Sign-in / Sign-up
Your question
Solved

Very high temps and throttling when running stress/benchmarking tests 4670K

Tags:
  • Overclocking
Last response: in Overclocking
August 23, 2013 9:46:06 AM

Just finished building a new haswell rig and it seems to run fine (no throttling at good temperatures (4.5 GHz stock voltage at around 50-60 deg) when gaming. The problem is when I try to use any of the benchmarking//stress tests, the temperature goes straight up to 100 degrees and the CPU throttles back to around 3.5/3.6 GHz. Why is this happening? Is there any setting in the BIOS that needs to be changed in order to be able to run benchmarks correctly? Btw, my mobo is a Gigabyte Z87x-UD3H

More about : high temps throttling running stress benchmarking tests 4670k

a b K Overclocking
August 23, 2013 9:48:11 AM

Are you using the stock cooler? If so that's probably why.
m
0
l
August 23, 2013 9:50:03 AM

cliffro said:
Are you using the stock cooler? If so that's probably why.


Nope. I have the 212 EVO and used Arctic Silver 5. And as I said, no problems when running games, just with benchmark tests.
m
0
l
Related resources
August 23, 2013 10:01:32 AM

Most games don't task the CPU.

You're lucky if the game uses 40% of your cpu where stress tests use 100% of EVERY core.
m
0
l
August 23, 2013 10:17:34 AM

Are you sure you're running stock voltage man? Pretty sure you aren't. You can't get 4.5GHZ at stock voltage. Noone can.
m
0
l
August 23, 2013 10:19:08 AM

My guess is you didn't turn overclocking from auto to manual in the settings, what happens is that it attempts to use your settings for 4.5 GHZ, realizes that stock voltage isn't enough, cranks up the voltage for you when the CPU starts needing more power, because of this, Temps are going to Rise, Hence the Temperature rise. GL.
m
0
l
August 23, 2013 10:32:35 AM

rchar081 said:
My guess is you didn't turn overclocking from auto to manual in the settings, what happens is that it attempts to use your settings for 4.5 GHZ, realizes that stock voltage isn't enough, cranks up the voltage for you when the CPU starts needing more power, because of this, Temps are going to Rise, Hence the Temperature rise. GL.


Yeah I just went through the BIOS and the Vcore was at Auto with a 1.31V default!! Way too high. I switched to Manual and 1.25V for the Vcore and x45 multiplier. There's still throttling but much less than before and there are times than temperature rises to 90 degrees but then comes back to 65 during testing. Maybe I'm missing something else
m
0
l

Best solution

August 23, 2013 10:39:12 AM

My guess is you didn't seat your cooler right then, Most likely though you won't be able to get a stable 4.5 clock at 1.25Vcore. you'll need something like 1.27...... I would reseat your cooler cause at 1.25 you should be getting low to high 70's and nothing above that.... Seriously make sure you check that though I remember one time I installed my cooler and forgot to take the plastic off the heat sync. Not sure what else to say... You should also turn turbo on to extreme I think it normally sits on auto. That could solve that problem. You'll find turbo somewhere in there I forget where exactly.
Share
August 23, 2013 10:58:02 AM

Ok things are getting better. I dropped the voltage to 1.2 and it's stable at 4.4ghz while barely touching 80 degrees. I'll try to lower voltage even more
m
0
l
August 23, 2013 11:02:45 AM

I highly doubt you are stable at 1.2.... If you are you probably got one of the best chips out there..... HATS OFF to you. Run a stress test for a couple hours to see before you lower voltage anymore than that..... That's already extremely low for the clock you got going.
m
0
l
August 23, 2013 11:06:52 AM

rchar081 said:
I highly doubt you are stable at 1.2.... If you are you probably got one of the best chips out there..... HATS OFF to you. Run a stress test for a couple hours to see before you lower voltage anymore than that..... That's already extremely low for the clock you got going.


Lowered to 1.15V. Still stable after 20 cycles of lynx. Now i'll do intel burn test. Yeah I think I may have gotten an amazing chip :D 
m
0
l
August 23, 2013 11:16:23 AM

wtf.... 1.15 is insane... I have done a lot of research on this, you have the best chip I have seen yet. Nothing even compares to that. You might be able to get 5.0 on AIR. LOL WTF, I doubt it though but I would try... Seems stupid not too really. I bet you crash on that burn test though...
m
0
l
August 23, 2013 11:16:59 AM

Guess what... My chip gets 4.2 at 1.257 LOL.... Not even close to yours.
m
0
l
August 23, 2013 2:42:41 PM

Ok I just left it at 4.4ghz with a 1.15v vcore. I still find it weird that at this voltage I'm getting around 75/80 degrees when stressing. In gaming things seem better with temps below 60 deg
m
0
l
August 24, 2013 1:39:19 PM

Ok I give up. I feel like I got lucky and unlucky at the same time. My CPU seems to be capable of great clocks at low voltages and stay stable but my temps when using prime95 or aida64 are not good at all at those voltages. Right now I am at 4.2ghz and 1.1V and I get 83 deg on stress tests. At 1.25V, which is not an incredibly high voltage, I can go up to 4.7ghz and I can boot and play games and everything but in stress tests I go over 100 deg and the CPU gets throttled. I've read this may be due to a bad TIM between the cores and the CPU lid. I could fix it by opening the CPU but I'm not going to risk it since I don't have money right now to get a replacement in case something bad happens.
m
0
l
a b K Overclocking
August 25, 2013 6:53:05 AM

Are you running a single fan on the 212Evo? Have you considered a Push/Pull setup? Maybe something beefier?
m
0
l
August 25, 2013 10:32:06 AM

cliffro said:
Are you running a single fan on the 212Evo? Have you considered a Push/Pull setup? Maybe something beefier?


I am running single fan but I just checked and it does not make big difference (probably 2 to 3 degrees). Since my cpu is only heat limited by now I have decided to perform surgery on it xD (replace the TIM inside the lid). Already ordered Liquid metal ultra and some razor blades to open the lid. It's actually not that hard when I saw the videos and I'm very good handling electronics (EE major). Most people see a 20+ difference after this procedure so if everything goes the way it should then I may be able to push the CPU even further without having heat issues.
m
0
l
a b K Overclocking
August 25, 2013 8:45:02 PM

Good Luck, and sweet if it works out to those temp drops.
m
0
l
August 26, 2013 8:58:33 AM

cliffro said:
Good Luck, and sweet if it works out to those temp drops.


I'm hoping it does. Right now I have the CPU running at 4.5GHZ and 1.2V. Have not had any BSOD, crashes or anything of that nature but can't be for sure if it's stable since I simply cannot run IBT or Prime95 without reaching 100 deg. I've been reading and it seems the problem is the distance between the die and the heat spreader due to a very thick layer of TIM. I think I may have one of the worst chips out there (thermally speaking) so I should get a pretty steep drop in temps after delidding. I hope it's at least 20 deg.
m
0
l