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.

[kpago]

i have change to z97mx gaming 5 only watt from auto to 88w.al the other setting left to default except xmp .xmp is anabled.so i had at results 4,2ghz turbo and 77c temperature.its a begin.also i have try play 2 hours diablo with watts to auto and turbo on 4,4ghz and the bigest temperature was 82c at first 2 cores.

[REnso1]

OK Gigabyte confirmed they didn't test with both limiters set so I have pointed that out and sent details for replication setup from default so it will be interesting to see if they get the same issue with both limiters set. 110W , 255A.

[SPark48]

Just to update:

I received my RMA chip and ran Prime95 and OCCT and average temperatures running stock (4.0/4.4GHz) was in the mid 50C's with a maximum of 62C. This is down from 90C-100C on my previous chip. With temperatures being in comfortable ranges, I decided it was time to try out the overclocking abilities and I was able to get to 4.896GHz (Core Ratio Limit x48 with 102 BCLK Frequency) with temperatures averaging high 70C's with a max of 84C, although that was maybe 5-6 spikes during a 15 minute test. I ran the test 5 times total and got a BSOD twice (CLOCK_WATCHDOG_TIMEOUT once and WHEA UNCORRECTABLE ERROR once) so it isn't 100% stable (yet), but I am confident that I will be able to adjust voltages a little bit or (if all else fails) bring Core Ratio Limit down to x47 to be completely stable.

To get the above temperatures, I did have to adjust settings to resemble ken-intel's as closely as possible without limiting the speed of the processor. Needless to say, I am much happier with the results with the new chip compared to what I was getting with the previous one.

Here is a picture of one of the tests I ran:

[idata]

Hi all,

I thought I'd add my 2 cents to this thread. I'm not an expert over-clocker, but have tinkered in the past.

I recently bought a i7-4790k and a ASUS Maximus VII Ranger board to go along with a new GTX 970 graphics card. I wasn't planning on over-clocking it yet and just stuck the stock cooler on it.

It was only when checking my graphic card clock speeds that I noticed that the CPU was going up to 95 degrees when running games!?!

I found this thread and using Intel XTU implemented the throttle limits suggest by @ken-intel:

Turbo Boost Power Max: 88W

Turbo Boost Short Power Max: 110W

Processor Current Limit: 105A

Interestingly, when stress testing, it was the processor current limit throttle that was triggered first rather than the power settings. In fact, the power throttle didn't have an impact until it was reduced to <60W.

I'm not sure I'd say this was *fixed*, but I'm now getting temps of 75 – 80 degrees (I wouldn't be happy with anything over 80) while getting approximately 4.2 Ghz on 30 minute stress test.

I'll probably get an after-market cooler in a few weeks and I'll be hoping to lose another 10 degrees of the temps, but in the meantime this will do.

Thanks to everyone who's contributed on this thread

Small question: I thought this chip was designed to work at 4.4 Ghz at boost for 8 seconds and then 4 Ghz after (a little surprised to be getting 4.2 GHZ after 30 mins)?

[kpago]

Guys I am saying again. I try everything until take a good thermopaste. And I mean in the best 3 at world. I saw 1 review don't remember where. It's not the money for this paste it's tseaper that others but is at top 3.use it and you see.not from 1st day because need work for 2 3 days this paste. Prolimatech pk-3 nano 1.5 gr.you will see fantastic results. From 100 c I have at Intel extreme utility hard test and ban which 77c now.i use after this again one paste from stock and again I had 100c.

[BSant8]

Hey all!

I keep seeing people saying that they were hoping that the CPU will run at 4.4Ghz with 100% load. It's not entirely correct. It's more of a marketing strategy. 4790K has a stock multiplier of 44x / 44x / 43x / 42x and a base clock of 100Mhz. That means that it will run at 4.4Ghz only when one or two cores are needed. When 3 cores are active, the boost frequency drops at 4.3Ghz and when all cores are at full load, the turbo speed will only reach 4.2Ghz. Some motherboards (e.g. Asus Z97 Maximus VII Hero) come with the default multipliers set to 44x / 44x / 44x / 44x. That's why some people see 4.4Ghz all the time. But that comes at the cost of higher voltage and higher temps. It still doesn't explain 80+ temps but it does make the CPU run some degrees hotter.

[SElwa]

Woolphy

You said "I keep seeing people saying that they were hoping that the CPU will run at 4.4Ghz with 100% load. It's not entirely correct."

I am actually running my CPU now at 4.4Ghz with 100% load with good temp using the stock cooler (Note: that is my replaced new CPU, the old one was faulty), I reached that result by tweaking Ken-Intel suggested settings in UTX, here are my new tweaked settings:

• Vcore: 1.150v
• Processor Current Limit: 161A (Note: if I set it to 105A as suggested I get 4.2Ghz at %100 load instead of 4.4Ghz)
• Set all multipliers to 44x / 44x / 44x / 44x

Using Ken-Intel suggested settings and with the above tweaking I get:

79c max using a UTX stress test

73c (steady, no more no less) using AIDA64 stress test (CPU stress test)

79c max using IntelThermalTest v0.2 (c) 2014 by Karwos

79c max using OCCT

All using the stock cooler at 100% at 4.4Ghz (all cores not just the first two)

Note: Any FPU stress test or any stress test that uses AVX instructions could/may increase the temp dramatically even with the use of aftermarket coolers (as it happened to some users) and I have noticed (by trying) that when that happens the frequency goes from 4.4Ghz to 4.2Ghz/4.0Ghz and the cause is not the thermal throttling only, but the power/current throttling too, maybe those AVX instructions draw too much power than the normal instructions but I don't know for sure that requires further testing.

[idata]

My replacement 4790K runs cooler than the previous one. Using the recommendations above, and under stress I now get a max temp of 75-79C after 5 minutes, rather than the old 99C after 1 minute. So things are much better.

One thing that I noticed, is that my old processor core # 3 ran 10 degrees hotter than the others under stress. The replacement, cores 1-3 run about equally, but core 4 runs 5-6C cooler than the others under stress, and when it cools down at idle, it runs about 2C hotter than the others. Seems strange to me.

What's the reason for temp differences between cores?

I'm happy that no thermal throttling, turbo throttling, or core amperage throttling is happening to me anymore. Stock cooler.

[jband2]

ShadiElwan Wich is the lote number of the new processor?? when you using ken-intel suggested setting with the new processor, wich is the new temps??? 73-79??? using 105A multipliers 44x 44x 43 x42x??

Thanks!

[jband2]

be a problem of internal dissipation , internal compound ?

https://www.youtube.com/watch?v=XXs0I5kuoX4 Fixing Haswell and Ivy Bridge CPU temps: IHS removal - YouTube

[SElwa]

bahianus

I am writing down almost everything written on the CPU Box maybe it could help:

--------------------------------------------------

S-spec: SR219

Product Code: BX80646I74790K

S/N: 3E433036A2639

Batch# : L434C341

MM# : 937428

EAN: 5032037065948

UPC: 735858285957

Made in Malaysia

--------------------------------------------------

You said: "when you are using ken-Intel suggested setting with the new processor, which is the new temps??? 73-79??? Using 105A multipliers 44x 44x 43 x42x??"

• Using ken-Intel suggested settings with the new CPU the temps dependent on the type of the stress test ranges from 73c to 79c (the old CPU no matter what I do was always always reaching 100c under any kind of stress test)
• Using 105A multipliers 44x 44x 43 x42x: It's between 71c to 78c with max short peak of 79c then returns to 78c

Very Important Note: believe it or not Clearing the CMOS (reset the bios and load default optimized settings) helped a lot. I remember changing something very trivial (can't remember what it was) in the bios settings then the temp reaches 95c using ken-Intel suggested settings again!!!!! So I cleared the CMOS (reset the bios and load default optimized settings) and tried the ken-Intel suggested settings again and the temp was normal again, I don't know why, so my advice is: update the bios first, then reset the bios settings (by clearing the CMOS) and load default optimized settings then go to UTX and enter ken-Intel suggested settings and if you can't find the "Vcore" in UTX before changing it through bios try first any third party-apps if you have Gigabyte try "Easy Tune" app.

Note 2: Enabling "Fast boot/Ultra fast boot" through bios or/and enabling "fast startup" in win 8 results in: "everytime you restart your PC the UTX tuning settings reset to its default values, so be aware".

Finally, here are my current settings:

• Reference clock: 100.1739 MHz
• Turbo Boost Power Max: 88.000w
• Turbo Boost Short Power Max: 110.000w
• Turbo Boost Power Time Window: 8.00000000 Seconds
• Processor Current Limit: 161.000A
• Multipliers for all Active 4 Cores: 44x
• Vcore: 1.150v

[DRing2]

I'm new to this and to be honest I fried my last CPU when I messed with overclocking. I could use some advise on this. My motherboard is a Gigabyte Z97X-SLI-CF and like others my settings were way off. These are the default settings running a stress test. I think the only thing stopping it from frying is the multiplier limiter. My temps stay around 70 with spikes to 75.

When I follow ken-intel's settings I stay around 4.06 Ghz. I got my speed up by using these settings, but I'm looking for advice on how to get to 4.4Ghz without burning up my CPU. Also, my Vcore is set to auto. Thanks.

[REnso1]

ken-Intel As per my previous, I contacted Gigabyte and gave them replication instructions to produce throttling to ~ 3.3GHz when the watts limit is in the order of 110W and the Amp limit is set 255A or below (to be precise any value less than but not equal to 256W). This issue has been replicated by another GA Z97MX Gaming5 user in this thread.

There was some misunderstanding at first but it looks like Gigabyte also replicated the issue as they took longer than usual to reply and asked me the following-

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

I am still having to underclock my CPU to 4x41 turbo with a Noctua NH-L12 on this mobo.

I feel a little underwhelmed by this answer so I wanted to check this with Intel and other I7 4790K owners before getting back to them.

• Is it possible to use both limiters together on other mobo models and brands?
• Do Intel have a sound justification for recommending using both limiters which I can pass on to Gigabyte?
• Does it help to use both limiters at the same time on other mobos as I cannot do it with their mobo so I dont know if it works better that way?
• Does the inability to use both limiters indicate the Gigabyte BIOS is not running to spec re power draw?

Any advice or insights welcome.

[KenF_Intel]

Hi BoolyBooly,

Thanks for your continued work on this.

I'm a bit surprised that Gigabyte is funneling these BIOS questions through you. Intel and Gigabyte have a strong relationship and I'm sure that someone at Gigabyte has a direct connection to someone at Intel that can answer these questions better than I can. I'm trying to find someone within Intel who works with Gigabyte to help us with this issue.

Here's what I can tell you on your questions:

• Is it possible to use both limiters together on other mobo models and brands?
  • Yes. I've used these settings on multiple brands/models. It's even possible to use these 2 limiters on other Gigabyte boards. I've used the settings I posted earlier on a Gigabyte Z97 Gaming 7 board. I did not have to change the Current Limit to 256 or higher to run at 4.2GHz with the XTU stress test.
• Do Intel have a sound justification for recommending using both limiters which I can pass on to Gigabyte?
  • I've asked some of our really advanced engineers for advice on this, but it seems obvious. One limits the total CPU power (Core and Uncore) and the other limits just the Core wattage while TurboBoost is active.
• Does it help to use both limiters at the same time on other mobos as I cannot do it with their mobo so I dont know if it works better that way?
  • Yes. Leaving the TurboBoost Short Power Max at the (very high) default settings results in quick overheating during Prime95 v28.5 Small FFT testing.
• Does the inability to use both limiters indicate the Gigabyte BIOS is not running to spec re power draw?
  • This is very difficult to answer. It's certainly expected that the 2 limiters function on the same board but there are a lot interactions/dependencies between these settings and with the hardware.

I will come back to this as I get more information.

[REnso1]

Thanks for your answers ken-Intel, I have been working on a reply to Gigabyte in the meantime but to get my facts straight I have been stability testing using tests of 3 hours or more in AIDAx64 beta and encountered hardware errors with my settings which are now fixed but testing is taking a while and I have had to rethink my undervolting which previously was keeping the temperatures down. I am currently having to restrict turbo to 41x to stay under 74.04°C with a Noctua NH-L12 heatsink with RAM set to XMP (QVL DDR3-2400 1.65v). At default turbo and auto settings and RAM at SPD 1333 1.5v, temps reach 78°C.

A further point of information, my retailer SCAN have already RMAd and tested this mobo and replaced it twice for causing overheating in their labs. So I will be asking Gigabyte if they can adjust their BIOS to use both limiters and follow spec more exactly.

[jband2]

Ducky89 when you run prime95 for 15 minutes the power up to 110.89w. The maximun power is 88watts for this processor. Set this in your bios and run prime 95 again.

[YKar]

Would it affect the clock speed if reduce the wattage?

[YKar]

Hi guys, I searched around and notice many suffers the same fate as me...I follow ken's settings and only manage to got 4.0 Ghz for CPU Frequency...in addition that the temperature was pretty bad which was about 85 Celcius when doing the stress testing = / Help needed...do I need to change the stock cooler to something like Cooler Master Hyper 212x?

[KenF_Intel]

Hi Ducky89,

Are you seeing the 4.0GHz speed when using the XTU stress test or is it coming from something else (Prime95)?

Deciding between the stock cooler and a higher end cooler is a bit like choosing a truck. If you are doing light to medium hauling and towing then a truck with a standard engine cooling system should work fine. But if you regularly tow a 5,000Kg trailer over a bunch of mountains then you may need a more robust engine cooling system.

What I'm suggesting is that you might want to monitor your temperatures and frequencies when you are running the software (games, video editing, productivity applications) that you use on a regular basis. If you experience throttling under those conditions then you may need to consider a higher end thermal solution.

I hope this helps.

[YKar]

Hi Ken, the 4.0GHz comes from the XTU stress test when I followed the settings that you posted in previous page. My temperatures idles around 40-60 Celcius in which I believe it's ridiculous = ( When I game, it could reach 90+ Celcius!! I previously had good experience with Intel CPU so I do not know what went wrong with this 1.

This is taken when I just boot up my computer for 11 minutes while idling. As you can see it reaches 62 Celcius maximum on 2nd core. My other hardware is fine though (Notice that GPU temperature is about 47 Celcius)

If this issue is resolved by changing the third party CPU cooler, I am more willing to do that. What I am afraid that even if I changed that, the issue persists = ( What should I do, Ken?

[KenF_Intel]

Hi Ducky89,

My lab system idles in the low 30's so your temps are a bit high. A spike to 62 is not great, but might be understandable if there was a background program that was loading and caused the spike and then finished whatever it was doing.

Here are a few things that customers have said helped with the temps:

1. Make sure the system is not using an XMP profile for the RAM. You might need to check this in the BIOS.

2. Make sure you have the latest BIOS for your system. We hope the board manufacturers will release BIOS updates with better default settings.

3. Clear the CMOS and then reapply the suggested settings.

4. Make sure the cooler is tight and the thermal material has spread evenly onto the top of the CPU.

5. Make sure the airflow through the chassis isn't blocked and that the chassis fans are working.

The settings provided are a guideline and might need to be tweaked for your system. What model of motherboard do you have? Can you post a picture of XTU during the stress test showing the monitoring section at the bottom? Especially the CPU Total TDP monitor and the IACore TDP monitor.

If all that fails to help then you have 2 options. You can RMA the CPU or go for a higher end cooling system.

Again, hope this helps.