• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

4790K running way too hot?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
When I run prime small fft the temp is hovering @ 80C!. Same thing for 2 CPU from different batch and H100i or H110 give same result. I tried at least 3 different TIM, (Glid, Artic Silver, Noctua not to mention the one that come pre-applied with the corsair W/C kit. CPU was cleaned with artic clean between TIM application. CPU has been check to make sure it is properly seated, same goes for the H100i and H110.
Click to expand...

So you have had 3 cpus all with same temps despite different tims, different cooling kits, different mounts. And your temps btw, are normal for prime 28 small ffts. There are couple hundred others getting same temps, many worse.

The ones that really complain are those on air with stock vids at 1.25 hitting 100C running prime 28.5 small ffts at stock settings. Im pretty sure one posted on this forum.
 
Mandrake4565 said:
I re read your first post running a test bench with 20c ambient temps. It may be that you just have a hot chip. Does the voltage stay at 1.175 the whole test or does it fluctiate higher? Is there a way to see the max voltage on Real Temp?

It does seem to be running hotter then the few 4970k's I've seen come through here but it is tough to compare to them because of the different cooling and ambient temps.

One thing I do know about these Haswells chips is some of them run really hot. The 4970k s were supposed to improve on this but how much on one of the hotter chips IDK. I can tell you from the experience I have had with my 4770k is mine is a beast to cool. I cannot do more then 4.4 on a full custom loop, see signature, because at 1.23 V I'm already in the mid 90's with an ambient temp of 22c. I have compared mine to another user on this forum and his running 4.4 with 1.25 v and the same ambient temps, running a Thermal take 2.0 AIO cooler his temps were 10c+ better across the board them mine.
Click to expand...

The voltage stay rock solid @ 1.175V does not fluctuate at all.
 
Tradermole said:
I already got 2 of the dud 4790k. Worst CPU EVER. I might try to delid one.

BTW: I got same horrible result using Intel Burin test.
Click to expand...

IBT test, if newest one with newest instructions pulls roughly same wattage as prime 28.5 small ffts.
wattage dissipated just from cpu package, 1.29v, 4.7ghz
prime 28.5 small ffts 175W
IBT or LINx (newest) 170-176W
Prime 28.5 large ffts 120W (hence temp difference with small ffts)
prime 27.9 small ffts 145W
prime 27.9 large ffts 120W
XTU 105W
AIDA64 100-105W
 
Mandrake4565 said:
I re read your first post running a test bench with 20c ambient temps. It may be that you just have a hot chip. Does the voltage stay at 1.175 the whole test or does it fluctiate higher? Is there a way to see the max voltage on Real Temp?

It does seem to be running hotter then the few 4970k's I've seen come through here but it is tough to compare to them because of the different cooling and ambient temps.

One thing I do know about these Haswells chips is some of them run really hot. The 4970k s were supposed to improve on this but how much on one of the hotter chips IDK. I can tell you from the experience I have had with my 4770k is mine is a beast to cool. I cannot do more then 4.4 on a full custom loop, see signature, because at 1.23 V I'm already in the mid 90's with an ambient temp of 22c. I have compared mine to another user on this forum and his running 4.4 with 1.25 v and the same ambient temps, running a Thermal take 2.0 AIO cooler his temps were 10c+ better across the board them mine.
Click to expand...

I think a lot of people dont run prime 28.5 small ffts. Everyone that I have seen that has reasonable temps, was because they did not hit small ffts, where wattage dramatically increases.

That being said, like you said there is a 5-10C difference between cpus, and wouldnt be surprised if a lot of that was tim1 differences, given it is mechanically applied with varying thickness because of varying adhesive thickness.
 
That being said, like you said there is a 5-10C difference between cpus, and wouldnt be surprised if a lot of that was tim1 differences, given it is mechanically applied with varying thickness because of varying adhesive thickness.
Click to expand...
Very true, it could be TIM mounting differences. I can say though, it was Johan and I that were testing our two 4770k's neither of us could believe how hot mine runs. I mounted and remounted my block at least 6 times using 3 different pastes, Gelid Extreme, Mx2 and Noctua. My temps changed 2-3 c tops and were still significantly higher then his, with he running more voltage on lesser cooling. Mines a darn good clocker on -10c cold water, as high as 5.3 but just runs hot.

My thinking is Tradermole has a similar chip to mine even though the 4970k is supposed to be improved, it may just be a hottie!
 
rge said:
IBT test, if newest one with newest instructions pulls roughly same wattage as prime 28.5 small ffts.
wattage dissipated just from cpu package, 1.29v, 4.7ghz
prime 28.5 small ffts 175W
IBT or LINx (newest) 170-176W
Prime 28.5 large ffts 120W (hence temp difference with small ffts)
prime 27.9 small ffts 145W
prime 27.9 large ffts 120W
XTU 105W
AIDA64 100-105W
Click to expand...

Is Prime 27.9 Large a valid stability tester?
how do you test the memory beside IBT?

Thanks (sure is a lot of helpful peoples here)
 
The whole thing is you will never equal prime95 loads in normal use.

I game at 55C on mine.


When I use prime I shoot right up to 99C but my rig is stable. 120 X 5, It is to much wattage going through such a small space.


Its a little bitty electric motor just a screaming.
 
Actually when I say small or large, i mean blend mode with small or large running. Though small and large very similar wattage draw regardless of mem being used.

So use 27.9 blend, it will run through small ffts on blend, but temps are 9-10C cooler on 27.9 blend versus 28.5 blend. If you get 27.9 blend stable for long run, and add .01v, you should be good for anything.

If temps are still too high for 27.9 blend, then you can run custom blend 28.5 or 27.9 and only run ffts that are 128 or higher. It is still better than running aida64 or xtu.

These are the least vcores I need to run each program stable, had posted these elsewhere on different bios, slightly different settings, didnt include pics with this repost.

4.7ghz 1.19V
AIDA64 crashed at about 10 mins
XTU crashed about 7 mins
prime 27.9 instant bsod
prime 28.5 instant bsod

4.7ghz 1.21v vcore
AIDA64 stable for 1 hour (see pic below), HWM shows ~100W cpu package power
XTU stable for 40 mins (didnt test 1 hr yet), HWM shows ~100W cpu package power
prime 27.9, bsod in 4 minutes
prime 28.5 instant bsod

4.7ghz 1.265V
prime 27.9 stable 1 hours, crashes after few hours 110W blend with large ffts, 145W blend during small ffts
prime 28.5 bsod after 10 mins

4.7ghz 1.292
prime 28.5 stable 24 hours 175W blend during small ffts
prime 27.9 150W during blend with small ffts
XTU variable load 100-112W
AIDA64 110-113W
 
rge said:
Actually when I say small or large, i mean blend mode with small or large running. Though small and large very similar wattage draw regardless of mem being used.

So use 27.9 blend, it will run through small ffts on blend, but temps are 9-10C cooler on 27.9 blend versus 28.5 blend. If you get 27.9 blend stable for long run, and add .01v, you should be good for anything.

If temps are still too high for 27.9 blend, then you can run custom blend 28.5 or 27.9 and only run ffts that are 128 or higher. It is still better than running aida64 or xtu.

These are the least vcores I need to run each program stable, had posted these elsewhere on different bios, slightly different settings, didnt include pics with this repost.

4.7ghz 1.19V
AIDA64 crashed at about 10 mins
XTU crashed about 7 mins
prime 27.9 instant bsod
prime 28.5 instant bsod

4.7ghz 1.21v vcore
AIDA64 stable for 1 hour (see pic below), HWM shows ~100W cpu package power
XTU stable for 40 mins (didnt test 1 hr yet), HWM shows ~100W cpu package power
prime 27.9, bsod in 4 minutes
prime 28.5 instant bsod

4.7ghz 1.265V
prime 27.9 stable 1 hours, crashes after few hours 110W blend with large ffts, 145W blend during small ffts
prime 28.5 bsod after 10 mins

4.7ghz 1.292
prime 28.5 stable 24 hours 175W blend during small ffts
prime 27.9 150W during blend with small ffts
XTU variable load 100-112W
AIDA64 110-113W
Click to expand...

Thanks will try that. now how to I test the stability of the RAM? (normal I use IBT)
 
Thanks every one will applied all the tricks and tips I got here and post the result in a few days!
 
outhouse said:
So do not run just small alone?
Click to expand...

I meant for him to just run prime blend, if it passes that for a long run, then memory and cpu should be stable, as goes through all small and large ffts in blend, if run long enough plus tests memory that way.

But wattage consumed on prime small ffts (on die not via blend) 12k 175W is roughly same as prime blend when same 12k fft hits 173-175W. I tried varying memory as well whether 2mb or 12mb and wattage same 173-175W.

And Tradermole, like outhouse said, which is most important point, prime temps dont matter as long as you can get through the stability test, because once your stable, your cpu will never see those high temps again in normal use.
 
Only time its safe to say you'd see max temps on normal use would be folding or benchmarking.
 
Welp I fired up my new 4690K, with cm 212 evo heatsink, 1.18v lets me go to 4.5GHz stable prime95 loading all 4 cores.

59 degrees according to speedfan.

So if you're getting higher than that, something is wrong.

Edit: Played with it a bit more:
4.6 GHz - 1.2 V - 65C
4.7 GHz - 1.25 V - 70C

I'm sticking to 4.5 because of the awful fan profile.
I bet the bugger could do 4.9 with better cooling.
 
Last edited:
Shell said:
Welp I fired up my new 4690K, with cm 212 evo heatsink, 1.18v lets me go to 4.5GHz stable prime95 loading all 4 cores.

59 degrees according to speedfan.

So if you're getting higher than that, something is wrong.

Edit: Played with it a bit more:
4.6 GHz - 1.2 V - 65C
4.7 GHz - 1.25 V - 70C

I'm sticking to 4.5 because of the awful fan profile.
I bet the bugger could do 4.9 with better cooling.
Click to expand...

Your cpu doesnt have hyperthreading so temps will be lower, but those are still very low temps. Can you post a screenshot of cpuz showing 4.7ghz with 1.25v, running prime 28.5 small ffts so we can see temps while running.

My cpu temps with HT off with prime 28.5 small ffts are 65C max at 4.7ghz 1.25v, but I am delidded so temps will be about 15C cooler at that wattage than one not delidded, plus I am on custom water.
HToff4700_125v.jpg
 
Last edited:
Hmm, those temps...

On Linux sensors gives a much higher number(+10C) than Speedfan does on Windows.
Prime95 in WINE works perfectly:

4.5 GHz - 1.18 V - 70 C - fan 50%
4.6 GHz - 1.2 V - 75 C - fan 75%
4.7 GHz - 1.25 V - 80 C - fan 100%

In the BIOS I lowered the thermal cutoff to 80C instead of the default 100C so it can't get too hot under load, I tripped it at 4.7 with the fan at a lower setting when sensors said I hit 81 C, so I guess it's accurate.

When I'm over >70% that 212 evo whrrs and wobbles so I'm sticking to 4.5.

In other news my GA-Z97X-Gaming 3's VRMs stays really cool even priming at 4.7, I guess 6-phase is plenty for OCing haswell, it was the cheapest board I could find locally that had more than 4 phase.

GIGABYTE says VIN needs to be >0.4 V above the CPU voltage... I wonder what happens if I lower it, if the CPU's regulation is linear it'd lower temps, but if it's switch mode it would do the opposite....
 
I lowered vin on mine from 1.88 to 1.7 and temps decreased maybe 1C (running prime small ffts and watching load temps and using GTL to drop vin while it was running and then switching vin v back and forth real time. I just keep mine at 1.88 since that is what I ran for 24 hours prime stable. And 1.88 vdroops to 1.85 under my LLC moderate settings. Some others on oc.net said they were getting larger temp drops by lowering vin...but cant duplicate their findings on mine, and some measured and found no difference.

edit: just ran it again using HWM to monitor cpu package power, since temps too difficult to monitor small change. decreasing vccin/vrin actually increase cpu package power.
 
Last edited:
What about case ventilation? What is the make and model of your case? Please outline for us the size, placement and orientation of your case fans with regard to intake and exhaust of air.
 
You must log in or register to reply here.

Similar threads

Voodoo Rufus
Haha! I did it! Win7 on Coffee Lake
Replies
0
Views
185
Voodoo Rufus
Voodoo Rufus
HankB
Comments on this build?
2
Replies
21
Views
943
DaveB
DaveB
Pepi93
Corsair CPU Liquid cooler troubleshooting...
Replies
2
Views
666
BugFreak
BugFreak
N
CPU unstable no matter what I do, and it's getting worse
2
Replies
25
Views
1K
mackerel
mackerel
L
1.45v on DDR3L ram
Replies
2
Views
547
lampino
L
Back