CPU temperature did not exceed 80 deg C, with 0 throttling. Compared to 100 deg c with ~33% throttling prior. See Photos below.

I'm assuming the CLP, still fluid, was pulled back in towards the die with the released pressure, allowing it to perform properly.

This leads me to believe, along with my initial observation of the factory applied thermal paste ( which had dried to a rubbery consistency) that somewhere along the line either us users, or Intel, is applying pressure to the IHS that is forcing the thermal compound out from the die.

Intel's thermal compound does not have any fluidity to it so that once it's been compressed out it will not flow back, and there will be an air gap between IHS and the CPU die.

I'm not aware if Intel's thermal paste hardens after being heat cycled from the consumer, or just over time after leaving the factory. The CPU I was initially using before RMA'ing had the same issue, however the person using it prior reported no temperature issues. He was not necessarily monitoring the temperature though.

This could have been caused by excess pressure when securing the water cooler block to the CPU. The cooler is currently almost loose, just enough pressure to keep it on, but it's barely snug.

I'm certain I could re-do everything and achieve even better results, but at the moment I'm quite happy.

1st test, pre-delid.

http://i.imgur.com/ac37lYp.jpg http://i.imgur.com/ac37lYp.jpg

2nd test, post-delid, no improvement.

http://i.imgur.com/0SryDXS.jpg http://i.imgur.com/0SryDXS.jpg

3rd test, post-delid with released cooler pressure.

http://i.imgur.com/wdaywGm.jpg http://i.imgur.com/wdaywGm.jpghttp://i.imgur.com/wdaywGm.jpg http://i.imgur.com/wdaywGm.jpg

[MM12]

I am not suggesting to perform this. This photo is merely to show the lack of thermal compound on the CPU die. The IHS has residue, but a majority of the compound is surrounding where it makes contact, and not where needed.

If you are installing a NEW CPU, apply your heat sink using very little pressure! This may avoid the overheating problem all together.