Another i7 4790K Temp Reading Question

Zi1ian · Jan 14, 2015

I'd like to preface this by saying my knowledge of computers is fairly limited. I've put together a computer before by following step by step instructions. For the most part, including the PC I am writing about, I have a friend help build for me.

I recently upgraded my computer's CPU and Motherboard to an Intel i7 4790K and the ASUS Maximus VII Hero. I also changed my CPU fan to the Corsair H100i. My question involves the i7 4790K's temperature readings. Most commentators recommended RMA, but some of those still had the same issues after RMA. I've found a few posts and comments on various forums, this one included, that noted that Intel has claimed that most programs are unable to accurately assess the i7 4790K's true core temperatures. And I'm hoping to see if anyone can either share with me a source or a similar experience to confirm that what I am experiencing (described below) is not an actual, functional issue.

After I upgraded to the CPU, mobo, and fan, I did not notice any functional problems. It ran and continues to run amazingly well. However, the reason for my upgrade was because my 13-month old Antec Kuhler 650 stopped working and fried my old CPU. So I've become a little paranoid about temperature. I installed Dragon Age Inquisition and played for about an hour. Not a single ounce of a problem- completely smooth, flawless, no lag, or any other concerns. However, before I exited, I figured I'd give RealTemp 3.70 a look to see the temperature and, to my surprise, all cores were listed in the 90's and apparently had been sitting there for probably the entire time (based on the timestamp for maximum temp).

Naturally, I started freaking out a bit trying to figure out what the problem was. Across my hours of Google searches, I came across varying stories of recent bad batches of the i7 4790K or other people posting about cheap components. But a small strain of posts described my scenario perfectly (like http://www.tomshardware.com/answers/id-2375114/corsair-h100-4790k-bad-temp-reading.html). Although RealTemp immediately reports drastically higher temperatures when under load and my CPU fan immediately kicks into overdrive, nothing else actually happens.

I ran two tests. First, I opened up my tower while the PC was running. I loaded up Dragon Age with Real Temp up. Immediately the temperature jumps from the high 30's to the 90's and up. However, my well-secured CPU fan is blowing cold air. It was still blowing cold air after 5 minutes, 10 minutes, 20 minutes, and even 30 minutes. All the while, I had no problems with the computer lagging even a fraction of a second. And previously when I was playing for an hour, I also had zero issues.

Next, I downloaded SpeedFan 4.50 and opened both Real Temp 3.70 and SpeedFan 4.50 to compare. At idle, Real Temp reported 39, 37, 38, 37 C temperatures @ Load <1%. SpeedFan reported CPU 37 C with 2012 RPM CPU fan, 670 RPM Aux Fan, and 959 RPM Aux 2 Fan. Sure enough, when I opened Dragon Age, Real Temp's numbers spiked to 95, 96, 98, 90 @ 54% load. However, Speedfan reported CPU 54 C. The CPU Fan, Aux Fan, and Aux2 Fan increased to 2818 RPM, 918 RPM, and 2642 RPM, respectively. The remaining increases for SpeedFan, if any, were negligible. SpeedFan also did note a change in the VcoreRef-Vin4 value from 0.35V to 0.62 V. None of the other Vcore values changed.

A lot of the posts I've seen have quickly dismissed SpeedFan's results over Real Temp's. Unfortunately for me, SpeedFan's results are the only ones that make sense. Any help in confirming or pointing me in the right direction would be greatly appreciated!

acvolta · Jan 15, 2015

Have you tried the software that comes with the board (I have the same one) and I found the ASUS AI Suite to be far more acculturate then and "after market" monitoring software. I recommend downloading the latest version from ASUS website.

Using the AI suite also dismisses the "programs are unable to accurately assess the i7 4790K's true core temperatures" theory because the program is designed to work with the i5-i7 series.

iamlegend · Jan 15, 2015

Install MSI afterburner and monitor the CPU temps when playing.
Verify if the temps reading are the same.

If you don`t know how to use MSI afterburner search it on the web on how to enable the CPU temps monitoring. It is pretty easy.

CompuTronix · Jan 15, 2015

Zi1ian,

What is your ambient temperature?

Are you overclocking?

What is your Vcore?

Are you absolutely certain that your waterblock is secure?

acvolta,

Your experience is not typical.

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

Intel desktop processors have Digital thermal sensors at the heat sources in each Core (Core temperature), plus a single Analog thermal sensor under the Cores for the entire processor (CPU temperature). For example, a Quad Core has five sensors.

Core temperature is 5C higher than CPU temperature due to sensor location. Intel's Thermal Specification on their website is "Tcase", which is CPU temperature, not Core Temperature. For example, Tcase for the popular i5 4690K is 72C. Tcase + 5 makes the corresponding Core temperature 77C. The relationship between Core temperature and CPU temperature is not in the Thermal Specifications; it's only found in a few engineering documents.

Utilities provided by motherboard manufacturers on the driver DVD measure CPU temperature, not Core temperature. The value from the single Analog sensor is converted to Digital (A to D) by the Super I/O (Input / Output) chip on the motherboard, then is calibrated to look-up tables coded into BIOS, which reads CPU temperature. Thermal code can vary greatly between BIOS vendors and version updates, and is often wrong by up to 30C. BIOS or CPU temperature may not be accurate.

Core temperature isn't read through BIOS; it's read directly from the processor, and is factory calibrated by Intel. Core temperature is the standard for measuring processor temperatures, because it's consistently more accurate than CPU temperature. This commonly means using freeware utilities such as Real Temp or Core Temp, which have been proven accurate over several years.

acvolta,

Please read the entire Guide carefully.

Thanks,

CT :sol:

sla70r · Jan 15, 2015

I7 4790k's seem to have been a bad release to be quite honest. I have the same cpu and have had a hell of a time trying to manage temps and keep them under 85-90 C. My friend has the same CPU and his temps hit 100 C under pretty much any game.... we put in a h100i and that helps with idle and lower load temps but any stress test still shoots it up to 100C. I regret leaving AMD to go intel .

The Original Ralph · Jan 15, 2015

Zi1ian :

i was getting ready to post along a similiar line, the temp readings being all over the map with non-oem temp monitoring utilities. I've got an Asus Z97M-Plus MB with an i7-4790 and even with the asus ai-suite i was seeing some variances / jumps in temps that didn't feel right, as well as general lag in reporting temp increases. Tried a few last night, and the first (CPUID HWmonitor) was all over the map and nowhere near what ASUS's utility was reporting - i took a snippet capture of the asus a1-performance utility and CPUID HWmonitor running side by side - it was taken while i was processing some video files, so no "gaming bursts" involved, just a steady load:

i think intel's stmt is probably accurate, as all the aftermarket freeware utilities seem to bounce all over the map in terms of readings - when CPUID recorded 99C, my cpu fan was running at normal (low) speed and case, up by the exhaust port, was barely warmer than ambient temps. My decision was to stay with the asus ai-suite for temp monitoring. Intel's stmt makes more sense than web stmts that there are bad batches of 4790(k) CPUs being released.

have you tried the ASUS A1-suite? - my only complaints with it, a)doesn't log & b) i've seen some variances that caught my attention but nothing like i've seen with the aftermarket utilities

melonhead · Jan 15, 2015

if temps are high, it is either somethign with the way it is setup, or a faulty chip(which can happen with amd as well). heck, perhaps your cooler isnt working properly whether it is a good heat transfer, or just mechanically working right.
i personally have the 4790k as well with the H105 cooler. with the fan speed at 25%, i idle at 34ish degrees at idle, and no more than 60 degrees under a gaming load, with average room temperatures. i dont regret leaving AMD at all.

CompuTronix · Jan 15, 2015

Guys,

The reason why there are so many conflicting reports of different temperatures, is that everyone tests their rigs with different stress tests, clock speeds, Vcore, coolers, ambient temperatures and measuring utilities that read two different types of processor temperatures.

Add to that the fact that almost no one bothers to look up Intel's specs, which aren't written to make any sense to anyone anyway. Then there's the term "load" that gets tossed around like gorilla poo in a cage. Load? Which load? What load? Load is a very subjective term.

Applications, rendering, encoding and gaming are partial workloads with fluctuating temperatures, and aren't suitable for thermal testing or comparing temperatures, but they're great for endless debate. Add a graphics card that recirculates heat into your case, and your perspective is lost among all the variables.

The only way to make sense of this mess is to test your rig using a methodology that reduces the variables to the lowest common denominators. The Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html - explains how to do it in Section 12 ... yet as the saying goes, you can lead a horse to water, but you can't push a rope.

CT :sol:

The Original Ralph · Jan 15, 2015

I'll take a look at that link - one thing, in my post above, those two monitor displays, were monitoring simultaneously, so it was reading the same load at the exact same time, which goes to the OP's point.

It'd also be interesting if someone were to contact the various monitoring software providers and ask them if they had reports of wildly vacillating temperature samplings on the i7-4790(k) CPU and see what they respond

sla70r · Jan 15, 2015

Their are literally 100's of reports about the i7 4790k and its heat. Are we to believe that all of us ( two rigs with i74790k personally) are doing something wrong ? 15 years of PC building experience, but its me and not the chip.....even though countless others are experiencing the EXACT same problem....kind of ridiculous.

melonhead · Jan 15, 2015

sla70r, well....im a 4790k user as well. not experiencing the problem. im sure that there are countless others that are not experiencing it either. in fact, i would say that there are more that are not experiencing it than there are that are experiencing it.
i dont care if you had 15 days of expereince of 15 years of experience. it does not exclude you from making mistakes. could it be your chip? sure? could it be an error, sure? but if it were the chip, there would be more with this issue than there are without. keep in mind that this doesnt mean that there will not be chips without issues, because there could be, and probably is. but it is no different than it would be with AMD, or any other manufacturer, regardless of product they make (motherboards, HDD, cars, trucks, boats, etc.- you get the idea).

The Original Ralph · Jan 15, 2015

sla70r :

sla70r - not saying you're cpu is or isn't bad, but i am responding to the OP;s point of varying tempurature indications from different monitoring software - and that sure would seem to tie in with intel's stmt that the 4970 is a difficult chip to read or monitor. After my last post, i ran Prime95 for 30 minutes, and ran both the asus utility and cpiud's - the asus showed 68C at peak, and steady at 68c. CPUID was showing 99C for both the cpu and the cores, and per the intel temp sticky, we should assume +5C for cores over cpu temp. Then the CPIUD monitor starting jumping down to 73C and within seconds back up to 99C - kind of erratic, while the other temps it was reporting, motherboard and my graphics card's temp it was reporting the same as asus.

That could still be a bad cpu, but i have to give the asus utility more credit than freeware i grabbed off the net and it's readings are stable and predictable.

like i said earlier, it'd be interesting to see what any of the monitoring utilities respond to being asked if they're experiencing complaints about erratic readings on their 4790(k) cpu.

If you really think your cpu is bad, iirc, they come with a 3 yr warranty - RMA it and see what the replacement does.

sla70r · Jan 15, 2015

melonhead :

Two CPU's , two different MOBO's, h100i cooler and a stock cooler. I would hope to god more would be with out issue than with issue, but a release should not have so many problems. I would be just as unhappy if it were my car or an AMD piece.

Zi1ian · Jan 15, 2015

Thank you everyone for your comments, feedback, and input. I appreciate all of you taking time to discuss my concerns.

I installed Asus Suite 3 and read the Intel Temperature Guide. I have not run any of the tests described in Section 11 and on. I do not have a device to precisely measure the ambient temperature, but I can easily order one to receive it in a few days. I waited until mid-day to estimate the ambient temperature around my PC. My office space heater is set to maintain at 70 F degrees, while my office's air system constantly blows cold air throughout the day (which I cannot control) directly above my head). I primarily use the space heater to counter the cold air in the winter. I would estimate the ambient temperature being around 22 -23 C. I understand that the 4790K's Tcase is 74.04 C +5, Tjunction 79 C, Tj Max 100. The CPU is not overclocked. Suite shows that it is on Performance Mode; ; turbo app is off; and optimized phase is on. I ran Fan Auto Tuning on Suite.

At idle, Suite reports:
CPU Temp - 31 C
Motherboard - 27 C
VRM - 37 C
SENSOR1 - N/A

CPU Fan - 2356 rpm (according to Fan Tuning, this value changes negligible amounts)
CPU Opt_Fan - 2848 rpm (was 782 rpm before I ran Auto Tuning)
Chassis Fan 2 - 894 rpm (was 1979 rpm before Auto Tuning)
Chassis Fan 3 - 422 rpm (was 586 before Auto Tuning)

CPU Core Voltages: 0.704 V - 0.720 V
DRAM Voltage: 1.507 V

Dragon Age Running @ Initial, Suite Reports:
CPU Temp - 87 C
VRM - 42 C
Mobo - 28 C
SENSOR1 - N/A

CPU Fan - 2356 rpm
CPU_Opt - 2806 rpm
Chassis 2 - 2772 rpm
Chassis 3 - 872 rpm

CPU Core Voltages - 1.232 V - 1.248 V
DRAM Voltage - 1.510 V

Dragon Age Running @10 minutes, Suite reports:
CPU Temp - 69 C (during play, I had seen it report between 66 C and 74C)
VRM - 46 C
Mobo - 31 C
SENSOR1 - N/A

CPU Fan - 2339 rpm
CPU_Opt - 2800 rpm
Chassis 2 - 2657 rpm
Chassis 3 - 821 rpm

CPU Core Voltages - 1.232 V - 1.248 V
DRAM Voltage - 1.504 V

Idle, 5 Minutes after DA was closed, Suite Reports:
CPU Temp - 36 C
VRM - 42 C (this temp decreased approximately 1 C per 75 seconds)
Mobo - 30 C
SENSOR1 - N/A

CPU Fan - 2351 rpm
CPU_Opt - 2842 rpm
Chassis 2 - 890 rpm
Chassis 3 - 456 rpm

CPU Core Voltages - 0.704 V - 0.784 V
DRAM Voltage - 1.507 V

Additionally, I played Dragon Age Inquisition for about 20-25 minutes this morning (5 hours ago). According to Suite, CPU cores initially spiked upwards to 80 C but then fell to ~65 and remained steady for the remaining 20+ minutes of play. The ambient temperature would've been around 19-20 C at the time.

Zi1ian · Jan 15, 2015

@Computronix

So if I understand it correctly, SpeedFan and Suite 3 are reading my CPU Temperature, while RealTemp is reading my core temperature. And since there could be upwards to 30 C difference between those two temperatures, RealTemp is likely accurate in its assessment.

If that's the case, barring some issue with the CPU fan's mounting, my only recourse is just to go through the painfully annoying and more than minimally inconvenient process of returning the CPU for RMA and hope I get a replacement that will function properly?

melonhead · Jan 15, 2015

i havent paid much attention to my voltages when playing, but the voltages during the times that it heats up seems to be a possible issue.
i vaguely remember looking at a thread for someone that did an overclock to 4.8ghz on this processor, that had voltages around 1.2+

id have to do some research on it, but im at work and dont really have that ability all the time. perhaps check into that to see if the voltage could be part of it. your idle temps are good, and have drastically lower voltage, and you dont get hot until it is in use and starts pulling more power.

melonhead · Jan 15, 2015

this might help you compare voltage with temps....
you will have to scroll down a bit to get to the 4790k results, but you can see tha the voltages would be for the higher speeds.
http://www.anandtech.com/show/8227/devils-canyon-review-intel-core-i7-4790k-and-i5-4690k/2

Zi1ian · Jan 15, 2015

@ CPUs Expert

Thank you for the link. I will definitely take a look at it this afternoon. Incidentally, I contacted Intel's support to find out the process and time frame on an RMA. The support staff ended up helping me continue to troubleshoot and answer some additional questions.

When I explained the scenario to the representative, his verbatim response was "Yes, we have been having that situation with this specific processor." He directed me to https://communities.intel.com/docs/DOC-23517 as the initial steps to try to resolve my issue, using the Intel Extreme Tuning Utility. He also instructed me to install and run Intel Processor Diagnostic Tool and provide him the results. The stress test result was a PASS but only barely- 1 degrees C below maximum. He reiterated his recommendation to follow that guide (which I will do shortly). I also shared with him my idle voltage vs. the voltage while playing Dragon Age; and he confirmed that that was normal. He said that it wouldn't be a problem until it exceeded 1.5 volts.

melonhead · Jan 15, 2015

the info that they go off of is going to be a range. technically, the cpu operates within its ranges, and seems to be all that they are seeing when you explain it to him.
i have a feeling that the MB is set to auto and it isnt directing power very well. you could probably lower the voltages manually, and still have the system to be very stable, and obviously run cooler off of the lower voltages.
when i get home, i will check my voltages to tell you what mine runs while gaming and at idle.

Zi1ian · Jan 15, 2015

I followed the instructions from the site Intel support gave me. I installed Intel's XTU tool and ran their diagnostic stress test again. It passed again- this time at 8 degrees below maximum, which is an improvement (although I haven't changed anything). The CPU TDP hovered around 39, never exceeding 40. Voltage was substantially similar to my other trials (0.709 idle, 1.267 stressed).

Next, I input all the manual tuning specs into XTU as instructed by the guide. The Intel Stress Test passed again at 16 degrees C below maximum. The TDP was perpetually at 37 W, with the voltage rainging 1.0820 - 1.1310 V. CPU temp 96-97 C. Thermal Throttling 0%.

The application of the results... when I immediately start up Dragon Age Inquisition, Real Temp displays the core temperatures jump to the mid 70's to low 80's (max was 82), which is a 18-20 degree difference between my earlier results. During actual game play, it settles around the mid 60's.

So this was definitely a significant improvement. From here, I guess I'll have to learn more about whatever it was that I actually modified from the instructions to figure out how to improve it further.

melonhead · Jan 15, 2015

i was doing some searching on the internet and found that a very large majority of those with temps being high are those with asus boards. in fact, quite a few of them being hero board. perhaps there is a, for the lack of a better term, compatibility issue where the voltages arent regulated as well with asus and this i7? almost wish i had an asus board to test it so that i can see for myself if there is that much of a difference with boards, or if it just happens to be coincidence.

during gameplay, mid 60s is pretty average...depending on ambient temps and what not

iamlegend · Jan 15, 2015

If anybody here is reading the Computronix` post (moderator) this thread is probably solved long time ago.

Don`t use Prime 95 in stress testing!
I have low 80`s (stock speed) in using Aida during this winter season.

The Original Ralph · Jan 15, 2015

Zi1ian - that XTU is a nice utility, tks for posting
i ran it as well, and mine passed the stress tests as well, but barely, hitting 99-100C, and throttling went as high as 39%
then out of curiousity i left XTU running while i edited a video - saw my CPU temps in the 55-56C range, ran the asus performance monitor utility and edited the same video again, and it showed the usual 65-67C temps. Running the two utilities simultaneously, the asus utility didn't show any monitoring, i suspect the XTU utility shunts it.

hitting the temps it did on the stress test, i suspect my heatsink needs to be clean and new paste applied - when i first assembled it i had place it on the cpu and then like an idiot lifted it back up and saw the paste had pulled up with it. Got a feeling putting it bad down on the CPU just created some air pockets between the cpu and heatsink.

again, tks for posting that XTU utility

Zi1ian · Jan 15, 2015

I agree that the Intel Temperature Guide is extremely helpful, particularly for a novice like myself, in obtaining a better grasp of Intel temperatures. However, I don't know that it solved my issue. It was absolutely good knowledge to have, but my core temperatures were still in the mid to high 90's at 54% CPU utilization. And it sounds like I was not and am not the only one experiencing that. And unfortunately, my PC already has a pretty decent CPU fan, excellent cable management, and an axial graphics cards, which were the three recommendations to improve temperature.

If anything, the change that improved my temperatures the most in the past 24 hours was the troubleshooting guide I received from Intel Support, which dropped my core temperatures dramatically. I'm just wondering whether I can drop them more with additional fine tuning in XTU or Bios. Simply reducing the voltage does not appear to be reducing the temperatures anymore than they've already dropped.

And consistent with what Melonhead posted, the Anandtech article author also performed his tests with an Asus mobo. Moreover, Intel alleged that his results were below average. So maybe there is some correlation of incompatibility separate from the possibility of a damaged RMA or not properly testing the cores. The curious thing about his article is that he reported the stock/default settings produced 81 C core temp under stress; but with fine tuning, he was able to drop it to 58 C under stress. Unfortunately, I don't have the technical knowledge to figure that out on my own, so I'm just hoping others might. And as a bonus, I haven't been able to find another thread that fully discussed this issue that didn't result in it being dismissed as a faulty CPU. So if someone else with substantially similar problems can also resolve their issue, then even better!

CompuTronix · Jan 16, 2015

Zi1ian :

SpeedFan measures CPU temperature AND Core temperatures.

Real Temp was designed from the ground up for Intel processors.

Running multiple temperature monitoring utilities simultaneously is not recommended, as they can interfere with one another, which can result in inaccuracies.

Another i7 4790K Temp Reading Question

Reputable

Reputable

Admirable

Intel Master

Honorable

Distinguished

Distinguished

Intel Master

Distinguished

Honorable

Distinguished

Distinguished

Honorable

Reputable

Reputable

Distinguished

Distinguished

Reputable

Distinguished

Reputable

Distinguished

Admirable

Distinguished

Reputable

Intel Master

Share this page