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.

[cpalm3]

What program are you using to verify your VCORE voltage? I think I'm going to mess around with the voltage settings on mine this weekend.

[SBuck1]

I use CoreTemp for both temperature and voltage monitoring. It has logging capabilities as well which can be useful if you're getting crashes.

[FFran20]

After reading the whole thread, and also having the same issue, I tried to test the cpu in a different way and also using a different software to monitor what is happening. A few thing were actually surprising. Also please note I am using the stock cooler. I will have my water cooling running once I get my video card back from RMA.

Here is some info on my build :

OS: Windows 7 64-bit

MB: Asus Maximus VII Gene (latest bios from asus)

CPU: Core i7-4790K stock cooler

RAM: 8GB (2 x 4GB) G-Skill Trident X Series DDR3 2400 mhz (running at 2133 and 1.65v)

GPU: Asus GTX Titan

Now, same as everyone else, cpu was running very hot on idle when I first booted up. Avg of 50c on idle in windows and in the bios. All settings in bios was on auto. Tried a few games, OCCT and Intel burn test. Games were running at around 75c to 85c depending on load. OCCT and intel burn test would just shoot the temp at 100c in seconds. I was using Core temp and CPU-Z to monitor the vcore. Core temps was showing between 1.25v to 1.30v depending on load. CPU-Z was showing 1.8v all the time, wich scared me the first time I saw it, so I stopped using it since it was (hopefully) wrong.

After some tweaking in the bios, I set the vcore to 1.142v, set the boost to 4.2ghz for all cores and left everything on auto(Hyperthreading on). Now on idle, temps are around 40c. In games, it ranges from 60c to 75c. But still unable to runs OCCT or intel burn test as the temp will shoot to 100c in seconds.

I then decided to try the Asus AI suite to monitor the cpu. It gives the power draw of the cpu wich other programs I was using could not provide and it also suppose to give more accurate info since it's made to work with the specific board.

The first thing I noticed, were the temps. On idle, the AI suite, gives a temp of 33c compare to 40c on core temp. The Vcore are pretty much the same. I then loaded some games and also some Folding, since it uses 100% and can push the cpu pretty hard. Here are some results. All numbers are avg. and all test were running for about 1 hour.

World of warcraft - cpu load 20% - power draw : 26w - temp on asus ai : 55c - Core temp : 63c

Battlefield 4 - cpu load 50% - power draw : 45w - temp on asus ai : 65c - core temp : 75c

Folding@home - cpu load 100% - power draw 89w - temp on asus ai : 87c - core temp : 100c

OCCT - cpu load 100% - power draw : 135w - temp on asus ai : 100c - core temp : 117c (all this happens in a couples of sec, I then shut it off right away to prevent damage)

intel burn test - cpu load 100% - power draw : 135w - temp on asus ai : 100c - core temp : 117c (all this happens in a couples of sec, I then shut it off right away to prevent damage)

The temps for games, seems alright for using the stock cooler. Same with the Folding. My old 2600k under water when Folding had an avg temp of 70c and it was using a much higher power draw and vcore(was running at 4.8 ghz, 1.4vcore, avg powerdraw of 150w at 100% load). For the benchmark software, well I have to idea what is happening. And all of this was achieved if some settings were manualy set.

Two things seems wierd here. The difference in temps between the 2 software. The higher the temps, the higher the difference. Now the Asus AI suite is "suppose" to be more accurate. How much to rely on it more then the others is hard to say at this point since the difference went as high as 17c. Vcore was still about the same. 1.153v on Asus AI and 1.141 on core temp.

Then, the power draw. Those cpus are made with a TDP of 88w, so how come it draws so much power "135w"????? The setting in my bios are set on auto, wich means the cpu should run at stock specs, but still draw all this power.

After all this, I am at lost. Do we blame Intel for bad cpus? The benchmark software? The boards? The bios? Plus, from what I can see, aint much left for overclocking wich I was hoping for.

Hopefully this will help some folks at bringing there temps lower and give more info to Intel to find a solution.

[AStar11]

AFAIK, CoreTemp does not read the Vcore, it reads the VID. Which can be considered the voltage the chip "wants" by default. Or similar. Asus AI suite reads the voltage correctly from my experience (Maximus VII Formula). (Coretemp isn't too far off the correct voltage).

AI suite, as far as I know, uses an average of all four core temperatures, or reads the CPU temperature from the socket, which has a sensor behind the CPU package itself. CoreTemp reads the temperature from the diodes in the CPU cores themselves, thus is more accurate for overall die temperature.

Definitely seems to me like either the CPU is getting too much voltage (higher volts would lead to the higher wattage readings, as it's used in the calculation) yet software doesn't read it correctly: which is a microcode / motherboard / bios issue. OR the CPU thermal packaging is indeed flawed, in which case the temperatures would spike under any load, which you are experiencing: higher temperatures result in increased electrical resistance on the die, thus more current draw = higher power consumption. Which is also factored into the wattage calculation by both AI suite and CoreTemp, afaik. However, these aren't 100% accurate. As for the CPU-z reading: 1.8V vcore would have likely killed the chip already, so I highly doubt that's the case, CPU-z might be reading the "CPU Input Voltage" Or "VCCIN" which is the voltage sent to the FIVR from the motherboard. On my board (Maximus VII formula) it does read the Vcore though, but not accurately enough for my liking, since Haswell chips have a voltage per core. Which AI suite reads, and CPU-z does not (it averages them all).

60-75C in games is probably acceptable under the stock cooler, tbh. That said: Intel should have rated this thing for 4.4 Ghz + all 8 threads on the stock cooler WITHOUT throttling under normal workloads. My 4790K @ 1.16V, 4.4 Ghz averages around 40-45C under games, and 58-62C under stress tests (Intel Burn Test) with H100i.

Also, to people using Corsair Hydro coolers: It has been known that the backplate for some H series and some motherboards leaves a little too much slack for 100% good contact. To solve this, simply putting some washers on the backplate screws between the board and the plate itself.. I had to do this with two H series, h80i and h100i, with two boards, my old Maximus IV Extreme and my new Maximus VII Formula. Worth a shot, since it's also one of the things I did to fix my issues. Aside from going through the bios and setting everything to optimised defaults and manually specifying the Vcore, under-volting the VCCIN, etc.

[DHall10]

Sponge wrote:

Also, to people using Corsair Hydro coolers: It has been known that the backplate for some H series and some motherboards leaves a little too much slack for 100% good contact. To solve this, simply putting some washers on the backplate screws between the board and the plate itself

Hello Sponge,

Forgive my ignorance but what washers did you use? I too have had issues with my 4790k and have since rma'd in hopes of getting a better one and want to do everything I can to make this thing work as it should. I am using a Corsair H110 and was getting temps up to 93C on Prime95 Blend and 87C on IBT all on stock settings on an Asrock Z97 Exteme6. Just for reference, using the same cooler, I was getting temps no higher than 79C on a 2600K clocked @ 4.5Ghz and 1.3V Thanks in advance and I hope we all get some results from Intel on this.

[YNomi1]

Hi all...

The same problem here. First of all my rig:

Case : Supermicro 747tq-r1400b (for those who don't know - a server case with huge space and 6 huge 5000rpm fans tha sound like a jet engine )

CPU : 4790K (L4)

Cooler : H100i (with arctic silver 5 thermal compound)

Mobo : Gigabyte GA-Z87-UD5H (F10c bios)

RAM : Corsair Vengeance Pro 2133Mhz 16GB (2x8GB)

GPU : No dedicated video card

OS : Windows 7 Ultimate 64bit

My cpu with stock cooling and f9 bios was running 39-45 idle (room temp about 22c) and when on load ( AIDA 64, Prime95 or intel XTU) was hitting almost instant 100-101c going to throttling.

When i upgraded the bios to f10c i was getting i little lower temps about 35-36 on idle and still 100c when on load but i was getting there more slowly (about a minute).

I went out and bought an Corsair H100i. Now things have improved BUT not very much. I am getting idle temps from 25-28c and the CPU can hold better when on load going to 78-85c but in some instances skyrocketing again to 100c.

Someone can say that i have an almost perfect cooling solution (server tower with 6x5000rpm fans + h100i) but still having problems.

Final conclusion is that i don't care what is the problem of the CPU, bad batch? cpu TIM?. I have spent TOP DOLLAR for a CPU that i was planning to use with stock cooling (so no overclocking) and now i find myself paying more money for water cooling, just to have the CPU operate at normal temps 70% of the time. Worse of all is that i must spent time that i don't have.

The original thread started 21st of August and has been 2 months and still no official answer about the problem !!!!!

Shame on You intel

[nsato4]

I'm also experiencing high temp issues

I7 4790

Gigabyte z97mx-gaming 5

CPU cooler - coolermaster seidon 120v

I idle minimum at low 30c, anywhere near 20% utilization jumps temps up to around 70-80c min.

I've reseated the cpu cooler a few times already with no difference. I've also put the stock intel heatsnk on it as well with pretty similar results. Though am going to try reinstalling the stock heatsink and try again. I believe the mobo is on F4 bios, so will try updating to F5 tonight as well as I think that's the latest.

Other then that I'm really at a loss. Perhaps try to manually set the cpu voltages.

I've noticed that the temps can just very rapidly at times from ~30c - 70c in the blink of an eye.

[SBuck1]

I talked to an Intel tech support guy online for a bit today. He had me run the Intel Processor Diagnostic Tool. He also pointed out that my motherboard was OCing my chip on the fly, taking the multiplier up to 44 from a stock of 40. (4400MHz over 4000MHz). I went into the bios and manually set the multiplier to 40. I changed nothing else. My temps are still high, but my machine is no longer crashing. I think what we may have here are "enthusiast" or "OC ready" chips that aren't really either. If a simple jump up 400MHz done automatically by the mobo can cause 100C temps with HUGE cooling, we have a serious issue. Bottom line, especially if you have a GB mobo like me, try changing the multiplier manually, at least until we hear an official word from Intel about wtf is wrong with these chips.

[APark7]

That's what Turbo Boost is supposed to do, isn't it? Wouldn't manually setting the multiplier to 40 essentially be disabling Turbo Boost?

[SBuck1]

I think Turbo Boost is just the dynamic frequency scaling? The x40 just seems to be an upper bound. The processor still idles at like 800MHz and then scales up to 4000MHz under load. Not sure though, I'm very new to Intel products.

[APark7]

The dynamic frequency scaling is a function of SpeedStep and/or C-States. Turbo Boost actually temporarily "overclocks" the CPU (down to specific cores if I'm not mistaken) to the higher multiplier when the performance is necessary. So if you are never going over 4000MHz, you have effectively disabled TurboBoost. And at a very minimum these CPUs should at least be able throttle up to their advertised boost frequencies without overheating.

[SBuck1]

No argument here, there's very clearly an issue with these processors. Hoping Intel will eventually get around to addressing this, or at least replace the chip.

[APark7]

Indeed. I do find it slightly depressing though that the tech you talked with from Intel didn't recognize that's what was going on. FWIW, most enthusiast motherboards often do have a way to enable the TurboBoost mode all the time, effectively giving you a permanent 400MHz overclock. But it sounds like yours was fluctuating though, so I doubt that was the case in this instance.

My Hyper EVO 212 was delivered today, so we'll see what that does to my temps this evening. Even if it does work, I'm still quite annoyed though that I'm having to buy a decent CPU cooler just to get this thing to run bone stock. And overclocking this thing at all is probably completely out of the question without a decent water cooling solution.

[MDitz]

It is depressing and sad, I agree. I wanted to buy this processor cause I need to upgrade but now I really dont know. It makes you wonder why it takes them such a long time to get back to us. Does it take that long for an engineer to look at it? I really think this is a design problem and if thats the case then that would be really bad. If you buy a processor, I mean how many people dont know how to change there voltage and stuff? Its always, unpack build in and go, then overclock if you want etc. With this one that is impossible unfortunately. My 2600K is so easy to overclock and so fast.. But its getting old a bit. Thanks for everybody's input, if I decide to buy I have at least a few things I can play with.

[MCron]

OK... so I have some new results. Promising results.

I'm at stock UEFI settings.

- VCORE/VCCIN/etc. stock

- Turbo ON

- Cores synced

- Batch: L421C128 (lets include the full batch# guys)

- 3.16.3 kernel (Fedora 20)

Load: Tested an ffmpeg 1080p h264 encode, uses all 8 cores at 100%

Intel stock cooler (properly applied, checked twice!)

Idle: 40C, 800mhz, 0.7v, 40% fan speed

Load: 100C, 4.4ghz, 100% fan speed (don't have a way to read voltage)

Zalman CNPS3900 Extreme with Noctua NT-H1 paste

Idle: 35C, 800mhz, 0.7v, 40% fan speed

Load: 75C, 4.4ghz, 100% fan speed

Instant improvement by cooler alone. The stock Intel cooler can in no way dissipate the rated TDP of the 4790k.

@allan_intel, the stock cooler is a serious problem and prohibits the advertised speed of the 4790k. Your customers should not have to buy a secondary cooler. Maybe you should have not included a cooler with this CPU.

[APark7]

I had some good results tonight with my Hyper 212 EVO as well. I'll post more details in the morning, but I did about a 20 minute gaming session of GRID Autosport with all settings maxed out and only hit a high of 47C between all the cores. A 10 minute stress test in OCCT resulted in a high core temp of around 75-76C (was seeing temps right at or just above 100C before). So far, I'm pretty impressed with what I'm seeing from this cooler. Definitely seems as if the stock cooler (or anything close to comparable to it) is garbage though.

[nsato4]

I was able to keep the cpu multiplier @ 40 while lowering the turbo multipliers to 40 from 44 which has seemed to keep my system stable for now. Temps stay below 75c for now.

[RKimb]

I've tried to find the datasheets for my chip (described by my retailer as Intel 4th Gen Core i7 4790K 4.0GHz 88W HD4600 8MB Quad Core CPU). The nearest model I could find seemed to imply that for normal continuous operation temperatures (Tcase) should be kept below 72C. But I'm not an expert in these things.

[APark7]

Intel ARK is an awesome resource for thees kinds of things. It contains details on just about every (if not all) Intel CPU that has ever existed. The detailed specs on the i7-4790K can be found below.

http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz Intel ARK - Core i7-4790K

The Tcase for this CPU is 74C. But remember, Tcase is not the max temperature of an individual core, but the max temperature at the IHS. For clarity, the integrated heat spreader is the part where you apply the thermal paste (and what gets removed when someone "delids" their CPU). This temperature is usually a bit lower (sometime significantly)than what any individual core is reaching.

[REnso1]

That's interesting, because when I started this thread the very same ARK page stated that Tcase was 72.72°C, I checked it very carefully at the time. So it has been revised upwards 1.68°C during the course of our discussion. Perhaps it was a typo in the first place or maybe there has been some kind of retest or calibration update, would be interested to know which.

[WMoor1]

You're not wrong. Here is a snapshot of the page from July 2nd.

https://web.archive.org/web/20140702014514/http://ark.intel.com/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)

TCase was 72.72°C.

I only landed here after reading tons of concerns about temps while researching this CPU. After reading this thread, it looks like the specs for my next build will be changing. If Intel at least responded, I might have more confidence in buying it but it's pretty clear there's a serious problem.