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.

[SPark48]

Love your analogies, lol.

Have you contacted Intel about these problems you are having? I honestly think there may be something wrong with your 4790 and you may need a replacement, especially since you are failing 2 of 3 PDT runs. There should be no reason for a 4790 to go above 80C, even with maximum load. Hopefully the processor isn't that old and you can get a replacement at the store you bought it from. Otherwise, getting a replacement from Intel will cost you $25, which is just ridiculous... Using your analogy, if my BMW Z4 has an engine problem within the first few days of purchasing the car, why do I have to pay money to get a replacement Z4? BMW replaces it for free. Makes no sense, but apparently that's how Intel does their RMA as I had to pay $25 to get a new 4790K shipped to me.

[LCCF]

okay - came home a bit early, and first i un-installed the rest of the asus utilities (EZ Update, USB 3.0 Boost, and System information) - had zero effect on temps

i then reset CMOS as well as RTC, went into BIOS and selected default settings, and dis-abled XMP and turbo boost - rebooted and ran XTU

idle temps were same 42-44C at avg 2% usage, and running the XTU stress test, saw 72-74C, 100% usage and .....drum roll.....3.80GHz cpu freq where i'd been at 2.80 max with current limit set to 56A to keep temps down,

went back into bios and enabled XMP - also looked for cpu power controls as i noticed turbo power boost was at 4094W and same for Short duration power boost, but for whatever reason, everything is on "auto" with no other selection possible that might load the settings screen or options.

rebooted, ran XTU and with XMP enabled, idle temps were at 48-51C, but under stress test they "only" went up to 74-75C for most of the stress test with some spikes to 79-81C. Much much better than i had seen to date, and the cpu frequency is now at 4.0GHz, which is the spec'd boost freq for my 4790 (locked) cpu. I went into XTU manual controls to adjust the watt limits and current limit, (84W, 105W & 257A) but saw no change in XTU stress test

I then ran the benchmarking in XTU and got the highest score yet, 976, which was 2 positions down from the highest scores in the comparison graph online for my cpu.

considering my cpu resides in a SFF case and i'm still using the stock intel cooler, i'm happy - i believe with an improved cooler, those temps will come down some.

did a short video file for 5 minutes, and highest temp i saw was a single spike to 78, with 99% of the video transcoding at 72-74.

I don't know whether the asus utilities were "glitching" the intel PDT program before, but i ran it 5 times this afternoon, and it passed all five.times. FIngers crossed, i'll try this again later tonite and probably do it again for a few days - i'll definitely keep Real Temp running as a temp monitor

thanks to ken_intel, sampark1980 and everyone that contributed to this thread - for anyone else reviewing this thread, the changes that i noticed brought improvement in temps were the following:

1) un-installing ASUS performance utility - dropped temp at idle by approx 6C and at 100% usage 3-5C - don't know if it was interfering with the bios as it has some ability to change bios settings

2) clearing CMOS & RTC and resetting to default settings.- brought the biggest improvement, brought my temps down to mid 70C while allowing cpu freq to climb to 4.0GHz with no limiting (temp, power or current) and allowing me to run XMP - XMP did make my temps jump up about 6C at idle, and 3-5C in 100% usage

again, thanks to all

[dadi]

Just get another cooler lol, almost anything above 40 bucks will do better than the stock one.

What speed is your memory now, using the xmp ? In xtu benchmark using xmp (from 1333 to 1866) gives me a big boost in the score, about 100 points; however, My best score is 930 or 932.

[LCCF]

i've already ordered the noctua NH-L9Ihttps://www.amazon.com/gp/product/B009VCAJ7W/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1 - I couldn't decide between a corsair H60 or an air cooled when i noticed one reviewer indicating it out performed his H60 in the indentical computer (he'd built two), so we'll see if the internet is playing nice this week. But it has consideraby more heat exchanger area (or fins) and a larger fan, so it should outperform this intel cooler. The H60 had the advantage of a greater clamping force on the cpu, and i remembered what some had said about their temps going down from the heatsink clamping down and compressing the cpu some - the theory being the glue compound inside compressing over time - hell i'd even researched the "delidding" process. There really wasn't anything off the table on this issue.

I still don't understand why resetting the CMOS didn't correct the issue last night, but i'm not going to worry about it too much - but i think you're on the money re the asus software interfering with the BIOS, and the Bios may have it's own bugs - i'm still remembering that 8X multifplier factor in the power settings.

remember, i'm running a 4790 (locked) not the K, but i'm hitting 4.0GHz with boost, steady with absolute no limiting, and with xmp enabled, as i said above "...74-75C for most of the stress test with some spikes to 79-81C." Converting a video file, temps are pretty much in the low 70s with a couple of small spikes up to 79/81 but those only last a couple of seconds, so i'll leave XMP enabled till the cooler gets here. I'm hoping to see a bigger improvement on the idle temps, those seem abnormally high, especially for 1-2% usage, but for now i'll live with it, and see if ASUS improves on the next Firmware update.

I started this thread with absolutely zero knowledge of most of the power settings in bios, and learned more than i really wanted to along the way - i know i fried more than a few brain cells learning this stuff - hopefully the computer won't have any other issues so i get a chance to regenerate those brain cells - at 65, they don't regenerate as fast as they used to.

Along the way i did some research into the i7-5820 & 5830K cpus (six cores) in case the issue didn't get resolved, and holy moly, the problems with the 4790k are child's play compared to the new Haswell-Es. Same with the X99 mother boards. Some of those owners already have their pitchforks in hand and kettles of tar boiling. Intel is definitely releasing product prematurely, at least imho.

On the benchmark testing, with xmp on, i hit 976 - before, running with current limit up to 256A and temps of 99-100 i hit 836, but with it limited to 56A i was only hitting 538 or thereabouts and 2.80GHz. But limiting current to 56A kept me at tolerable temps.

[dadi]

And again, what speed does your ram have with xmp enabled ?

Just curious, you have a good score in the xtu benchmark and the ram seems to play an important role in that.

[LCCF]

ram is 1600, but if i dis-able xmp, it drops to 1333, ram is Gskill Ripjaws, and the part number was one on ASUS's qualified vendor list.- i remembered asus had said to be sure the ram was XMP rated and this particular ram is.

but that brings up another bug, i assume in the bios - dis-abling xmp didn't always drop the ram freq, if turbo was left enabled, it would still be 1600 some of the time, other times it would drop to 1333 regardless of turbo boost status. I noticed it was varying unpredictably on that freq change before the CMOS clearing.

here's a shot of a test i just ran - it took it 2.5 minutes into the test for temps to climb above 75C. Another thing i've noticed - before, when it was over heating, TDP never went above 58W, even with the limit up at 256A. Now with temps down to a reasonable range, TDP is up to 68-69W - so earlier I had higher temps from less heat, and now I've got lower temps from more heat. It could just be an illusion - i have no idea.

[MM12]

My specs:

i7-4790k

Asus z97 Gryphon, bios v 2101 (12/14 latest)

H60 Cooler

HX650 PSU

It seems to me that this problem is in fact due to the CPU.

I'm on my 2nd CPU after RMA'ing the 1st. I just spent some time on the Intel chat with a representative who told me that this problem was definitely not related to the CPU. Infuriating, the problem persists even on a 2nd completely different computer.

The issue is common enough that it really should escalate to being a full recall.

What appears to be happening is Intel cannot produce enough of the i7-4790k to the labeled specifications.

The i7-4790k is, more or less an i5-4690k with hyperthreading/larger cache.

If you underclock the i7-4790k to run at the same speeds as the i5, you'll achieve passable temperatures.

I'd wager that Intel is unable to produce enough 4790k's at their marketed specifications, and is instead selling "i5-4690k"'s with hyperthreading, larger cache and overclocked to try to meet i7-4790k specs.So some chips will perform properly, where it seems many others wont, and Intel is counting on most people not noticing. Those who do notice, they RMA. (The RMA process went very smoothly actually)Unfortunately I'm on my 2nd bad chip, and I'm hesitant to go for a 3rd.

I've done almost everything suggested. Updated bios, cleared CMOS. XTU is reporting normal voltages/wattages. Set BIOS CPU voltage lower until it wouldn't boot. Didn't help. I'm currently underclocked to 39x at 1.0v that seems to be preventing overheating. A glorified i5-4690k.

I don't think there is a solution other than replacing the chip.

I would absolutely love to be proven wrong.

[SPark48]

I feel ya man... It's not just this thread, or even this forum. There are hundreds upon hundreds of these forums with thousands upon thousands of people having the same problem with these processors. You said it exactly correct that Intel is "counting on most people not noticing". I went to Micro Center the other day and there was someone buying a 4790K so I asked him if he knew about the overheating issue and he said he already bought 2 of the same processors before and he has not had a problem. I asked what he uses to stress his CPU and monitor his temperatures, and he had no idea what I was even talking about. I told him about HWMonitor and I bet you anything that dude has either joined a similar thread and/or started an RMA process. It's crazy/scary/stupid/(enter your favorite adjective here) how many people buy this processor just looking at the price or mesmerized by the "i7" and don't have a clue how to check if it is running correctly. This is why Intel is getting away with this. What got me even more upset is that a few months ago Intel settled a class action suit for publishing embellished benchmarks for a Pentium 4 processor. I was part of this class action and I got a check for a whopping $15 after 12 years!!! I'm afraid something similar is going to happen where we're going to get a check for $20 in 2030 for the overheating 4790Ks... In this case, I'm one of the lucky few to get a decent performing processor (after RMA of the first one), although Prime95 still gives me 100C+.

Speaking of RMA, when you say yours "went very smoothly", did you have to pay $25 for shipping? When I did RMA on my first chip, Intel charged me $25 as "shipping" to send me the new chip and send them the old one. I was desperate for the new processor so I went ahead and paid it without much argument, but I keep thinking, why the heck did I pay $25 for a chip that wasn't working right straight out of the box? I'm just curious if other people that had to RMA their chip had to pay for it as well.

[MM12]

The first time I was told to just mail it in, with no charge. I just printed out their instructions, wrote the order # on the box, and sent it. It took about 7 days from sending it in to receiving the CPU. The 2nd time the rep said I could either do that same thing, or pay $25 for an overnighted CPU, along with a hold for the amount for the CPU on a credit card until the defective unit was received.

Both seem reasonable enough.I fortunately have a G3258 that I put back in when I sent it back so I'm not without a computer. I love that little processor. Mine is an OK overclocker but it's been fun to work with. This i7 has been a nightmare.

They said if after the 3rd time, if I can prove it is the CPU at fault, they will consider a refund. I may go that route. I'll have 3 separate computers to prove my case on. I only bought the i7 because it was supposed to be the top of the line. I'll "settle" with the i5 and keep $100 instead of waiting 12 years for $15

[SPark48]

When you did it the RMA for the first time, did they send you a free shipping label, or did you have to pay for the shipping to send it back? You know, if you buy anything from Amazon, Newegg, Google, Apple, (enter any major "tech" company name here) and if their product is not working correctly, they will send you an email with the return shipping label for FREE so you can send it back to them at no cost and then send you out a new one at no extra charge... Even if you do the "quick"/overnight method, in the case of Google and Apple, they will hold the value of the item on your credit card, send out the item to you first, then when you get it, it already has a return shipping label inside (FREE), so you just put the broken one inside, stick their FREE shipping label on it, ship it back, then they release 100% of the hold on the credit card. In my experience, Intel has been the only (major) technology company that charges their customers for RMA on a chip that is not functioning correctly. It really is absurd if you think about it. Why are we paying $25 extra ON TOP OF the $300+ we already paid for the processor to get it replaced if it isn't working right? The onus should not be on the customer if the processor itself is at fault. Don't you think? It's not the $25 that's important here and I'm not trying to be petty. It's just the principle behind it. We're losing $25 for the RMA; what is Intel losing for doing it? Not sure what happens to the "broken" chips they get back, but for all we know they run some tests on it, refurbish it, and sell it again. I've always been a big fan of Intel and it actually says a lot that in the 15+ years of buying their products, this is the first time I had to do an RMA. But this whole situation with these 4790K's has been a disaster for me from beginning to end. The most frustrating thing is, with AMD's drop in quality chips, Intel pretty much has a monopoly on the market. Until AMD or someone else comes out with some decent processors to contend with Intel, we're forced to bend to their demands.

[MM12]

No label, I just went to UPS and shipped it. $15

[CPapa1]

I am also very frustrated by all of this. This is my first Intel CPU after many years of being a die-hard AMD fan (since the early socket A days!). Based on all the hype, I decided to jump in the deep end with the i7-4790k.

This was the most expensive CPU I ever purchased, and it's been the most frustrating system upgrade I can remember. This CPU ran scary-high temperatures with the factory HSF. I had to spend money on aftermarket cooling just to get back to a range where I felt safe. I'm irritated that I spent (extra!) big bucks on a premium CPU, only to have to spend even more dollars on cooling just to maintain safe non-overclocked performance. What was the point of all this?

Ken's /docs/DOC-23517 Troubleshooting Intel® Core™ i7-4790K / i5-4690K overheating article is a great resource, but I'm running linux so I can't run it. I have an Asus Z97-A motherboard. Exactly what settings need to be fixed in the BIOS? If it's a motherboard vendor/BIOS problem, why is Intel not working aggressively with the various vendors to resolve the apparent miscommunication in factory configuration?

But, I lean toward a mechanical problem with the CPUs themselves. I've read many posts on other forums where folks delid their CPU, remove the adhesive to bring the integrated heat spreader down just a smidge closer to the die, then reinstall the lid with the same factory thermal interface material (TIM), and the CPU temperatures are significantly reduced. This points to an issue with the TIM not making proper contact - air gaps, perhaps? - and also suggests why we are seeing temperatures peg to 100C in a fraction of a second.

I wish Intel would stand behind its product by first determining the source of the problem (default BIOS misconfiguration? mechanical? both?), then communicating remedial action with the community. I too feel like they're hoping that everyone's CPU will quietly thermally throttle themselves into the appearance of sufficient functionality. This treatment has left a very bitter taste in my mouth.

[dadi]

link to page 29 in this thread, I have photos in that page with the exact settings on a asus motherboard ( maximus vii hero)

edit : just go to page 29, for some reason can't link you to the exact post

[LCCF]

danyxp - it occurred to me, for you to comment what you did about my benchmarks, what kind of benchmarks are you seeing with an i7-4790K?

[dadi]

I did a fresh windows install yesterday, now it shows my speed and temp at bench finish, before it was empty. I find the test not so reliable, I get results from 910-920 with the same settings. my best was 968 with oc to 4.7 Gz; below screens with default settings (intel defaults, not asus) and xmp ram at 1866 Mhz.

[LCCF]

i don't know enough about this stuff to state authoritatively, but your dynamic cpu voltage offset is showing a heckuva negative value - when i dropped my values negative, my temps went down as well as my benchmarks??

is that an intel default value?

and have you tried a straight 0.000000000mV? for the hey of it?

and your core voltage is showing "0.0156xxx" where mine is 1.016 and ranges from 1.012 up to 1.017 under adaptive mode?

[dadi]

I'm using adaptive voltage from bios, with a negative offset off 0.064 (-0.064) and an extra 0.016 added for turbo. Those values are different from cpu to cpu, you have to test to find the lowest stable voltage. I actually can go using all cores on 4.5Ghz with these values, but the temps get 5-10 celsius higher. My voltage during testing is maximum 1.1490v.

I have taken the the values fro a post here : https://rog.asus.com/forum/showthread.php?51142-Z97-Vii-Hero-4790K-Overclock-Help/page2 Z97 Vii Hero + 4790K Overclock Help - Page 2

"

First in bios load optimize. Reboot. See inside ure bios whats the voltage is. This is the vid for ure cpu. Mine is 1.04v.

Ok now things will get interesting.

My 4.4ghz stable is +0.096v of this. So if i run manual i will get 1.04+0.096v=1.136v. The default adaptive auto is 1.184

So if i want to run adaptive i need to set offset to "-" 0.064v with additional turbo 0.016v

4.5ghz is offset on auto or adaptive "-0.016v with additional turbo 0.016v

4.6ghz is adaptive "+" 0.032 with additional turbo 0.016v

4.7ghz is adaptive "+" 0.080 with additional turbo 0.016v

"

If I leave the default voltage from bios, the cpu gets a higher voltage, resulting in higher temps with no benefits.

[LCCF]

have you tried narrowing down the benchmark comparisons in the online xtu comparison graph to just MBs like yours, then looking at their settings?

That benchmark is low - it's too bad the MB mfgrs don't offer a counterpart to the intel PDT utility, to run a check on the MB board's components

[LCCF]

i just ran another test only to compare the memory utilization values - mine's showing 4838 MB this morning, and it's usually up there in the high 4000s, while i notice your's is showing 1367 MB - mine's at idle showing the 4838 score, same as yours.

Is it possibly your memory? - did you select one from ASUS's QVL (qualified vendor list), and even if you did, it'd be interesting to see the results if you had access to some different mem boards to try for a short test.

PS - was researching some on the asus utilities when i noticed this:

5-Way Optimization - Auto Tuning (Mainstream and WS Only)

Auto Tuning is an automated overclocking routine that tunes a system overclock for the end-user. We've had this feature for a few generations now and have been enhancing it year-on-year based upon end-user, media and our internal feedback.

Auto tuning works by running a Prime based stress test on the system. Previously, there was no control over the duration of the stress test. We ran a quick 15 second test and then increased processor clock frequency in steps. For Z97, we've added the ability to set stress test duration up to 1 hour at each step (we can extend this further via an update). Plus we've added a memory stress test to the mix for more stringent overclock testing.

Users can also target frequency or a temperature to tune to. Overall a far more comprehensive set of auto tuning parameters to help make overclocking easy for newcomers and help seasoned users evaluate the overclocking potential of a CPU effortlessly. "

If you've got the asus utilities still on your system, it might be worth a try to see what it shows

[LCCF]

not sure about linux, i'm a computer idiot, but clearing the CMOS, per ken's instructions, seemed to have had the biggest benefit. Don't know if you've got CMOS in linux, would assume you do - if so, do clear it and see what results.

another solution for securing settings, if you do the XTU benchmark test, then when finish hit the "compare online" button

once there, stay anonymous if you don't want to set up an identify, but clik on the "analyze" (center circle) - xtu will have uploaded your system info, including MB, OS etc

on the graph you'll see a number of "dot" representing other users' scores. your's will be identified with an red dot and an arrow -

up top you can select other parameters to change comparisons, ie only i7 cpus, Z97 chipsets, etc - one of them will be your MB model - clik on it and make sure it greys out

- then you'll only see results of users using the identical MB and i7-4790k CPUs -

if you clik on the dots above your's, ie to the right, it'll show you their benchmark and the "compatibility" to your setup - i found a couple showing 100% compatibility - down below the graph will be a download button - you can download the settings in XTU that user had selected - it will download to a file on your desktop, clik on it and it will load into XTU and set those settings.

A couple that were higher than mine did indeed increase my benchmark test, but also increased my temps. But it might give you some starting points to adjust from.

hope that helps

[dadi]

CMOS is not depending on the OS, it's on the motherboard

On some models it's a button on the mb, on others (older models mostly) you have to play with a jumper on the mb or even remove the battery from it.