Help! 4770K Hitting 100c @ Stock On Water [SOLVED]

[Rvenger]

I did the same thing as Kenmitch and bumped my old trusty 4770k to 4.2ghz. No increase in voltage either but I am now using a Z97 board that appears to run cooler. :hmm:

EDIT: appears that OP edited post

 

[Gr3y Wolf]

You probably just need more vcore as your still pretty low for the clock speed. Just up it 0.010v and test again. If your getting bsod then maybe try 0.020v's.




On your first post in this thread click on edit. Then click on go advanced I think it's called. It'll let you edit the title.

Im aiming for as low a voltage as i can get, so im increasing by .002v each time, kinda time consuming but worth it for lower temps

I just changed the title, thanks for telling me how.

 

[Gr3y Wolf]

I did the same thing as Kenmitch and bumped my old trusty 4770k to 4.2ghz. No increase in voltage either but I am now using a Z97 board that appears to run cooler. :hmm:

Well im aiming for 4.5Ghz... and ive been doing some reading that suggests that because my RAM is 2400Mhz, that it can effect how high a clock (i almost missed the 'L' out then, haha ) i can get on the CPU.

wasnt aware that Z97 runs cooler... ah well, i think my Z87 is good enough.

 

[crashtech]

Ive seen something about a ring bus in the bios, but havnt changed it because i didnt know what it was for... is that important for overclocking?

You asked about cache, and the ring bus is how Intel connects the CPU cores to the L3 cache.

 

[Gr3y Wolf]

You asked about cache, and the ring bus is how Intel connects the CPU cores to the L3 cache.

Thanks for the info... but does altering the ring bus help with OCing?

Btw, im at 4.5Ghz, although it has taken 1.21v to get there, max CPU temp was 76C.

 

[Kenmitch]

Im aiming for as low a voltage as i can get, so im increasing by .002v each time, kinda time consuming but worth it for lower temps

I just changed the title, thanks for telling me how.

I guess if nothing else it'll keep your adventure going. They'll be a point when you hit the thermal wall....At least during stress testing.

No problem.

 

[Gr3y Wolf]

I guess if nothing else it'll keep your adventure going. They'll be a point when you hit the thermal wall....At least during stress testing.

I DO like adventures... maybe thats why i play Skyrim so much

By thermal wall, i suspect you mean when i get close to the TjMax of 100c when the CPU starts to throttle. I think i could probably push 4.6Ghz and stay under 90... but i'll cross that bridge if and when i come to it

 

[Ed1]

Im aiming for as low a voltage as i can get, so im increasing by .002v each time, kinda time consuming but worth it for lower temps

I just changed the title, thanks for telling me how.

Does your MB support 0.002 increments ?
My Asus Z77 smallest is 0.005v

On software reading ,it seems not able to read that fine , I see this scaling for example
1.104, 1.112, 1.120, 1.128
seems to be 0.008v steps for software here .

 

[Gr3y Wolf]

Does your MB support 0.002 increments ?
My Asus Z77 smallest is 0.005v

On software reading ,it seems not able to read that fine , I see this scaling for example
1.104, 1.112, 1.120, 1.128
seems to be 0.008v steps for software here .

Not actualy sure if it supports so small an increment, but it allows me to enter the value manualy and sticks to it - except monitoring programs show that it is running at a higher voltage than what i have set.

 

[Ed1]

Not actualy sure if it supports so small an increment, but it allows me to enter the value manualy and sticks to it - except monitoring programs show that it is running at a higher voltage than what i have set.

if it sticks, it probably does , when in bios on right there should tell you smallest amount supported .
PDF manual should say too .
Edit; here manual for your MB .
http://dlcdnet.asus.com/pub/ASUS/mb/LGA1150/MAXIMUS-VI-HERO/E8459_Maximus_VI_Hero.pdf

going down to CPU voltages its little confusing , first section option ,CPU core voltage it say 0.003 increments but below that for manual and offset modes it then says 0.001 increments .
So its pretty fine .

 

[Gr3y Wolf]

if it sticks, it probably does , when in bios on right there should tell you smallest amount supported .
PDF manual should say too .
Edit; here manual for your MB .
http://dlcdnet.asus.com/pub/ASUS/mb/LGA1150/MAXIMUS-VI-HERO/E8459_Maximus_VI_Hero.pdf

going down to CPU voltages its little confusing , first section option ,CPU core voltage it say 0.003 increments but below that for manual and offset modes it then says 0.001 increments .
So its pretty fine .

Thanks for the link.

Yeh, .001 is the lowest i have been able to change it - not sure why it says .003, somewhat confusing, but i have noticed that if an OC is 'almost' stable in prime, if i then up the voltage by .001 that is sometimes enough to stop me getting errors, so it seems that an increment of .001 does actualy take effect - even if monitoring programs show a higher voltage.

 

[Ed1]

Thanks for the link.

Yeh, .001 is the lowest i have been able to change it - not sure why it says .003, somewhat confusing, but i have noticed that if an OC is 'almost' stable in prime, if i then up the voltage by .001 that is sometimes enough to stop me getting errors, so it seems that an increment of .001 does actualy take effect - even if monitoring programs show a higher voltage.

yup , software is only so accurate , your voltages seem good for your clocks .
They match up pretty close to mine but I haven't tried to tweak voltage to low as i normally add 10mv for safety margin .
But I am on 3570k so not exactly same, Haswell tends to be more temp sensitive .

 

[Gr3y Wolf]

Ok, this is wiered...

Yesterday my temps at 4.5Ghz @ 1.21v never went above 75C in Prime... Now today i run the test again - havnt changed anything, and now my temps are up 10C to 85C... What is going on????

 

[Kenmitch]

Ok, this is wiered...

Yesterday my temps at 4.5Ghz @ 1.21v never went above 75C in Prime... Now today i run the test again - havnt changed anything, and now my temps are up 10C to 85C... What is going on????

Strange indeed if you haven't changed anything.

Only thing that comes to mind is TIM pump out. Seems like it's way too early to be the issue to me at least.

 

[Ed1]

I never used AIO units, but if the pressure of the water block is greater than the pressure of the CPU socket clamp might it be the IHS shifted ?
This might be the reason you found it with such high temps, might of been good then he built it too .

I have seen many reports of guys that delidded and used reg paste but after a while temps started to climb .

 

[Gr3y Wolf]

Strange indeed if you haven't changed anything.

Only thing that comes to mind is TIM pump out. Seems like it's way too early to be the issue to me at least.

I doubt it being TIM pump out - at least not over night.

 

[Gr3y Wolf]

I never used AIO units, but if the pressure of the water block is greater than the pressure of the CPU socket clamp might it be the IHS shifted ?
This might be the reason you found it with such high temps, might of been good then he built it too .

I have seen many reports of guys that delidded and used reg paste but after a while temps started to climb .

Even if this was the case, which i doubt - i dont think that would happen in less than 24hrs... I havnt even moved the PC or even bumped it.

I dont get it - doesnt make sense, less than 24hrs pass and temps go up by 10C, even though i havnt changed a thing.

 

[Termie]

Honestly, you've been lucky so far, but the guy sold you a junked CPU. I'd probably leave the extreme overclocking tests in the past and just settle into a low-voltage overclock, perhaps 4GHz or 4.2GHz. Your heat spreader installation is insecure, and the TIM application method you used simply isn't going to be stable over the long term. Regardless of whether it was overnight or over a fortnight, the temperatures were bound to climb.

Sorry to be the bearer of bad news, but I'm not the only one delivering it. Better to just enjoy the system rather than pushing its (unstable) limits.

 

[Gr3y Wolf]

Honestly, you've been lucky so far, but the guy sold you a junked CPU. I'd probably leave the extreme overclocking tests in the past and just settle into a low-voltage overclock, perhaps 4GHz or 4.2GHz. Your heat spreader installation is insecure, and the TIM application method you used simply isn't going to be stable over the long term. Regardless of whether it was overnight or over a fortnight, the temperatures were bound to climb.

Sorry to be the bearer of bad news, but I'm not the only one delivering it. Better to just enjoy the system rather than pushing its (unstable) limits.

As undesirable as it is to hear... i have a feeling you may be right... at least until i get some liquid metal ultra under that heatspeader

 

[Gr3y Wolf]

Well this aint going well at-all, just tested again... now over 90c.

Soon i may as well remove the [SOLVED] from the thread title...

 

[Kenmitch]

Well this aint going well at-all, just tested again... now over 90c.

Soon i may as well remove the [SOLVED] from the thread title...

What about lower over clocks?

Some people use liquid metal and crazy glue the lid back on.

If the lid shifted maybe a piece of paper between the tabs on the spreader and locking mechanism would work. The adhesive was most likely about paper thin.

 

[Gr3y Wolf]

What about lower over clocks?

Some people use liquid metal and crazy glue the lid back on.

If the lid shifted maybe a piece of paper between the tabs on the spreader and locking mechanism would work. The adhesive was most likely about paper thin.

Hmmm... thats not a bad idea, didnt think of that - thanks for the tip - i'll have to try that.

 

[Ed1]

read this thread IDC did on delidding, lots of good info .

http://forums.anandtech.com/showthread.php?t=2261855&highlight=

his measurement turned out after deliding there be a gap between PCB and IHS of 0.14mm or 0.0055" .
So a pice of paper is generally around 0.004" so you still got 0.001 of gap (thats w/o paste on die ).

 

[Gr3y Wolf]

read this thread IDC did on delidding, lots of good info .

http://forums.anandtech.com/showthread.php?t=2261855&highlight=

his measurement turned out after deliding there be a gap between PCB and IHS of 0.14mm or 0.0055" .
So a pice of paper is generally around 0.004" so you still got 0.001 of gap (thats w/o paste on die ).

Thanks for the link and info. Man... why did Intel have to mess up so badly in the first place that people even felt they had no choice but to delid... it just creates more problems than its worth. Ah well.

 

[Ed1]

Thanks for the link and info. Man... why did Intel have to mess up so badly in the first place that people even felt they had no choice but to delid... it just creates more problems than its worth. Ah well.

Also when you place IHS over die you need to offset it a bit towards the hinge as when you move clamp arm down the tabs move down as it clamps, moving the IHS a bit .

Maybe that is a issue too .