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.

[aanwa1]

i got similar results to you too with both limiters set to 110W and 255A

my mobo is Z97X Gaming 5

[SPark48]

ShadiElwan,

It's your processor so you can choose what you want to do, but if it was me, I would NOT be satisfied with a chip running in 90C-100C on stock settings even with a stock cooler. When Intel says that the recommended heat for this processor is 74C, that SHOULD (and that's a BIG should) mean that your processor should run at a max of 70C-78C at stock settings with their stock cooler; at least that's what it should mean from a marketing/advertising standpoint. If you're running 90Cs with stock cooler, that means with aftermarket cooler, you could see into 70C-80C range depending on the cooler you get, meaning if you decide to overclock the processor, you could be back up into the 90C-100C range again! That brings back the same point of why buy an unlocked processor if you cannot safely overclock it? Why not pay 40 bucks less and get the 4790 instead?

With the new RMA chip I received, at stock speeds I got temperatures in mid 60Cs with high of 71C with stock cooler and with Corsair H60 I got low to mid 50Cs with high of 62. Correct me if I'm wrong, but this is the correct temperatures that the processor should run, not 90C-100C on stock settings with stock cooler.

Hope it works out for you one way or the other.

[SElwa]

sampark1980

You said: " I got temperatures in mid 60Cs with a high of 71C with (stock cooler) ". May I ask you a few questions if you don't mind?

1- What is the stress test type you have done to get that temp (FPU stress or CPU stress)? And with what program? Test duration?

2- What is your Vcore and CPU frequency during that stress test?

3- What is your room temperature?

4- Is the air flow inside the case good or not?

5- Your motherboard model?

Note: I get below mid 70Cs @4.2Ghz (Vcore 1.109v) during the normal stress test only using XTU, AIDA64 Extreme, @4.4Ghz (Vcore 1.150v) I get 79c using XTU and steady 73c using AIDA64 Extreme except when I use torturing test using prime95 28.5, AIDA64 Extreme (FPU test) I get ~97c but as I understand those tests (FPU stress test for example) are not a real-life scenario, for the normal usage in the worst case when the CPU is fully utilized I get max peak of 81c (for a short time returning to 79c or 69c-75c dependent on what apps/games are running) @4.4Ghz all using stock cooler, room temp 26c.

Update: using IntelThermalTest v0.2 (c) 2014 by Karwos I get 79c and passed the test.

[kpago]

I had same problems .i used this paste and now i have 77 c at turbo with h 60 watercooler.with stock paste i had 100c.take Prolimatech Thermalpaste PK-3 Nano 1,5gr .

[SPark48]

ShadiElwan,

1) The 3 tests I have run (successfully) are OCCT, EVGA OC Scanner, and Intel XTU to get those temperatures. I ran the tests both with stock cooler and then again with a Corsair H60i cooler. I also ran the tests 3 times each. The tests were run for a duration of 5 minutes the first time, then 10 minutes, then 15 minutes. When I get the time, I guess I can run it for longer. I just fall in the category of if the processor can stay stable under torture for 15 minutes, I don't want to keep torturing it for 12 hours more, especially if it is a daily use computer. I could be wrong, but in my experience, usually about 15 minutes is sufficient time to see if a processor will stay stable.

2) VCore was set to auto in BIOS and it was running at 1.15V. It was at stock frequency of 4.0GHz+turbo.

3) Room temperature was 70F which equates to 25C.

4) I actually have 2 cases that I ran these tests in to make sure the new processor was good. One is a Silverstone PS07 with 2 120mm intake fans and 1 120mm exhaust fan (all 3 are Corsair Performance). The other case is a Silverstone GD09 with 3 120mm intake fans (Corsair Quiet) and 2 80mm exhaust fans (Arctic F8). Airflow inside case isn't the best by any means, but it is adequate with good aerodynamics/flow of air. I know some schools of thought think "chaos" is better for cooling, but I've always built computers with good intake-exhaust flow of air through the case.

5) The Silverstone PS07 has a Asus Z97 Maximus VII Gene and the Silverstone GD09 has an Asus Z97-Pro (Wifi/NFC Model). Both motherboards ran the same settings in BIOS and both gave very similar results as far as stock temperatures go. I was able to overclock to far higher frequencies with the Maximus VII, though, with stable OC of 4.896 GHz.

Hope that information helps!

Update: I didn't even bother reading your update until now. Sorry. I haven't tried the AIDA64 test, but the Prime95 is definitely giving me problems (99C+). It looks like both of our "new" processors are running very similarly. As you said, I don't know a single application that will run those Prime95 and AIDA64 stresses in a real-life scenario. My previous processor WAS actually giving high temps running some applications. Games such as BioShock Inf, NBA 2K15, were pushing temperatures to 86C. The new (post-RMA) processor, on the other hand, only goes up to high 60C's running these same games. All things considered, I'm pretty satisfied with this.

[SPark48]

pagos,

That's great to hear, but it is highly doubtful that just the change from stock thermal paste to the PK-3 dropped your temperatures +23C. Perhaps it was the way that your cooler was sitting or a slight error during installing the fan the first time.

I am currently using Arctic 5, but I have also ran these same tests with IC Diamond 24 Carat and Arctic MX-4 and the difference in temperatures were negligible. The important thing, though, is that you're running good temps now.

Speaking of IC Diamond thermal paste, has anyone used this brand before? It received a lot of good reviews so I thought I would try it out. The thing is, it is a LOT thicker/dense than most of the other pastes on the market. It comes out more like a glue substance rather than a paste. Is that how this paste usually is? Or did I just get a bad/old batch? The viscosity of it just feels so different from other pastes. Anyone know?

[jband2]

I'm handling the RMA with intel , install the intel extreme tuning program , and the temperature reached 78 degrees at a speed of 4.2GHz processor , by setting the values of the test image INTEL . The PASS test was still no solution to this issue ?? microprocessor is the problem ? motherboard ?

[SPark48]

bahianus, I'm not 100% sure what you're saying so sorry if this response doesn't make sense. I will try to summarize what's going on so far, at least from my point-of-view.

My first 4790K passed the Intel PDT, but when trying to run Intel XTU stress test and other third party CPU stress tests, I would get temperatures of 90C-100C at stock settings. ken-intel gave us some settings to use in XTU and using his settings, I was able to get my temperatures in the Intel XTU down to acceptable limits. The problem is, his settings also limited the processor to 3.6 GHz. I contacted Intel and at first they refused to do an RMA; since I was passing the PDT they said there was "nothing wrong with my processor" even though their own Intel XTU at stock settings was giving me 100C. After some back and forth with them, they reluctantly agreed to do an RMA for me. The "new" processor I received is running in 60C's at stock settings on the same tests (except for Prime95 Small FFT which is still giving 99C) without having to limit the processor with the settings suggested by ken-intel.

Here are a few things that I firmly believe so far:

1) Some 3rd party stress tests (ie Prime95) are giving instructions to the CPU that probably no other application will give to the processor, and these instructions are causing the CPU to run at extremely high temperatures.

2) Stock (out-of-the-box) settings of many motherboards, regardless of brand, are unnecessarily high and causing the 4790K to run hot. Clearing CMOS has fixed this issue in most cases.

3) Still, all things considered, some (many) of the i7 4790K's that were manufactured by Intel are "faulty" and giving temperatures that are way too high and need to be replaced by Intel.

Other than that, the other problems reported here I don't believe are specific to the 4790K. For example, proper case cooling, proper installation of CPU fan, contact issue with some Corsair liquid cooling systems, thermal paste differences, etc. If these were the issues that were causing the 4790K to overheat, we would be seeing it in more processors and not just the 4790K. But this heating issue, as far as I know, seems to be limited to the 4790K. There are some reports of the other Devil's Canyon (i5 4690K) having some issues, but nowhere near as widespread. However, the Haswell family as a whole seems to have quite a few reports of overheating, so it may be a bigger issue; I don't know for sure. However, most of the overheating chatter seems to revolve around the 4790K.

[HFigu1]

I was able to RMA my I7 4790k because of a CPU Frequency fail. Got the new chip today no overheating issue, but I STILL FAIL CPU Frequency Test! I submitted a ticket to Gigabyte for the motherboard for this issue now.

[jband2]

sampark1980

My 4790K passed the Intel extreme tuning utility, but when trying to run Intel XTU stress test and other third party CPU stress tests, I would get temperatures of 80-90C at ken-intel settings.

The problem is, his settings also limited the processor to 4.2 GHz, when this run at 4.4 GHz.

then I guess my processor has temperature problems

[TVere1]

Hello abcanw!

I think i could use your hel here. I have the sam mobo as you do ( GB z97x Gaming 5 with BIOS version F5 ). I was following this thread, and I did reset the default optimized setting in the BIOS, however I discovered that the The Turbo Boost Power Max stays at 88 W and the Processor Current Limit also stays at 105 A according to the Bios. As for the RAM I have Kingston HyperX Fury 1866 Mhz, and in Bios XMP is disabled and I cant change that option. And this is what my Bios says. According to XTU ( after I reset The Bios ) the The Turbo Boost Power Max is set to default ( 39xx Watts ) and the Processor Currant Limit is also at the default setting ( 1000 A ). So I dont know which data I should belive in. I also have some problem with the latest version of XTU. It doesnt allow me to set everything that ken-intel told us to alter. For me it loks like this :

So I decided to use an older version ( XTU version 4.3 ) and this version has all the features we need :

After reseting the BIOS I used the settings that ken-intel provided for us and run the stress test. My result are here:

With these setting I dont have throttling issues as you see, but the overheating remaind as the main problem. The TDP is good and i have 4.2 GHz on all 4 cores.

Any idea what the problem could be. I use the stock cooler and I bought a quite good thermal paste.

I would like to know how to do this PDT test. Which software should I use and where can I find it. I am totally amateur in this field, so if you could help my I would much appriciate it.

Thank you in advance!

[REnso1]

@Szamkivetett, well I expect some GA Z97X Gaming5 user can advise you best, since I am using a GA Z97MX Gaming5, the micro ATX version and it doesnt have an F5 BIOS... yet ... and it definitely does show strange behaviour with the limiters which the Z97X doesnt. So apples and oranges re settings.

However I can give you one piece of advice which is that the Intel heatsink is not adequate for stress testing or for intense processing. If you want to reduce temperatures you will either need to disable turbo and possible underclock to 3.8 GHz or less, otherwise you will need to get the best heatsink you can afford and which will fit in your case. If you dont know about how to do this this you should be aware you may have to take your computer to bits to fit it and you would be wise to use a static strap while handling components or at least ground yourself before handling them and dont wear fleece or synthetic sole shoes and then shuffle on a carpet or you could generate a spark which could fry a chip. Strap is safer, very cheap, connect to electrical ground/earth.

I dont understand why your RAM profile is not available for you, have you tried different forms of input eg + - keys on the numpad when in BIOS, it ought to give you a profile to select and its worth using. Is the RAM a model # which is listed on the QVL (qualified vendors list) for the motherboard?

[aanwa1]

Hello there Szamkivetett

when i saw that the latest version of XTU doesn't show all the options and settings for z97x gaming 5 mobo i decided to tune things in the bios directly and leave Gigabytes's EasyTune and Intel's XTU because they were unreliable so i advice you to test setting directly in the bios to be more sure.

when you reset your bios to default settings the The Turbo Boost Power Max and the Processor Current Limit along with most of the settings will be set to (Auto) and dont stay at 88w and 105A (in the bios not XTU).

In the Bios when i set the both limiters to 110W and 255A the processor will run at 3.4Ghz, same results as GA Z97MX Gaming5 mobo, so right now im waiting for response from boolybooly who is contacting Gigabyte about this issue.

while waiting for an answer i run default optimized settings (which make the processor runs at 4.2Ghz and at 77 degree max, my cooler is scythe katana 4) ,then i wanted to change my memory frequency to 1866 Mhz so i enabled XMP but that somehow made the processor run at 4.4Ghz and above 90 degrees, so i turned off XMP and changed the memory frequency manually to 1866Mhz and that made the processor run at 4.2Ghz 77 degree max again.

I don't know why even with using the default optimized settings you are getting high temperatures, but its probably what boolybooly mentioned in the previous comment (stock cooler) or could be bad cooler or thermal paste installation, so i advice you to turn off turbo boost in the meanwhile.

[REnso1]

You are right abcanw the temps for Szamkivetett's test are way too high. Ken-Intels test with stock cooler produced 74.1°C so 96°C is too high. But you need to compare like for like Szamkivetett so you need be running the latest version of XTU and change settings in BIOS if they are not available IMHO.

I have been in touch with Gigabyte whose rep did not seem able to understand what I was saying, as follows, you decide. If you can do better abcanw I would suggest you start your own eSupport query with them as they would probably benefit from hearing this from other voices. I seem to confuse them for some reason, perhaps I am including too much detail.

Thankyou for attempting to replicate with AIDAx64, and the screenshots which are helpful. I think I see why replication did not succeed. The previous replication attempt did not apply the TURBO POWER LIMIT (WATTS) 110W limiter specified by Intel, as mentioned in the second part of my first post, this must be set approx 110W to comply with Intel rep recommendations in the linked thread on the Intel site and this should then replicate the unexpected behaviour with the amps limiter FYI. - To replicate, load defaults and set CORE CURRENT LIMIT (AMPS) to 255A and TURBO POWER LIMIT (WATTS) to 110W, the clocks will then not exceed 3.3GHz in AIDAx64 for example. - The fault is only apparent when both limiters are set. I had to recheck this myself in the light of the difficulty with replicating the fault. With only the core current limiter set the clocks are normal. I also cleared CMOS using the mobo jumper and rechecked the fault, which still occurs with BIOS F4 defaults after clearing CMOS when both limiters are set as above. - Thankyou for your suggestion re cooler. However in this case it doesnt follow logically from the results as the temperatures with a Noctua NH-L12 are below 55°C when operating at 3.3GHz. Also when only one of the limiters is set the clocks are normal and temps rise above this without restricting the clocks. Please see screenshots FYI. This is why I believe the problem is more likely related to the BIOS.

Dear Customer, Our BIOS team would like to know why you need both functions run at the same time? Can you use the default Auto setting?

Please tell your BIOS team that Intel support recommend using both limiters in advice here URL https://communities.intel.com/message/260210# 260210 https://communities.intel.com/message/260210# 260210 In addition the BIOS may need to be reexamined and a version made to run to Intel's spec. Intel rep ken-Intel stated "...feedback and our internal testing of certain motherboards have shown that some motherboard processor power settings are not set to Intel default specs and are thus requesting more power than expected." My retailer SCAN replaced this mobo model under RMA for the same order twice, both times for causing CPU overheating. My current mobo is still causing overheating with an I7 4790K CPU hand picked by SCAN for lowest temperatures eg optimised default settings with XMP (QVL 2400 RAM) enabled reaches 78°C when testing with AIDAx64 beta with a correctly fitted high quality third party Noctua NH-L12 heatsink. Intel ARK states Tcase is 74.04°C. Therefore I have to underclock the CPU on this motherboard to keep temps in spec. NB I am not trying to overclock currently turbo is completely disabled. Other parties in the thread have experienced overheating with this mobo model. Since the limiters do not work correctly and I cannot follow Intel recommendations it is not possible to progress with this mobo/CPU without cooperation from Gigabyte BIOS team. An alternative BIOS in which the limiters work correctly and power is drawn exactly to Intel spec is needed, a "green" BIOS if you like. Also please see ken-Intels further post which suggests there are channels of communication between Gigabyte and Intel which could assist Gigabyte BIOS team https://communities.intel.com/message/263736# 263736 https://communities.intel.com/message/263736# 263736 ken-Intel also states that other motherboard models are able to use both limiters, use of both limiters is justified because they limit different aspects (core and uncore) of the package power draw, using both limiters does help, it is expected that both limiters can function at the same time. I hope that helps

Dear Customer,

Are you saying the CPU Temperature reading is incorrect using both limiters?

Could you please clarify what is the major concern issue on the last reply?

The previous reply from Gigabyte asked on behalf of the BIOS team why both limiters are needed and my last reply answered that specific question for the BIOS team. I am sorry if it was not easy to understand, I will try to explain again...... Temperature needs to be controlled because this mobo/CPU combination causes overheating. By Intel's recommendation both limiters are needed to control power draw and therefore temperature. However you cannot use both limiters with this mobo as advised by Intel because there is an additional replicable fault with the F4 mobo BIOS which makes it run at a low clock when you apply both limiters. Specifically, with the Gigabyte GA Z97MX Gaming5 using both limiters [eg Amps<255A (Intel recommend 105A) Watts=110W] causes the CPU clock to reduce to approx 3.3 GHz in stress testing, which the clock should not be as it should be 4.2GHz (ie default clock for 4 cores in turbo mode). So there are two faults, the limiters dont work together and the mobo overheats. I dont know if these are related...... I am sorry that the lack of formatting in these reply boxes makes it so difficult to communicate clearly. If you still do not understand this reply, would you please pass this reply and my previous reply to your BIOS team as they should be able to decipher it...! Thanks very much</strong>

[REnso1]

OK I can add the following recent reply from Gigabyte.

Dear Customer,

We tested on 4790K with GA-Z97MX-Gaming 5 (1.0). The CORE CURRENT LIMIT (AMPS) and TURBO POWER LIMIT (WATTS) might need at least 160 for turbo ratio to be fully worked.

I am testing this with defaults and XMP and finding that 160 setting for both limiters does allow the CPU to operate at 4x 4.1GHz in XTU. Which is interesting because the minimum turbo is 4x 42 meaning the combined limiters are restricting the clock to below the default turbo settings. Temps max at 69°C with a Noctua NH-L12. This is all due to current limit throttling which is averaging 96% during the XTU test, power limit throttle remains at 0% and averages 0% the whole time yet it cannot be set to 110W without limiting the clock to 3.3GHz.

Obviously these limiter settings are different to ken-Intel's recommendation. I am wondering if this is due to to the mATX GA Z97MX Gaming5 using what http://www.eteknix.com/gigabyte-z97mx-gaming-5-lga-1150-motherboard-review/6/ one review described as a "low-end 4 phase VRM". I would be very grateful if ken-Intel or someone with the necessary experience could say whether that makes sense.

[KenF_Intel]

Thanks for your continued help with this BoolyBooly. We would not have made the progress we have without your persistence.

I won't be able to comment on the VRM component, but a fair question for Gigabyte is why the Gaming 7 I have does not have the same 160W requirement as your Gaming 5.

FYI, I have created a document that I hope to use as a 'one-stop shop' for helping others that run into overheating due to BIOS power settings.

The document is here:

It is also part of the Featured Content on the processor community main page ( ).

I will keep that documented with information I get regarding settings that work with different motherboard brands/models.

[idata]

Ken,

We really do thank you for all the help that you have given us for this crazy problem situation.

I do have one question, however. If the recommended maximum temp of the 4790K processor is 75C, why doesn't the internal thermal throttling cut in before 100C? To me, this implies that 100C is an extreme, but acceptable limit for the processor.

Is there any way to adjust the thermal throttling limit (downward) through XTU? If not, it might be a good idea for the next processor generation, and XTU.

Thank you again.

[SBuck1]

Ken,

Have you had any success contacting intelligent people at Gigabyte? I've been pounding my head against the wall with them trying to get them to understand there's an issue with the "latest" BIOS for the GA-Z97X-SOC. Additionally, has Intel made any progress identifying or rectifying the issue with the thermal adhesive between the chip and lid? The internet is covered with accounts of people fixing their 4790K overheating issues by de-lidding.

[REnso1]

Thanks ken-Intel for your helpful advice and encouragement. I have unmarked your previous as correct answer and instead marked the post with the link as correct so people can find /docs/DOC-23517 the document easily, which is a useful resource for people trying to set up an I7 4790K system so thanks, I am sure people appreciate the effort you have gone to to provide that.

In other news I have sent a question to Gigabyte about the higher than normal BIOS power limiter settings with the GA Z97MX Gaming5 and will relay any reply here when I get one.

[kpago]

hi.just buy i74790k with gigabyte z97mx-gaming 5.i have the same problem but i try some thinks and my opinion is the problem begin with xmp profile.i have f4 bios update.run aida64 test with xmp disable.i have corsair vengeance 2x4gb 1600mhz.so when dint use xmp and my memories run at 1333 mhz the temratsures was about 77c system stabilty tests and turbo cpu 4,4ghz.i use 2 tests aida and intel extreme diagnostic.when change and anable xmp i had 100c with same tests.at banchmark with intel test my pc shut down.so if the problems is with memorie powers???i dont know just asking for more opinions.i have 1 day my cpu.thermopaste is ok and i use h60 corsair watercouler.

[REnso1]

Update, Gigabyte got back and said they had turbo working no problem but did not specify any settings.

This was a bit confusing as I am seeing a repeatable problem my end using the settings in ken-Intels screeny. So I cleared CMOS (again, just to be sure) and tested step by step after loading defaults and found the amp limiter issue only applies if the wattage limiter is also set. This may explain why Gigabyte replication was not successful, they may not have applied all the settings specified and just focussed on the headliner, so I will have to check that with them because if they cannot replicate it means its just my setup.

This setup, if the mobo is using default settings with just amp limiter set it works OK below 256A eg 255A and is limited by the turbo multipliers. (In this instance XMP does not appear to make a difference.) If the wattage limiter is set 110W and the amp limiter is below 256A eg 255A, the CPU will run at 3.3GHz in 4 core stress tests with other settings (eg vCore and offset) at default, but if I set each limiter individually or set the amps at 256 or over the clocks will reach their full turbo setting eg 4x 4.2GHz.

I would expect the logic to be that if the turbo wattage is set 110W and Amps are set 255A they would not interact and only one of the limiters would restrict the clock at any one time ie the one whose conditions were being met. As it is, both set as above will not pass 3.3GHz in a stress test. So somehow the two limiters are interacting and providing a greater constraint when both are set than when each individually is set.

Is that how it is supposed to work?