Overclock.net banner

1 - 20 of 233 Posts

·
Registered
Joined
·
945 Posts
Discussion Starter #1
So I finally killed it. Long story short I ran with stable 4.5 ghz overclock for more than a month. My chip was delidded with CLP between IHS and the die. Core voltage was ~1.32. Temps never went above 75C on Prime95. Few days ago I decided to try Prime95 with AVX.

I noticed that with AVX the voltage jumped to ~1.38, I was not too worried because temps didn't go any higher. I observed it for few minutes and left it run for few hours. When I came back my PC was dead. Would not turn on at all. Fans would spin and turn off right away.

Turned out that CPU was dead and when I took it apart it was physically broken with cracks on the die surface, see attached picture.

So what do you guys think happened? Did temperature slowly go up while I was away and CPU failed to throttle down as it should?
Can voltage be the cause of this damage even when temps are ok? Is Prime95 with AVX simply not safe to run? Could delliding with vise and hammer somehow do some micro damage and temperature finished the job?

Will appreciate any input.

P.S In the photo vrms are covered with IC7 diamond and stuff on the chip is dried CLP.

 

·
Registered
Joined
·
600 Posts
Very weird. 1.38 isnt that bad of a voltage, and those temps seem fine. There is a chance that the temps creeped when you were away because the way prime95 runs, after about 3 hours the intensity of the calculations increase, resulting in a temp increase. That's why i would never run prime for many hours without being there to check on it regularly. Especially att 75C under prime before ACX, there is a good change the temps could increase after several hours, possibly significantly resulting in the meltdown. Im no cpu expert, but thats all the input i can provide.

*edit* Also using the vice and hammer method for ocing is the safest way, and shouldnt really cause damage. There might've been some nicks or deep scratches if the vice was too tight, and that actually couldve caused your issue.
 

·
Premium Member
Joined
·
17,348 Posts
wow i wanna know what could have caused those cracks, thats very unfortunate. do you plan on getting another 4770k to replace it?
 

·
Registered
Joined
·
945 Posts
Discussion Starter #5
Yea I got it today, very happy that Mobo seems fine with new CPU, was worried that I could damage it too.
 

·
Registered
Joined
·
834 Posts
What was your cooling? Direct die?
 

·
Registered
Joined
·
1,107 Posts
Intel always says it's voltage that kills the chip and not the temps... I guess voltage jumped to the roof while you were gone. Did you use fixed voltage or adaptive? Did that 1.32 came from the 'auto' voltage settings?
 

·
Registered
Joined
·
945 Posts
Discussion Starter #14
Quote:
Originally Posted by Bartouille View Post

Intel always says it's voltage that kills the chip and not the temps... I guess voltage jumped to the roof while you were gone. Did you use fixed voltage or adaptive? Did that 1.32 came from the 'auto' voltage settings?
Adaptive with offset.
 

·
Premium Member
Joined
·
10,755 Posts
Quote:
Originally Posted by Bartouille View Post

Intel always says it's voltage that kills the chip and not the temps... I guess voltage jumped to the roof while you were gone. Did you use fixed voltage or adaptive? Did that 1.32 came from the 'auto' voltage settings?
Voltage degrades the chip. Last i checked, it didn't um, crack it in half
 

·
Premium Member
Joined
·
17,348 Posts
adaptive setting during a stress test can raise automatically even above the setting you put your voltage at, ive accidentally stress tested with adaptive on and my volts shot up from 1.25 to 1.4+ on its own.
 

·
Premium Member
Joined
·
11,043 Posts
Quote:
Originally Posted by Bartouille View Post

Intel always says it's voltage that kills the chip and not the temps... I guess voltage jumped to the roof while you were gone. Did you use fixed voltage or adaptive? Did that 1.32 came from the 'auto' voltage settings?
I was actually thinking the same thing, it must have spiked higher and higher. Although still strange considering CPU's have fail safes to prevent damage and actually shut the system down when temperatures hit a point. Unless Haswell's fail safes don't work?
 

·
Registered
Joined
·
635 Posts
Was that Antec 620 really good enough to cool "appropriately" an OC'ed Haswell?

Anyway, seems like I'll be sticking with "manual" voltage setting....
 

·
Premium Member
Joined
·
17,348 Posts
Quote:
Originally Posted by delavan View Post

Was that Antec 620 really good enough to cool "appropriately" an OC'ed Haswell?

Anyway, seems like I'll be sticking with "manual" voltage setting....
adaptive setting is known to do that during stress tests, manual should be used during intelburntest/prime95 etc due to the spike in voltage adaptive gives the cpu. in gaming/every day use adaptive should be best id think.
 

·
Registered
Joined
·
1,107 Posts
Quote:
Originally Posted by Mhill2029 View Post

I was actually thinking the same thing, it must have spiked higher and higher. Although still strange considering CPU's have fail safes to prevent damage and actually shut the system down when temperatures hit a point. Unless Haswell's fail safes don't work?
That's the thing with adaptive voltage. It bumps the voltage to the offset you give it to from the stock VID. But if I remember correctly, even at stock the chip overvolts itself when it sees avx instructions, so maybe that offset was too much? Also P95 has multiples 'stages' during the test, so maybe he just took a quick look and went to sleep, and during the other stages the voltage was even higher. correct me if im wrong. I hope this doesn't happen with fixed vcore
 
1 - 20 of 233 Posts
Top