4770K, the nightmare continues?

[proppsie]

Hi everyone!

I had a topic about my 4770K CPU before, and i had to accept that it's a really bad chip.

The highest oc. i could achieve on it was 4.2GHz with 1.25V, but recently i had to leave it at stock settings since my temps went f***ing crazy.

Now its running stock, turbo up to 3900MHz obviously and i ran some OCCT stress-tests.
First test results:
Temp max values in celsius:
Core 0: 96 (what the hell?)
Core 1: 85
Core 2: 78
Core 3: 80
Package: 87

CPU Vcore reading is: 1.26V - is this even normal?

Second test results were the same voltage wise, with H110 stock fans:
Core 0: 94 
Core 1: 72
Core 2: 69
Core 3:71
Package: 82

As you can see the core 0 temp is ridiculous , even the core 1 is too high for stock settings, voltage is on auto etc. And there is a Corsair H110 cooling this chip.
The CPU was delidded and thermal paste used is an Arctic MX4. Still using the IHS with the replaces TIM.

Edit: room temps went up since it's almost summer time, but not this drastically (from 21-22 to 23-24)
What should i do? After deliding, and testing temps were fine, max values were around 82 degrees with the 1.25V on the core and 4.2GHz oc. But after like 15 days temps started going absolutely crazy.
I have checked the paste and all, it was fine, not too much, just the right amount...

Any ideas what i should do? Set voltages to manual? What should i check? I'm seriously out of ideas, and i'm also pretty disappointed.

Thank you in advance!
 

[DavidTheWin]

You could try manually setting the stock multiplier and voltage in the BIOS rather than leaving it on auto

[Freaky_spider]

do you use special mounting gear for delidded CPU? the hight of the CPU is lower and thus the normal mounting hardware doesnt work

[Constantine]

Did you check your temps using other tools? Open hardware monitor's notorious for mislabeling temp sensors for me

[proppsie]

OCCT uses an integrated HW monitor right? So i'm guessing that could also be inaccurate. ASUS AI suite shows completely different temps. Did not check there before.

What other Software do you recommend using for temps?

I checked the contact with the cpu and the waterbloc, it seems to be completely fine btw.

[JuztBe]

OCCT uses an integrated HW monitor right? So i'm guessing that could also be inaccurate. ASUS AI suite shows completely different temps. Did not check there before.

What other Software do you recommend using for temps?

I checked the contact with the cpu and the waterbloc, it seems to be completely fine btw.

Try core temp

[Jacky]

if u didnt delid it u probably could have returned it and got a new chip.

if your worried about the temps you could always underclock the cpu

[Freaky_spider]

if u didnt delid it u probably could have returned it and got a new chip.

no, why would intel take it back? even if he hadnt delidded it

[Jacky]

no, why would intel take it back? even if he hadnt delidded it

it is running way to hot for a normal chip at stock. and he could have opt for the overclocking option where they will send a new chip if somethings goes wrong during a overclock

http://click.intel.com/tuningplan/

[Freaky_spider]

it is running way to hot for a normal chip at stock. and he could have opt for the overclocking option where they will send a new chip if somethings goes wrong during a overclock

http://click.intel.com/tuningplan/

You realise you have to pay for that to apply...

[Jacky]

You realise you have to pay for that to apply...

it is better then having a chip that runs at 90 deg c at stock

[Freaky_spider]

it is better then having a chip that runs at 90 deg c at stock

yes, but as i said before, you need special mounting hardware when delidding, and he probably doesnt have that. so thats why it runs so hot

[proppsie]

yes, but as i said before, you need special mounting hardware when delidding, and he probably doesnt have that. so thats why it runs so hot

Just for clarification, i did delid the CPU, and replaced the TIM, but the IHS is back on it, and it makes good contact with the H110 block.

[Freaky_spider]

Just for clarification, i did delid the CPU, and replaced the TIM, but the IHS is back on it, and it makes good contact with the H110 block.

oww, yea owkay, 

 

maybe its your H110 that has a faulty pump and runs lower? could be it

[Recon-UK]

Stop using the cooler, grab a NHD15, that is some seriously FUKT temps mang.

 

For clarification, my H100 went on the fritz but kept working, Corsair coolers are not of teh best quality, see video:

 

[proppsie]

oww, yea owkay, 

 

maybe its your H110 that has a faulty pump and runs lower? could be it

It might be, ill test it with another cooler now.

I'm sorry if it was not clear that im still using the IHS. ill edit the post.

[proppsie]

I replaced the TIM again, now max temps don't exceed 70 degrees, just like the first time i did this. (core0 is only 69 max)

Could it be that with time this gets worse so quickly? (sorry for the double post)

[Faceman]

Further evidence as to why you don't buy an expensive CPU cooler until you know what kind of chip you are dealing with.

 

I'm sorry that you are having problems, there is an Intel hardware replacement do-dad specifically for overclocking.  It is $20-$25 and if you "overvolt" your chip, they will send you a new one.

[ProKoN]

Hi everyone!

I had a topic about my 4770K CPU before, and i had to accept that it's a really bad chip.

The highest oc. i could achieve on it was 4.2GHz with 1.25V, but recently i had to leave it at stock settings since my temps went f***ing crazy.

Now its running stock, turbo up to 3900MHz obviously and i ran some OCCT stress-tests.

First test results:

Temp max values in celsius:

Core 0: 96 (what the hell?)

Core 1: 85

Core 2: 78

Core 3: 80

Package: 87

CPU Vcore reading is: 1.26V - is this even normal?

Second test results were the same voltage wise, with H110 stock fans:

Core 0: 94 

Core 1: 72

Core 2: 69

Core 3:71

Package: 82

As you can see the core 0 temp is ridiculous , even the core 1 is too high for stock settings, voltage is on auto etc. And there is a Corsair H110 cooling this chip.

The CPU was delidded and thermal paste used is an Arctic MX4. Still using the IHS with the replaces TIM.

Edit: room temps went up since it's almost summer time, but not this drastically (from 21-22 to 23-24)

What should i do? After deliding, and testing temps were fine, max values were around 82 degrees with the 1.25V on the core and 4.2GHz oc. But after like 15 days temps started going absolutely crazy.

I have checked the paste and all, it was fine, not too much, just the right amount...

Any ideas what i should do? Set voltages to manual? What should i check? I'm seriously out of ideas, and i'm also pretty disappointed.

Thank you in advance!

 

 

 

i would suggest you re-apply mx-4 on the delid. also make sure the ihs is seated fairly centered under the cpu hold down latch.

 

I have had similar results with the temps increasing on the delid over time, however judging by the large difference in core individual temps from your chip I would have to say you used to much \ too little mx-4 or did not apply the tim very evenly.

 

even if your application is good, pushing the chip to hit 95+C may cause the tim to loose its viscosity and possibly "run" out the sides of the die and create poor contact \ thermal transfer. especially  if your board sits vertically.

 

intel will accept a dead delided processor for rma, as long as their is no physical damage to the pcb or ihs and you have adhered the ihs nice and centered. just kill it with volts.

 

your load voltage is perfectly normal. same thing an my boards as well. I input 1.38V and the true voltage is 1.4V. HW monitor Vcore reading is accurate for this platform.

[proppsie]

i would suggest you re-apply mx-4 on the delid. also make sure the ihs is seated fairly centered under the cpu hold down latch.

 

I have had similar results with the temps increasing on the delid over time, however judging by the large difference in core individual temps from your chip I would have to say you used to much \ too little mx-4 or did not apply the tim very evenly.

 

even if your application is good, pushing the chip to hit 95+C may cause the tim to loose its viscosity and possibly "run" out the sides of the die and create poor contact \ thermal transfer. especially  if your board sits vertically.

 

intel will accept a dead delided processor for rma, as long as their is no physical damage to the pcb or ihs and you have adhered the ihs nice and centered. just kill it with volts.

 

your load voltage is perfectly normal. same thing an my boards as well. I input 1.38V and the true voltage is 1.4V. HW monitor Vcore reading is accurate for this platform.

Very informative, thank you!

Yeah, i have tried reapplying the MX4 multiple times now. With different amounts of it, and the core 0 is always 8-10 degrees hotter than the other cores. Now i'm back at 4.2GHz with 1.25V. Max temp is 77 on Core 0 while the other cores are around  67-70.

The board sits vertically yeah. ( Here is my setup btw: http://linustechtips.com/main/topic/151194-my-dreambuild-nzxt-h440-matte-blackred-780ti-kngpn-edition-give-me-name-suggestions/ ) I dont know if i should try killing it and RMA-ing it though. Its really not a golden chip but yeah... i'll see how it does in the next month or so and if it really sucks ill say bye -bye to it. Maybe wait for a 4970K in september.

Thanks again. All of you! ^^

 

[ProKoN]

Very informative, thank you!

Yeah, i have tried reapplying the MX4 multiple times now. With different amounts of it, and the core 0 is always 8-10 degrees hotter than the other cores. Now i'm back at 4.2GHz with 1.25V. Max temp is 77 on Core 0 while the other cores are around  67-70.

The board sits vertically yeah. ( Here is my setup btw: http://linustechtips.com/main/topic/151194-my-dreambuild-nzxt-h440-matte-blackred-780ti-kngpn-edition-give-me-name-suggestions/ ) I dont know if i should try killing it and RMA-ing it though. Its really not a golden chip but yeah... i'll see how it does in the next month or so and if it really sucks ill say bye -bye to it. Maybe wait for a 4970K in september.

Thanks again. All of you! ^^

 

 

only try to kill it if there's little or no indication of it being delided.

 

PIV- 2.6V

 

Vcore-1.8V +

 

RIP Haswell package

[Recon-UK]

only try to kill it if there's little or no indication of it being delided.

 

PIV- 2.6V

 

Vcore-1.8V +

 

RIP Haswell package

A real shame intel decided to fudge the damn IHS up, Haswell could have been equal to Sandy in OC headroom.

[ProKoN]

A real shame intel decided to fudge the damn IHS up, Haswell could have been equal to Sandy in OC headroom.

 I have a theory , if the ihs was soldered the Overclocked Haswell parts may have rivaled the 6 core parts.

 

it is a real shame, agreed.

[proppsie]

So i decided to wait a couple of weeks and see if temps get worse with time. And i was right. I'm not sure what could cause this. It might be what ProKoN said. As you can see in my previous post, temps looked fine again after reapplying the TIM. 

Now here is a screenshot of the temps i'm getting again, only after two minutes of testing:



As you can see the Core 0 is reaching stupid temps again. And it's always that core...
Room temp is 23.5 degrees celsius.


 

[Faceman]

So i decided to wait a couple of weeks and see if temps get worse with time. And i was right. I'm not sure what could cause this. It might be what ProKoN said. As you can see in my previous post, temps looked fine again after reapplying the TIM. 

Now here is a screenshot of the temps i'm getting again, only after two minutes of testing:

test.PNG

As you can see the Core 0 is reaching stupid temps again. And it's always that core...

Room temp is 23.5 degrees celsius.

 

At this point, I would contact Intel and inform them that even at stock settings, your chip is crazy hot. It is ok for cores to be different temperatures, a 10C difference between cores is normal. You are well past the normal range and have a defective chip.