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.

[RVito]

Another one with overheating problems, I was getting 100 C on load using stock cooler.

I followed ken-Intel and the load temperature was better, lower than 80 C, but freq was capped at 4.2 GHz and I bought a 4.4GHz processor.

My system specs: (I have updated the BIOS to the 2012 version later, but nothing changed about the temps)

My CPU tuning:

What I basically changed from ken-Intel config was set the 3 Active Cores to 44x and 4 Active Cores to 43x, set the Current Limit to 257A to avoid Current Throttling and lower the Core Voltage and the Offset until I found a stable value.

This gives me 4.3GHz at temperatures around 86 C

I still cannot get to 4.4GHz under 99 C

It seems to me that mine is one of those faulty processors, should I ask for a RMA?

[REnso1]

Vitoi - As I understand it, to determine whether your CPU is overheating for sure you need to compare like for like with the same settings as kens and the same test and version of XTU and see if you get temps below Tcase maximum of 74.04°C with the Intel stock HS. If the settings are the same and the heat is significantly higher (accounting for ambient) then you can say it is overheating. Though then you have to decide if the CPU or mobo causing that or even both!

Default turbo settings are 4x42 3x43 2x44 1x44 ; so x44 is the maximum multiplier which only applies with low active core counts, so if 4x44 causes high heat with an intel cooler this does not necessarily mean it is overheating compared to the normal heating for this CPU. To compare like for like you need to put the turbo multipliers at the defaults.

If your ambition is to run at 4x44 in stress testing then you will need a much better cooler than Intel stock, even if the CPU is running with normal temps.

Hope that makes sense.

[RVito]

boolybooly thanks for clarifying that, I really tought it was supposed to run 4x44 because my motherboard default was that.

With ken's settings I have an average temp of 85°C, with peaks of 90°C. That's 11°C more than Tcase maximum with 23°C room temp. Idle temp is 47°C with 3% CPU utilization.

[REnso1]

I am not an expert but from what I have understood in trying to fix my own overheating CPUs, comparing your test /message/260210# 260210 with Kens which I believe was Intel stock cooler, it looks 14°C too hot Vitoi.

I can see your reported core voltage is 1.115v where set voltage is 1.108v so higher by 0.007v. Makes me wonder why the board is overvolting. You said 2012 BIOS so how old is this mobo? What model and does it have a BIOS with I7 4790K support? If not maybe try contacting the manufacturer about that? I dont know if it would be an issue or not but several Z97 mobos I was tracking got BIOS updates to run the Devil's Canyon cpus so maybe.

Also as I understand it, for this test RAM should be run at SPD rather than XMP for a like for like comparison. That can make a difference, though I found my first CPU which was RMA'd was more prone to heating with XMP than my second replacement CPU, though it does still make a couple of degrees difference.

You could also try reseating the cooler to improve contact and if available use optimised thermal compounds to be sure.

Once you rule out the seating XMP and BIOS support, if temps remain high like that, I would suggest discussing what it takes to get an RMA with your retailer then apply whatever tests they use yourself, to check it will meet their requirements for replacement when they test it.

Is all I can think of for now, hope that helps.

[ak25]

How can your total tdp be 37 ?

[RVito]

I have no idea.

[RVito]

2012 is the BIOS version, not the year. Here is my setup:

Asus MAXIMUS VII GENE Micro ATX LGA1150 (BIOS version 2012, latest from asus.com [2014/10/17])

Intel Core i7-4790K (stock cooler)

Kingston Beast 16GB (4 x 4GB) (XMP not enabled, currently running at 1600MHz 1.5V)

And just to rule out case not having enough air flow:

2x Corsair Air Series AF140 (intakes)

1x Corsair Air Series AF120 Performance Edition (exhaust)

I've completely ran out of options here, my retailer (dabs.com) policy for faulty items is to arrange the return with the manufacturer and I don't really know if I'm willing to go through that process right now, I'm moving to South America in 30 days and I'm not sure if its enough time to get my RMA accepted and a new processor in my hands before I move.

[RChiz]

Same would happen to me, total TDP around 40W while stress testing XTU with default settings including 88W limit. Max frequency would only be 4.1GHz because I would hit current limit throttling at default 105A. Raising current limit to 120A allows me to reach stock 4.2GHz fully loaded. Even stressing with OCCT only gets up to 54W. On heavy stress testing I will always hit current limit throttling if I set current limit for any value up to 255A. If I set current limit of 256A or higher, I will always thermal throttle.

I've been wracking my brain for weeks trying to figure this out on my Maximus VII Hero. I finally got fed up and bought a cheap MSI Z97 PC Mate board to try. Guess what, it works fine. I actually can hit the 88 W limit when stress testing and my temps are 20-30c lower under load. Something is not right on some of these boards or are just defective. I've tried 3 different bios, CMOS cleared too many times to count, uninstalled AI Suite. I now need to find a new feature rich motherboard that won't have problems. The PC Mate is fine for troubleshooting but is too spartan for my daily driver.

[RVito]

Wow, good to know that... bought the most expensive mATX board willing for top quality and that's what we get.

[RChiz]

Yeah, I think the Gene is pretty much the M-ATX version of the Hero and we seem to be having the same problem. I've tried bios 1104, 2012, and 2201. I'm wondering if it's some bios setting that needs to be changed or actual defect. Maybe we are on the path to figure something out.

[DVu3]

Wow, I haven't touched my Hero board yet to check it out. It would be suck to not be able to use it at full potential. I have a paper due tomorrow so don't have time to test it yet, but I do hope that it will run fine though. My Asus Z97 AR board is running quite hot with the i7 4790K for now, without Turbo boost, it's already at 90C most of the time, some occasion hit 95-98C. The stock cooler seems to absorb the heat since it's hovering around 40C (it should be closer to the temp of the CPU, right? Just weird!). In my previous setup, I tried reseat the cooler 3 times but it's always 27-28C on the cooler. Seems like there are problem with the IHS which may have a gap so heat cannot transferred out efficiently.

After tomorrow I'll check the Hero board out and see what's gonna happen.

[RChiz]

I'm glad you said something about the Asus Z97 AR, I was thinking about it as possible replacement. My 4790k idles around 32c but shoots up to high 80's during non-avx stress and 99+ with avx load like IBT or Aida64 FPU. I started with Hyper 212 then went to H80i and now using Swiftech H220-X so cooling isn't my problem.

[DVu3]

I'm running computation for chemistry with it and it peaked to 100C in seconds with Turbo boost. Once I transfer it to the Hero, I will use the Hyper 212 Evo to cool it to see if there is any change. I do hope to get a "good" CPU per say, I hate to go back and forth with MicroCenter to exchange for the CPU all the time. It's like 50 miles round trip every two weeks (15 days return period) until the "good" CPU.

For the AR, I haven't played a lot with it, tried ken's setting and it peaked to 100C, then change to TB off and I've been running it like that since. I'll have the GSkill 2400MHz RAM tomorrow so let's see how it works with the crazed CPU. Probably 70C idling and 150C full load running at 2400MHz XMP for the RAM. (Not that the CPU is gonna go higher than 100C)

[RVito]

Now that you stated that motherboard was your problem I payed attention to this VCORE value on the motherboard. Is this normal?

[RChiz]

The thing is that the CPU is probably fine. It's the motherboards. Don't know if its something wrong with bios setting or defective component like power vrm.

[DVu3]

The Vcore changes based on the load of the CPU. The 1.8V value looks like the Vinput than the Vcore, the 0.7 - 1.199V is probably the Vcore.

The CPU does have variation between them. My previous one sucked badly. While the new one is marginally better. I have another one that I haven't tested yet. I have till Friday to test it before deciding which one I'm keeping.

[RChiz]

That is probably supposed to be CPU input voltage not Vcore.

[RChiz]

Sorry, you beat me to it.

[SPark48]

Vitoi and slaveonepunx,

I'm not sure if you've read my posts before, so her is some background info:

I also have been having the same problems with the 4790k overheating. I tried it on multiple boards, all from ASUS - Z97-Pro Wifi, Maximus VII Gene, Maximus VII Hero, Maximus VII Formula. My 4790k was passing Intel PDT, but as soon as I ran a third market CPU stress test (OCCT, Prime95, etc.) my temperatures would go up to 90-100C almost instantly. This was with no overclocking at all. I first tried to reseat HS several times with the three of the best reviewed thermal pastes on the market, yet nothing helped. I purchased a Corsair H60 cooler and that got it down to mid 90C's, but obviously I felt that something was wrong if a K processor was giving 90-100C straight out of the box with no overclocking on an aftermarket cooler... So I contacted Intel about it and they said that since my CPU passes Intel PDT, there was nothing wrong with my processor... Needless to say, I was not happy with that answer so I ran Intel XTU's stress test and sure enough, same incredibly high temps. So after several nasty back-and-forths with their CSR, Intel begrudgingly agreed to do an RMA for me.

A few of things I wanted to point out to you guys is:

1) Vitoi, you mentioned that you're going out of country in about 30 days. Intel's RMA process took me a total of 3 days from the time they approved the RMA process. All I did was call them about their express RMA service and basically you have to pay for a new processor on a credit card, they will send the processor to you first, and then when you get the new one, put the old one in the box and ship it back. They overnight the CPU to you so I got it the very next day. Once they get your processor back, they refund your credit card. So the process is quick and you should have plenty of time. However, keep in mind that Intel charges $25 for shipping to do an RMA. Not sure if you have experience doing RMAs with other companies, but Google, Apple, Amazon, Best Buy, all of these places take care of 100% of the cost of shipping, but for some reason, as big as Intel is, they want to charge their customers to ship back a product that was made defective... Because the overheating processor is somehow OUR fault... So basically RMA is very quick, but it will cost you $25.

2) slaveonepunx, I noticed you said that you think that the problem is the motherboards and not the processor. Well, I got my replacement chip and used it on the SAME motherboards mentioned above that I was getting crazy temps on. At stock settings, the new processor ran in the mid 50C's on max load using the H60 cooler. Now, my processor is overclocked to almost 4.9GHz and OCCT, Intel XTU, and AID64 give me maximum temperatures in low 80C's on max load. Still kind of hot, but at least it is stable and acceptable to run in short bursts. To do this, I first had to clear CMOS as the ASUS default settings were the main culprit for the high temps. After that, I changed the settings according to ken's recommendations, and then increased multiplier and voltages until I was stable. Right now, the processor is in the Maximus VII Formula and *knock on wood* it's been running grea (other than Prime95 Small FFT test). I attached CPU-Z and HWMonitor below for reference. As I stated previously, I believe the problem is both that the motherboards have incorrect stock settings that don't play nice with the 4790k and also that some of the processors themselves are defective. If you haven't already done so, I highly suggest clearing CMOS, then changing to ken's recommended settings. If that still gives you problems, I suggest an RMA.

3) Weird thing I noticed is that for me, on AIDA64, running ONLY the FPU test will give me 90C+ temps, but if I run CPU+FPU together, my temperatures are OK. Is anyone else seeing this? I admit I really have no idea what these tests are doing to stress the CPU, but one would think that CPU+FPU together would be more stressful than just the FPU? I'm guessing this is the same thing with Prime95 and it's Large+Small FFT test, although I haven't had the courage to try it because I'm happy with where I'm at now and don't want to p***-off the processor.

I hope that helps you guys out a little bit.

[RChiz]

Thanks for the suggestions sampark1980. Each time I flash a new BIOS I use the clear CMOS button afterwards and keep it held down for at least 10 seconds. I set defaults and only change from AHCI to Raid as well as disable multicore enhancement so the cpu acts like Intel intended. It hits current limit on anything set 255A or less so the default of 105A doesn't work. TDP never goes above 50W at stock settings. I was hoping it was just a bad CPU but since I tried the 4790K in the MSI mobo and it acted normal along with staying at reasonable temps and reaching its designed TDP in the 80W range I'm thinking its the ASUS board.

I guess the only way to be 100% sure is to open the G3258 that I bought bundled with the MSI board and drop it into the Hero but I'd like to be able to return the processor to apply the money to a new mobo. I mean, if the 4790K works right in the MSI board, it has to be the Asus board right? I can't get the board to a power load close to the 88W design even under Prime when voltage and temps get crazy. What causes my max TDP to never get to 60W?

I can't RMA the board because I was a knucklehead and bent some pins installing the socket cover incorrectly. I'm so used to the old socket covers that go under the clamp that I wasn't thinking and tried to do the same on this one that goes on top.

I think AIDA64 runs cooler with CPU+FPU because it cycles the test from FPU to CPU. You will see the temps get hot when its testing the FPU then the temps will drop as it tests CPU and it keeps cycling like so, You would see the voltage change if set as adaptive or offset as the test went from FPU to CPU. That has always been the case for me on this and previous CPU's.

[SPark48]

slaveonepunx,

That makes sense about the AIDA64 cycling between tests. That would explain the acceptable temperatures where as FPU alone was 90C+.

I apologize, but reading back on your posts, apparently you aren't having the same problems with overheating on 4790k, you are actually having problems reaching max power of 88W? I know this is a really dumb question, but did you make sure the processor says 4790K instead of 4790S? Your wattage numbers are suspiciously close to what I am getting on my 4790S server machine... I think for sure there is something wrong with your board then if you're not getting full power to your processor while the MSI is giving full power. Did you speak with Intel CSR yet?