CPU Throttle

[ChristmasCat]

Hey, guys! As a noob at PC hardware, I thought I should ask some experts for help. My new rig has an i5 6600k in it, an ASUS Z-170 E motherboard, and a Cooler Master Hyper 212 Evo. I can successfully overclock my chip to 4.5ghz, rock solid, with temps in the low 60's. However, whenever I try to go to 4.6ghz, all seems to be going well; until I start the Aida 64 stress test that is. Every time, the CPU reaches a multiplier of 46, and reaches 100% load, works for about 5 seconds, before throttling back down to 70% or so, and the test says my system is unstable. I know the issue is not heat, as the CPU never even gets to 65 degrees C. I also checked the core voltage, not at the max I set of 1.35, sitting at about 1.33 or so at full load, and I have a 600w power supply with no GPU currently installed, so no issue with power I think. Have I reached the max overclocking potential of my chip, or am I doing something wrong?

Thanks!

[Imakuni]

2 minutes ago, ChristmasCat said:

not at the max I set of 3.5, sitting at about 3.3 or so at full load

I hope you mean 1.35, no? Because 3.5 would either insta melt the CPU or be a completely different voltage altogether.

[meenmeen1103]

Could possibly be the VRM temps causing the throttling?

[ChristmasCat]

oh god, ya. Sorry, ya, 1.35.

That would have been bad...

[ChristmasCat]

It could be. Do you know of a way to monitor that? 

[ChristmasCat]

They were cold to the touch during the whole test, if that means anything.

[ChristmasCat]

2 minutes ago, meenmeen1103 said:

Could possibly be the VRM temps causing the throttling?

The VRMS were still cold to the touch, if that means anything. The board has decent VRMs, so I don't know. 

 

[Imakuni]

4 minutes ago, ChristmasCat said:

oh god, ya. Sorry, ya, 1.35.

That would have been bad...

(Btw, quote / mention people when you respond to them, otherwise they might miss it).

 

Anyway, you've gotta hit the limit eventually... seems like you can't get 4600mhz on 1.35v, even when going for a potato test such as Aida64. Looks like your wall is 4500mhz.

[ChristmasCat]

1 minute ago, Imakuni said:

(Btw, quote / mention people when you respond to them, otherwise they might miss it).

 

Anyway, you've gotta hit the limit eventually... seems like you can't get 4600mhz on 1.35v, even when going for a potato test such as Aida64. Looks like your wall is 4500mhz.

Alright, cool. Thanks for all the help! (I'm pretty new to the forum, thanks for the tip.)

[meenmeen1103]

I'm not sure what all aida64 shows as far as info, but HWiNFO64 is fairly detailed. I'd run that along with the stress test and look for a temp that idles around 30-40C and hits ~70C when throttling is experienced, or just generally stops climbing at the same time the system throttles. Not all mobos will report the same sensors under the same names in all programs, so I did this to find that my system was doing what you described.

[Imakuni]

3 minutes ago, meenmeen1103 said:

hits ~70C when throttling is experienced

That's... a rather conservatie temp for a 6600k. Straight up throttle should only happen well into the 90s, and we usually deem the low 80s to be perfectly safe as well. You don't really need to restrain OP to 70C.

[meenmeen1103]

If nothing seems off there, you at least know definitively that it isn't limiting due to temps and you're just at the limit of the silicon's abilities

[meenmeen1103]

1 minute ago, Imakuni said:

That's... a rather conservatie temp for a 6600k. Straight up throttle should happen well into the 90, and we usually deem the low 80s to be perfectly safe as well. You don't really need to restrain OP to 70C.

Used that as a reference, but wasn't referring to the actual CPU temp. In my case, what HWMonitor listed as CPUTIN and what HWiNFO64 lists as the CPU temp under the mobo sensors actually seemed to be much closer to what an IR thermometer read from the VRMs than the CPU socket, with a difference greater than 10C. What the programs reported for other components were usually within 2C from what the IR read.

[meenmeen1103]

11 minutes ago, Imakuni said:

snip

Basically, if OP sees a high temp (without actually defining 'high') that caps out at the same time the throttling occurs, it could be suspect for causing the throttling. Not that it's a sure-fire way of explaining it, or that the issue is even thermal throttling, but it's a variable I experienced at least. Something else worth mentioning: this didn't happen with the same brand hardware.