4790K overheating with NH L12

[Sylvester]

I was researching delidding this morning when Gigabyte got back to me saying they cannot replicate the issue which has thrown the cat among the pigeons once again. What if I have a dodgy CPU? I really need to get verification from SCAN now on their testing procedure.

Gigabyte offered this screenshot as proof of normal function in their test set up. But they dont say anything about their BIOS set up. I will have to go back to the Intel HS/F now to get a like for like comparison. And I need to buy AIDAx64 as the test process has taken so long the free trial has run out.

[watercooled]

In the screenshot, it looks like the CPU is turbo-ing to x42 on all cores, which is what you'd expect until you run into thermal issues. But that would be with MCE off - as you've found out, it seems it's turned on by default on their boards.

You should be able to use Prime95 rather than aida for testing, rather than spending more money.

I would still advise against delidding unless it's something you really wanted to do anyway. And as you say you'd be stuck if the CPU is faulty besides an IHS issue.

[Sylvester]

Actually, AIDAx64 does seem to run at 4.2 with default settings. I just checked. I am trying to provide replication guidance for Gigabyte, and if they cannot replicate then it must be the CPU. SCAN have offered to take another look at it, since they only applied the Intel Processor Diagnostics tool test last time.

Screeny below was default BIOS with SATA on RAID which for some reason is necessary to boot with the Crucial M550 1Tb. I seem to remember I could not see it when installing Win7Ultx64 unless the mobo was set to RAID. I used Ceramique on the Intel HS/F which I later checked and replaced with NT-H1, seating and contact region were faultless. NT-H1 did improve temps but still got throttling in AIDA at default even with fan manually set to full speed in BIOS.

[Bonebreaker777]

Fantastic perseverance and yes at the same time dreadful experience.

[RobbieRoy]

Fantastic perseverance and yes at the same time dreadful experience.

I second that! Don't give up Sylvestor!

[Sylvester]

Thanks everyone your interest and encouragement are much appreciated, I wont quit now, I want to get to the truth and get my new PC working. I will report any developments.

I am also encouraged by the good CS from Gigabyte and SCAN. They are both focussed on problem solving and cooperative where they can be which helps.

I will wait for Gigabyte to get back on the replication, there is just a slim chance that setting SATA to RAID might be doing something unexpected, else I may have been unlucky enough to get an overheating mobo and an overheating CPU in the first place and then missed the CPU on the first RMA because the mobo took the blame.

[watercooled]

I agree, kudos to Scan and Gigabyte for trying to replicate the problem etc.

[Redroth]

Hi Sylvester, just a quick update from me after phoning up scan end of last to get more details about there testing procedure before contacting Asus as regards to my problem. I found that they had only tested my motherboard with a 4770k and no haswell refresh chip. They have agreed to take everything back under RMA and test with a 4790k. I would have personally expected them to test the board with a haswell refresh originally. Ill get back to you when I actually get some feedback from scan, but it might be worth checking with can if they actually checked your motherboard with a 4790k originally.

If that is the case and both our boards are faulty they chances of us both getting a bad CPU and MB from different manufactures and the same supplier at the same time.

[philipmathews]

Sounds like the CPU is faulty, they were supposed to have the thermal paste issue on the CPU cap on these particular CPUS.

[Sylvester]

Redroth, I know they didnt test with a Haswell refresh but they still found a mobo overheating fault with a 4770K, so there must have been something out of whack for that to happen. It could be that the second mobo has a fault or that the Intel test was not rigorous enough to bring out the problem with my CPU, hard to say, so another RMA is in order.

Gigabyte got back to say they did not replicate my levels of heat. Their screeny below.

UPDATE: SCAN has started an RMA.

[Sylvester]

OK I have some kind of answer for how the CPU could have passed Intel's Processor Diagnostic Tool test (which is the method used by SCAN to test it) while still overheating. It seems that the test will pass a CPU even if a core reaches 100°C.

[Sylvester]

UPDATE : FYI SCAN got back to say they replicated the CPU overheating and when they tested replacement CPUs using AIDAx64 and Intel retail HS/F they discovered a bad batch of three overheating 4790ks before the fourth CPU proved to work within spec under the same conditions. Here is their screeny showing the good CPU operating within spec at 4.4 turbo, running AIDAx64 under the retail Intel HS/F.

EDIT apprarently this was not made with an Intel HS/F but with a Zalman CPNS700.



So all these CPUs will be returned and I hope they will stimulate Intel to consider its manufacturing process and package design.

I hope this will be useful information anyone else with thermal issues.

EDIT a test using AIDAx64 and the Intel HS/F cannot be used as a standard to test for RMA, see below. Sorry I was wrong to assume they used an Intel HS/F to test just because that is what they told me to use. I did not realise they used a different cooler as they did not specify the details of the test, I will need to check Gigabyte's result now as well as they also did not specify a cooler.

[Redroth]

I also phoned scan (funnily enough getting the same person who talked to you) to say they had completed similar tests on my MB. I have been told they will be testing CPU's tomorrow to find a stable one to send back to me.

[Sylvester]

update: replacements ran hot in AIDAx64 99°C just under throttle temps with the Intel HS/F. OK that is better than 100°C but not the same as the 82°C screeny I was sent above.

So I checked if I had been sent the right parts and asked how the screeny was made, apparently the 82°C screeny was made using a Zalman CNPS9700 cooler, not the Intel HS/F. So I was mistaken in assuming they used the Intel cooler, that was because they told me to use the Intel so I assumed it was a standard bench method. I will edit my previous post to reflect that, sorry if it mislead anyone, I am in the middle of testing but I thought I better make this clear the moment I knew about it.

So to summarise, SCAN accepted my RMA based on Intel HS/F AIDAx64 tests (on their instruction) and agreed it was overheating after testing themselves.

They then replaced the components but used a non Intel cooler to test these before sending them.

They are also recommending that when running the AIDAx64 test that a better than Intel cooler is required to prevent overheating for this model of CPU (4790K) even if they are good CPUs.

So while an Intel cooler can be used for testing it is not the recommended cooler for stress testing and a thermal throttle event in AIDAx64 using Intel HS/F does not guarantee an RMA.

I have tested with the Intel HS/F and the SCAN tested replacement CPU does suffer thermal throttling in AIDAx64 at default settings with 4.2 turbo.

[MrRockliffe]

Please remember these are synthetic benchmarks and your cpu wouldn't normally encounter such a high load

[Sylvester]

A comforting thought, nevertheless the first CPU evidently did have a cooling problem as the replacement CPU is running significantly cooler. I replicated SCANS test results with the Noctua NH-L12. Max 81°C which is one degree cooler than SCANs 82°C with the Zalman CNPS9700. This is at 4x 4.4.



And for comparison with the tests I did on page one of the thread, this is OCCT, not underclocked, running at 4x 4.4 turbo, previously it would shut down at 85°C after a couple of minutes even at 4x 4.2 and I had to underclock it to 3600 w/o turbo to get it to complete. Its a different story with this CPU.