Is my 4770K 'broken'?

[24HZ]

Well I was running a CPU intensive app (think folding but not quite) on my OC'd 4770K and now one of the cores gets way hotter under load than the rest. This only started happening after 48 hours of load with the app, not right away and didn't pop up either when I was stress testing the OC a few weeks ago.

As I've posted before, I'm running direct die with an H110. 4.5@1.35. ~78C average under load with my app before this happened. Nothing else has changed and all cores still exhibit the same temp ranges except core 3 which now goes over 100C. I've tried loosening the mount and tightening but to no effect. Core 3 idles ~3C hotter than the other cores as well.

I can't remount now since I don't have TIM handy (I took the machine to work, TIM at home).

Just curious if this is an expected side effect of pushing the CPU too hard?

 

[inf64]

Is it erroring (sic) out at core #3 when it hits over 100C?

 

[24HZ]

Its just throttling right now. When I push it hard and it errors it errors hard with a BSOD so I don't know what core is doing it.

 

[inf64]

So you are not sure it is the core #3 that is causing the BSOD? I recommend you lower the clock by 100Mhz, run OCCT with logging and see which core will fail first and at what temperature.

 

[Mushkins]

One core idling a little higher than the others shouldn't be a concern, as idle doesnt mean "do nothing at all." There's OS stuff going on in the background, and its probably putting a very light load on that core.

I cant think of a way to damage a single core that would make it still work but create significantly more heat. My best guess is there's a bubble in your TIM or the HS mounting is off in some way. Remount when you get a chance and try again.

 

[ShintaiDK]

One core will always be higher temperature than the others due to the layout. In this layout core 3 from the left would be the hottest.

 

[saad_ahmed]

yes inf64 is right you have to first check which core is failed on the top by reducing frequency and do note the temperature.

 

[NTMBK]

Make sure it's not broken certain instructions- try finding some apps which use TSX, or VT-d, see if they're working.

 

[24HZ]

Yay it was just a bubble or something that formed in the TIM. I removed the H110 block and reapplied NT-H1 and now all cores are back to normal temps.

 

[computerparts1]

Haswell is broken period.

 

[Idontcare]

Haswell is broken period.

Classic example of a needless trolling post. Don't be surprised if this kind of off-topic posting doesn't land you infractions in the future. Moderators and members alike are getting tired of the incessant trolling going on in CPU. Just a heads-up.

Yay it was just a bubble or something that formed in the TIM. I removed the H110 block and reapplied NT-H1 and now all cores are back to normal temps.

Glad to hear it was something simple and was easily resolved :thumbsup: