4670k temp issues

Toggle sidebar Toggle sidebar
G

Guest

Guest
Hello,

Having some issues with overclocking my 4670k. I seem to be able to get 4.4ghz (with 4.1 uncore/cache) relatively stable at 1.25V, but the temps get insanely high, so I've gone down to 4.2ghz (and 4.2 uncore) at 1.21V which seems to be stable currently, AIDA64 has been running for about an hour as I'm posting this, with an average temp of approx 65C and max 80C. It's stressing CPU/FPU/Cache/Memory.

While these temps feels fine, the temps when running Small FFT in Prime95 don't. Even with this current OC (4.2 / 1.21V) temps instantly jump up to 85 during the initial 12k test. Large FFT temps are about the same as AIDA64. If I try the new Linpack that runs AVX2 it gets to 100C fast.

Non-synthetic tests like ROG Real Bench or Cinebench get to ~65C. Ambient temp in my room at floorlevel is quite hot, ranging from 26-28C. Idle CPU temp is in the low 30's.

My rig:

CPU: i5-4670k
Mobo: Asus Maximuis VI Hero
Cooler: Corsair H80i running 2 Gentle Typhoon AP-15 fans (Radiator is slighly loose in that I can tilt it ever so slightly inside the case as some screwthreads were broken and I 'fixed' it with tape. Could this affect temps in any way?)
RAM: 8gb Corsair Vengeance 1866mhz, running XMP profile
GPU: XFX Radeon 7870, stock settings of 1000/1200
Case: Fractal Design Define R3. Case fans are 2 Front intake, 1 Bottom intake, 1 Top exhaust. The CPU cooler is set up as Rear exhaust.
PSU: Corsair TX650

Hopefully that's enough information, and hopefully someone can give me some tips on lowering my temps as I'm rather at a loss when I see people running worse coolers at higher freq/volt much cooler.

I got the H80i few days after the new processor/motherboard as my old Hyper 212 Evo was running too hot, but in honesty there wasn't really a difference between that one and the H80i, even though I took the time to sort out cablage for better airflow as well as get better front intake fans and add a bottom intake fan at the same time.. I'm not looking for a 4.6+ OC, but it just feels like something is really 'off' with these temps.

Sorry for the wall off text, hopefully it's readable. :)
 
Solution
Novuake
LOL pic 1 is a little much, the BEST method has always been the pea method. a 3/4 pea sized drop in the middle and let direct downward pressure spread the paste in a circle.

Try it, its worth it. Power technology is just for interest, since OCing Haswell is still pretty experimental, its related to power tuning to lower voltage for reduced power consumption, but to be honest, do not think that is a problem.
Sort by date Sort by votes
G

Guest

Guest
Less vcore and p95 small FFT gets a BSOD after 2-3min. :/ Definitely didn't win the silicone lottery, but I've seen people that need 1.25V+ to get 4.2 stable so it could be even worse in that regard. But from what I've seen, those people still get lower temps than me, often with a worse or the same cooler, which is what makes this so frustrating. Now sure, delid could be a option, but I don't have the spare money currently incase I mess up and CLU, which is what I'd want if I did delid for best results, doesn't seem to be available in Sweden atm.

And yes, I'm OCing in the BIOS. If memory serves, the changes done for this specific OC is:
Cpu mult: 42
Min / Max Cache Ratio (Ring bus / Uncore): 42
Vcore: Manual 1.21V
Vring: Manual 1.2V
CPU Spread Spectrum: Disabled
Initial CPU Input Voltage: 1.7V
Eventual CPU Input Voltage: 1.65V

Edit: Not sure if this is useful - but report from the 1h AIDA64 I just finished. It doesn't show Chassi3 fan for some reason, and the CPU fan there is the top exhaust (as it's plugged into CPU_OPT). The H80i fans don't show either, but are working fine in Corsair Link. Water temp went up to approx 34C according to said Corsair Link, not sure if that is normal or not for 100% load over an hour.

Sxj9Krg.png
 
Upvote 0 Downvote
G

Guest

Guest
Just edited the last post with a report from the mentioned 1h AIDA64 test, it shows voltages there as well. While AIDA64 reports it as 1.210V, HWMonitor says 1.216V and CPU-z the same. I never run them at the same time for the record. I'm also avoiding offset/adaptive until I'm done running AVX/AVX2 tests so the solution is sadly not as simple as that. :(
 
Upvote 0 Downvote
G

Guest

Guest
Hm, Power Technology? Not sure I've seen that option - any idea what name it could possibly have in an ASUS BIOS? :)
As for Turbo - does that have any effect when OC'd on Haswell? I'll gladly try obviously!

Reading another post just now I saw this:

2x too much: http://cdn.overclock.net/2/26/26d296bd_NGRAT.jpeg
Good amount: http://cdn.overclock.net/5/5f/5f5816c4_k3q1D.jpeg

This is what the poster claimed at least. When I reapplied TIM after remounting my H80i, my amount was more in line with pic #1 as I used the method Arctic Silver suggests for Haswell processors (Arctic Silver 5 was all I had home), so it was thinner, but also got a bit closer to the edges.

I'd say what I applied was quite a bit less than pic1, but also quite a bit more than pic2. I've read up on applying TIM in the past and always thought I had it down right, but could this possibly be my issue after all? That would explain why the issue has persisted from my old 212 Evo if so, heh.



Edit: Tried disabling Turbo, and it removes the OC completely (sets Mult to Auto). If I change that back to 42, Turbo is reenabled.
 
Upvote 0 Downvote
Novuake

Novuake

Champion
Aug 27, 2011
14,308
0
56,960
LOL pic 1 is a little much, the BEST method has always been the pea method. a 3/4 pea sized drop in the middle and let direct downward pressure spread the paste in a circle.

Try it, its worth it. Power technology is just for interest, since OCing Haswell is still pretty experimental, its related to power tuning to lower voltage for reduced power consumption, but to be honest, do not think that is a problem.
 
Upvote 0 Downvote
Solution
G

Guest

Guest
Well, it was worth a shot. :) Sadly did not help, as Turbo could not be disabled without removing OC completely, and shutting off all the C-states / EIST didn't change temps at all. I suppose I will try reapplying TIM once more tomorrow and use less than previously, even tho the more I look at pic1 the more I realize mine was never THAT bad, heh.. So while I doubt that will magically solve everything, I can always hope - or perhaps someone else comes up with a miracle solution

Thanks for helping me brainstorm some, appreciated! :)
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom