i6700k running very hot

Chas__mazzucco · May 23, 2016

I did take the plastic off and i did put on thermal paste. Yet, my cpu is getting to 88 degrees celcius when i ran prime 95. I thought I was going to overclock this cpu but now i'm thinking i might need to underclock it. I just started monitoring CPU temps today after building last week and the numbers just seem very off. I get about 30-40c idle with a very light load (a few programs up). Not sure what my ambient is but im in NJ and my house is around 65 - 70 farenheit. One idea I have is that i didnt screw my Noctua C14-s on tight enough? Or maybe I just lost really hard in the silicon lottery.

Supahos · May 23, 2016

Which version of prime? A few of them done play well with skylake. Try a different cpu stress test and check temps

treewithahat · May 23, 2016

Prime95 has multiple different tests, one of the tests stresses the CPU to its thermal capacity, and is not really a good judgment of heat. If you have any demanding games or a different CPU benchmark, try running one of those and check the temperatures.

Gnuffi · May 23, 2016

first off, 88c is safe, ofc far from optimal if on stock settings
but
A.
6700k on some motherboards actually overvolts alot on auto settings, so check voltages, if its over volting to 1.5v that might be why, and a manual overclock with lower voltages would then fix the issue
B.
for the love of god dont test with Prime95, its almost gotten to teh point where its almost intentionally trying to be a chip killer
use Realbench, Aida64, Handbrake, OCCT
see what temps and more importantly what voltages you get if running Realbench and Aida64 at teh same time in 15-30min test

if you are running hot, and your chip+mobo doesnt do those heavy over volting like others had, then you most likely need to re-seat your cpu cooler and/or check thermal paste

synphul · May 23, 2016

Are you running p95 version 26.6? The latest version will create an unrealistic workload and much higher temps on newer intel cpu's.

http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

You might try tightening the cooler down a bit more. Given your ambient temps you should be lower than that with decent case airflow. That cooler benched around the same as an nh-d14 so it shouldn't be the cooler (aside from mounting). If you have a few programs up running with the cpu at 10-20% load then it's not idle, it's under light load. Idle is 0 cpu load and leave it that way for a few minutes (say 5min) with a temp monitoring program open like realtemp. That will be your idle.

It would help to know what your vcore is, using a program like hwinfo64. You may need to set the vcore manually if the board is applying too much on auto/default. Vcore will raise temps. Losing on the silicon lottery usually refers to overclocking and how far one can oc their cpu while staying within the vcore limits of the particular chip. For example only reaching 4.4ghz at 1.35v vcore when others are reaching 4.7 or 4.8ghz at 1.35v vcore.

Edit: I would disagree, 88c is NOT safe. Not for 24/7 operation or typical 100% load scenarios. No it won't immediately kill the chip but it's far from 'safe'. Mid 70's would be considered safe so long as real world loads (not stress tests) don't exceed around 70c for the skylake chips.

Intel warns that running cpu's sustained over the tcase temp can cause premature wear on the cpu. The 6700k's tcase spec from ark is 64c and says "Case Temperature is the maximum temperature allowed at the processor Integrated Heat Spreader (IHS)." Tcase is 5 degrees C lower than the temps measured at the cores themselves which sets max core temps around 70c. At stock (not overclocking yet) these shouldn't be serious concerns so I think it's important to find out why the cpu's running so hot already.

Chas__mazzucco · May 23, 2016

Hey everyone, so i found out 28.6 that I was running was bad. I uninstalled that and got intel xtu. on a 10 min test I averaged about 72-74 and hit 81 at one point but also bounced down as far as 60 every now and then. Are these temps acceptable for me to start OC or should I reapply thermal paste/ RMA CPU?

EDIT: I am in bed on laptop rn but when i check HWIF64 my core was 1.28 i think (i have a photographic memory)

Gnuffi · May 23, 2016

no reason to RMA cpu,
do the steps as advised above
check voltages while under stress (aida 64 is good for this, has voltage and temp gauge while running the stress test), recheck temps
then if still issues, re-seat cooler

treewithahat · May 23, 2016

Try keeping a set voltage. Gnuffi was also right about the variable voltage. Try to set a reasonably low voltage @ stock speeds and see if temps improve (disable C states, EIST functions, and turbo boost). Then, start applying frequency increases and voltage increases as necessary until temperatures become uncomfortable or the CPU becomes unstable.

Chas__mazzucco · May 23, 2016

How do i see if I am overvolting in the bios?

treewithahat · May 23, 2016

Chas__mazzucco :

There should be a voltage option in the BIOS. It varies by motherboard, so look at the manual or find a guide online for your motherboard.

Gnuffi · May 23, 2016

if voltages are on auto in bios you likely wont see it set as over volted while in bios
you will first notice this while actually being in OS and during stress tests, monitoring the voltages on 100% loads sustained peak, which is why i recommended AIda64(its a free trial program) because it lets you do this, without having to switch to HWmonitor and gauge the volt yourself
and while voltage measurements inside OS is rarely 100% accurate, tehy are a fair estimate,, so if it on stock settings, jumps to 1.4volts or more when only boosting to 4.2GHz you can bet safely its sucking too much power than needed
also Turbo boost does not need to be disabled, enabled is fine
else you run at set clocks all teh time, which is either stock low, or manual overclock high, both not needed

Chas__mazzucco · May 24, 2016

Gnuffi, i see you have a similar set up to me (asus board, noctua, i7) what temps are you getting at 4.6 ghz and whats your voltage? I'm willing to bet my issue is the over voltage because those numbers seem ridiculously high and I made sure I installed the CPU correctly. I'll check that once I get home.

Gnuffi · May 24, 2016

my voltage is 1.296, but in bios set 1.3 with CLL4, and im peaking at about 74c core temp, but thats with my Noctua cooler set to low'ish rpm, and my case being near dead silent
and is why i didnt attempt a higher overclock because my current setting allow me to game in near silence

and like i said, ive seen several times on different boards the 6700k sucking too much volt on stock settings, which could likely explain temps for some
i recall a user posting with volts above 1.5 in auto i think
bad for business if it indeed does that, and should then be manually lowered

doesnt even need to overclock, just set the boost multipliers to stock 42, but change the voltage from auto to manual, and lower it, 1.3 should be no problem,
i think its common around 1.28 stock volt

Gnuffi · May 24, 2016

synphul :

agree 88c is not safe for 24/7 operation, but since it was done in a stress test on prime95 even, i consider it safe in that scenario
far from optimal, but safe, since Prime95 intentionally tries to fry the chip there is no way it would ever encounter that stress and temps again outside the test
never did i mean it was safe for continuous operations, but it was a nice gauge to the issue, but had to bear in mind Prime95 exerted stress
not like he had killed/was killing his CPU right now by running that test, only if it was in prolonged continuous use would those temps be unsafe, for a peak its safe, even if far from optimal

Chas__mazzucco · May 24, 2016

So i tightened up my HSF, temps are lower but still less than optimal. The HSF was about 3 full spins on each side shy of being fully tightened. I reran XTU and got the same high (81) but that was a weird spike, for the majority of the time I was at 54-74C. My ambient is 21C and when I closed all apps (save like 5 low impact background ones and XTU) i was running at 31C. Should i reseat my HSF???

my voltage was jumping from 2.96-1.33 IVC(whatever that stands for)

Gnuffi · May 25, 2016

a 20degrees variance on load is alot, and strange, it should be somewhat stable on full load unless you have a fan curve set that engages at "X"c degree temp in the 54-74 range
do me a favour, run Aida64+Realbench at teh same time in a 15min test, and see what it cranks the temp up to steady or if they jump around there
and see if Aida64 +HWmonitor agrees with the Vcore voltages (despite inaccuracies its a good gauge)
because while 74c seems plausible for 1.33 with your cooler, depending on RPM, its not exactly optimal if your c14 is running at max, to keep it at 74c on stock

Chas__mazzucco · May 26, 2016

I do have a fan curve set up like that so Ill try just setting them all to 80%. So i turned CLL to 3, it was on auto, The voltage still wont stay at 1.28 though! itll always be at 1.3 something, I remember noting that i was a whole 30mhz (.03ghz) above my manually set voltage, what gives? Even if i turn turbo boost and multi frequency off itll still hit 1.3+... I'm thinking of upping my frequency and ratio so i can at least utilize the voltage to give me more performance.

also, my build is ->

and when gaming the cpu cooler is sucking up hot GPU air so I will get like 70C while gaming. Whats the best way for me to stop this? I can return my GPU for another two weeks. (or do i just go AIO H100i)

treewithahat · May 26, 2016

That cooler isn't gonna get you very far in overclocking. If you can fit an AIO, go for it.

Gnuffi · May 26, 2016

treewithahat :

its a decent cooler, maybe not nr1 top of the line air coolers, ut its solid enough
you only need AIO for heavy overclcoks, and even a good quality air cooler will handle that fine on this chip

Gnuffi · May 26, 2016

Chas__mazzucco :

those types of cooler blow inwards on to the cpu usually, unless you mounted the fans different, which will never give the same quality cooling as cpu coolers that blow from front to back (visualise the difference in airflow if you can)
you can get an AIO if you case fits it, but you can get big air coolers that will cool the same as teh h100i
the thing is since AIO is big int erms of rad, and air cooler is big in terms of direct size, you need to make sure your case will fit/allow either of them
you cooler should be decent enough if your exhaust fan can help removing air if the cpu fans are blowing inwards onto the cpu, (since its not the best direct, exactly because it sucks hot gpu air in)
it doesnt mean it wont or cant work, just that those coolers are not totally optimal, and then accounting for overclocks but
since you CPU is already at 1.3+volts with 70c, temps, just see what you can increase teh multipliers to and stillr emain stable
just remember to proper stresstest after doing increments, because if you dont even get a basic stress test stable with a modest multiplier remaining on the voltage, you need to retry until you get a basic stresstest stable before you even do the full on long stresstesting for hours
but leave the GPU off on the stresstest, get CPU independently first,
even tho my cooling might be better, and GPU less hot, even with a GPU overclock and and low "on GPU fan", i my temps still never rise nowhere near the temps i get just in stresstesting my CPU, even when gaming for 12+ hours straight
check if you cant atleast get a 44 multiplier going to start with, should be doable, and you might not even need LLC set if having everything on manual,
i only first needed to apply LLC 4 when i switched to adaptive mode to not run the full voltage all teh time

Chas__mazzucco · May 26, 2016

Thanks for your help Gnuffi, its been really helpful. So if I go and buy a used D15 off ebay and sell my c14s off ebay, would that be a good course of action?

Gnuffi · May 26, 2016

Chas__mazzucco :

its a much better for cooler for sure, and performs on par with top end AIO coolers like h110i and such big ones
only caveat is, its a big beast, sitting directly on teh CPU, so need to make sure Case allows the size (it will be specified in case speccs how large cpu coolers are supported, compare with NH-D15 spec size) and that your RAM clearance allows it
the NH-D14 is a tiny bit smaller than the D15 and will for some fit better, while performing almost on par
also it doesnt necessarily have to be such big beasts, there are smaller ones out there that are fine, but those 2 are definitely in teh top, but they pack the size too in order to be it
so if clearance is an issue another air cooler might be/fit better, or maybe even AIO if radiator space is there

so check your case spec cooler clearance and your RAM height clearance, before purchasing!

and if you do swap cooler, please remember to proper remove the old thermal paste, and not just slab the new cooler on (have friends that sadly did this)

but the C14 isnt a terribad cooler even if not optimal, suggest you try and see if you cant get it to be/perform somewhat decent first, before replacing

Gnuffi · May 26, 2016

are those the noctua c14 standard fans? looks different to me
if not using the standard fans, and either sub par fans, or fans mounted wrong, that would also impact the temps

i6700k running very hot

Commendable

Expert

Reputable

Honorable

Champion

Commendable

Honorable

Reputable

Commendable

Reputable

Honorable

Commendable

Honorable

Honorable

Commendable

Honorable

Commendable

Reputable

Honorable

Honorable

Commendable

Honorable

Honorable

Share this page