i7 4790k very high temps.

pipboy 3000

Distinguished
Apr 9, 2011
34
0
18,530
Hi, hoping for a bit of advice.

I recently purchased an i7 4790k second hand from ebay to go in my Gigabyte H97 motherboard.
Before anybody says, yes I know the H boards are not usually used with K chips.

After installing it and attaching a H55 cooler to it I ran a stress test and noticed that the CPU was instantly jumping up to 100C. (YIKES). I've reapplied thermal paste and also checked that the cooler is properly attached, which it is.

As an extra test to see what could be causing the issue I installed the CPU to my other computer which has a Gigabyte z97 motherboard and H75 cooler and temps are just as high.

Both boards have up to date bios'. CPU idles at around 30C. Cooler fans at full speed don't make any difference. VCore is at 1.188 according to bios and multiple hardware monitoring software.

Do I have a faulty chip? Anything else that could be causing the issue?

Thanks.
 
Solution
pipboy 3000,

Whoa there ... hold off on pushing the button on replacing your CPU. I seriously doubt that you have a "faulty" chip that is "used up" as elbert has suggested, so I wouldn't condemn it just yet. Keep in mind that the i7 4790K was launched the 2nd Quarter of 2014, so the oldest CPU's are just now approaching 30 months.

Is your Vcore in "Auto" or "Adaptive" and not in "Manual" or "Offset"? Some test utilities will cause Vcore to spike on the 4790K, which in turn will cause Core temperatures to spike to 100C, which is "Throttle" temperature. Remember that the 4790K is a factory overclocked 4770K by 500 MHz, and runs at higher Core voltages. Have you monitored Vcore during stress testing?

If it reaches 1.25...

pipboy 3000

Distinguished
Apr 9, 2011
34
0
18,530
Thanks, I've retried it with the stock intel cooler and the problem persists. Temps instantly increase to 100C in the first few seconds whilst prime 95 small FFT is running. Gotta be a faulty chip. I've arranged to return the CPU to the seller.
 

CompuTronix

Intel Master
Moderator
pipboy 3000,

Whoa there ... hold off on pushing the button on replacing your CPU. I seriously doubt that you have a "faulty" chip that is "used up" as elbert has suggested, so I wouldn't condemn it just yet. Keep in mind that the i7 4790K was launched the 2nd Quarter of 2014, so the oldest CPU's are just now approaching 30 months.

Is your Vcore in "Auto" or "Adaptive" and not in "Manual" or "Offset"? Some test utilities will cause Vcore to spike on the 4790K, which in turn will cause Core temperatures to spike to 100C, which is "Throttle" temperature. Remember that the 4790K is a factory overclocked 4770K by 500 MHz, and runs at higher Core voltages. Have you monitored Vcore during stress testing?

If it reaches 1.25 or higher toward 1.3, your H55 is beyond its cooling capability. The H55 is low-end liquid cooling which overlaps with mid-range air cooling, and an H75 isn't much better because they don't have enough cooling surface area. High-end air rivals an H100.

Another unasked but very important question; what is your ambient temperature?

You mentioned Prime95, but you didn't say which VERSION ... keep reading ...

Guys,

In case you didn't get the Memo, do NOT run any versions of Prime95 later than 26.6. Here's why:

Core i 2nd through 6th Generation CPU's have AVX (Advanced Vector Extension) instruction sets. Recent versions of Prime95, such as 28.9, run AVX code on the Floating Point Unit (FPU) math coprocessor, which produces unrealistically high temperatures. The FPU test in the utility AIDA64 shows similar results.

Prime95 v26.6 produces temperatures on 3rd through 6th Generation processors more consistent with 2nd Generation, which also have AVX instructions, but do not suffer from thermal extremes due to having a soldered Integrated Heat Spreader and a significantly larger Die.

Please download Prime95 version 26.6 - http://windows-downloads-center.blogspot.com/2011/04/prime95-266.html

Run only Small FFT’s for 10 minutes.

Use Core Temp to measure your temperatures - http://www.alcpu.com/CoreTemp

Your Core temperatures will test 10 to 20C lower with v26.6 than with v28.9.

Please read this Sticky: Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

CT :sol:
 
Solution

pipboy 3000

Distinguished
Apr 9, 2011
34
0
18,530


Thanks, I'll try this tomorrow and report back. Is there any other stress test I can use just to be sure that the chip is not overheating. Also, I previously had the 4690k and I'm sure I used to run the latest version of prime95 stress test on that too, however it never reached anywhere near 100c. Why the discrepancy between the two haswell chips?
 

Yup prime95 VERSIONS LATER THAN 26.6 is an issue. Wish you would have posted that earlier.
 

CompuTronix

Intel Master
Moderator


Because the 4690K runs at 3.9 GHz, has 6 MB Cache without Hyperthreading (i5), while the 4790K runs 500 MHz faster at 4.4 GHz and has 8 MB Cache with Hyperthreading (i7), which requires a significant increase in Vcore, and in turn adversely affects Core temperatures.
 

pipboy 3000

Distinguished
Apr 9, 2011
34
0
18,530
Ah ok, and sorry to answer there other questions you asked earlier in the thread, which I missed.

1. Ambient room temp is 16c, it's getting cold here.

2. I haven't checked whether V core is adaptive but I did keep an eye on it during the few seconds of prime 95 stress testing and it didn't go over 1.188.

3. I'm not sure which version of prime 95 I downloaded, I'll check tomorrow and report back.

Thanks for your help.
 

CompuTronix

Intel Master
Moderator


Guys,

Test utilities such as LinX, LinPack, Intel Burn Test (IBT is NOT an official Intel utility) and versions of Prime95 later than 26.6 load your CPU beyond 120% TDP (Thermal Design Power) at default BIOS settings. Intel tests their processors at 100% TDP with the intended default settings specified for motherboard manufacturers. So when the end user is running thermal and/or stability tests, the goal is to replicate Intel's test parameters as closely as possible by using compatible test methods.

This means that when your 88 Watt TDP i7 4790K is installed with the stock cooler on a motherboard with default settings, then Prime95 v26.6 Small FFT's is run, a utility such as Core Temp or HWMonitor should show that Power (Watts) is within +3% of TDP, which is approximately 91 Watts. Of course, the values change when we change BIOS settings and install after-market coolers to overclock our processors, but this shows that our test methods are valid and appropriate.

Prime95 version 26.6 Small FFT's meets Intel's TDP specification more closely than any other utility. Also, Small FFT's is a steady-state workload, which is perfect for evaluating thermal performance and Core temperatures. Just take a look at Section 13 in the Intel Temperature Guide, which I linked above in a previous post. You can see that the Charts show what a steady-state workload looks like compared to a fluctuating workload.

Thermal testing with Prime95 Version 26.6 Small FFT's will give you a valid thermal baseline.

Asus RealBench is excellent for stability testing because it stresses CPU, memory and GPU. Although it runs a fluctuating workload, it also runs your CPU within 3% TDP during peak workloads, but it uses a realistic AVX workload with code from HandBrake, yet provides peak Core temperatures consistent with Prime95 v26.6 Small FFT's.

Asus RealBench - http://rog.asus.com/articles/news/realbench-v2-43-new-version-available-now/ - (Click on "download" shown in bold print).

The problem with this topic is that there's so much confusion concerning Intel's specifications, and thus so much misinformation on the Internet, and so many people using a multitude of haphazard approaches to testing without any solid basis or understanding, that it's nearly impossible to compare apples to apples. This is why I wrote the Intel Temperature Guide. Give it a read.

CT :sol:
 
What do you use the CPU for? I'd test it with whatever tasks you use it for and see if the temps are normal under that load. I switched from 4690k to 4790k and I noticed the i7 uses more power than the i5. I had to get a better cooler just to use the i7 with turbo boost enabled.


While playing Battlefield 1 with:

Arctic Freezer Pro 7 Rev 2 - ~80C at 4.4GHz

upgraded to

Cryorig H7 - 60C and below at 4.4GHz

(I'm pretty sure the low quality Acrtic Freezer wasn't making good contact though.)

**Also, since you bought it used you could try running Cinebench to see how it scores.
 

pipboy 3000

Distinguished
Apr 9, 2011
34
0
18,530
Hi,

So I've carried out further tests using prime95 26.6 this time and these are my findings:

Notes:
(CPUID used to monitor temps)
(Open air test bench set up)
(Vcore didn't exceed 1.188 on any of the stress tests below)

Using the stock intel cooler first

Ambient temp: ~19C
Idle: 27C

*Prime95 version 26.6 10 min Small FFT Max temp*:
Max temp: 100C


I quickly changed to the H55 AIO cooler

Ambient: 21C
Idle: 28C

*Prime95 version 26.6 10 min Small FFT stress test *
Max temp: 60C

*Asus real bench 15 min stress test using up to 8GB RAM
Max temp: 58C



I'm surprised at how ineffective the stock intel cooler is though, initially when I tested prime95 v266 I though I had the same issue as temps instantly increased to 100C but after switching to the corsair AIO cooler they seemed fine. Perhaps the cooler is faulty?

Should I run further stress tests or is the above ok?
The above figures look promising, is it safe to say that the chip is fine and that I screwed up by installing v285 of prime95?

Thanks for all your help.

I'll be reading the temperature guide that was linked above.
 

CompuTronix

Intel Master
Moderator


LeKeiser,

Although OCCT is useful for stability testing, it's not well suited for testing processor thermal performance. There are two tests; CPU OCCT which uses a version of Prime95 later than 26.6, and CPU LinPack. Neither test can provide a valid thermal baseline. I'm also not a big fan of AIDA64 for similar reasons; additionally, there are too many test combinations which many users find confusing.



pipboy 3000,

Proof once again that Prime95 version 26.6 Small FFT's is the proper test for measuring thermal performance, and later versions should not be used. As I suspected, your i7 4790K is fine.

Here's an interesting article from Intel which addresses the Prime95 version issue: Troubleshooting Intel® Core™ i7-4790K / i5-4690K overheating - https://communities.intel.com/docs/DOC-23517

Here's the recommended operating range for Core temperature:

80C Hot (100% Load)
75C Warm
70C Warm (Heavy Load)
60C Norm
50C Norm (Medium Load)
40C Norm
30C Cool (Idle)
25C Cool

Core temperatures in the mid 70's are safe, so just keep it under 80. For Prime95 V26.6 Small FFT's, 60C on an H55 is excellent, which leaves you ample headroom for overclocking.

In our technical world, everything revolves around specifications. So when discussing this topic, it's extremely critical to be very specific. It's not enough to simply say which utility is being used; which version can also be important.

Further, since we get questions from many climatic zones worldwide, it's equally important for Members to include their ambient temperature, which can range from 15C to 35C. Many Members offering answers or troubleshooting advice frequently neglect to ask concerning ambient temperature, and sometimes jump to the wrong conclusions based upon incomplete information. Ambient temperature is a very crucial variable, as all computer temperatures rise and fall with ambient.

Since your ambient temperature is only 1C below Standard which is 22C, If I were you, I would declare victory and have a beer!

CT :sol:
 

pipboy 3000

Distinguished
Apr 9, 2011
34
0
18,530
Thank you very much. Glad I finally got the answer. It would have been a nightmare if I returned the chip only to have the same problem on the replacement I bought. Lesson learned, stress testing isn't as black and white as many online guides make it out to be. Thanks again.