[Now Delidded!] 3770k Overclocking Help Please

C6ckneyGeezer · 11 Apr 2017 at 18:53

Hi all! I have was running a 3570k @ 4.2Ghz, 1.195vcore and high LLC. This gave me 1.2vcore under full load. I was running like this for a good 12 months if I remember, no issues at all and a max load temp of about 80-85*c.

I have recently bought a second hand 3770k. I thought I would try the same sort of settings. So I tried 1.195vcore and high LLC. This gave me 1.2vcore under full load the same as before. Started at 42x multiplier and got all the way up to 45x before I see a WHEA error. So I backed it down to 44x and now testing more.

The questions I have is temps! This 3770k seems slightly hotter than my old 3570k, I am using HWMonitor to monitor things.

The reason why I decided to back of the multiplier and not up the voltage is due to the loads temps I am getting.

Idle Temps

Core#0 - 36*c
Core#1 - 37*c
Core#2 - 39*c
Core#3 - 30*c

Why is Core#3 so low?

Full Load Temps

Core#0 - 78*c
Core#1 - 89*c
Core#2 - 88*c
Core#3 - 75*c

Now it seems under load Cores#1 and #2 are on average 10*c higher than #0 and #3. How comes? Anything to worry about?

It averages out to about 90-91*c full load after about an hour of Prime95 small FFT's.

Of course I cleaned both CPU and my Cooler Master 212x with TIM Clean and applied some new Thermal Grizzly. HS not seated correctly?

Any help and advice very much appreciated! Thank you!

Edit: Just an update. I have got the voltage down to 1.164 under load, as of now everything is fine. Temps have dropped to a max of 85*c average. A worthy note is though, all 4 cores are now with in 2-3 degrees of each other?

opethdisciple · 12 Apr 2017 at 14:34

C6ckneyGeezer said:
Hi all! I have was running a 3570k @ 4.2Ghz, 1.195vcore and high LLC. This gave me 1.2vcore under full load. I was running like this for a good 12 months if I remember, no issues at all and a max load temp of about 80-85*c.

I have recently bought a second hand 3770k. I thought I would try the same sort of settings. So I tried 1.195vcore and high LLC. This gave me 1.2vcore under full load the same as before. Started at 42x multiplier and got all the way up to 45x before I see a WHEA error. So I backed it down to 44x and now testing more.

The questions I have is temps! This 3770k seems slightly hotter than my old 3570k, I am using HWMonitor to monitor things.

The reason why I decided to back of the multiplier and not up the voltage is due to the loads temps I am getting.

Idle Temps

Core#0 - 36*c
Core#1 - 37*c
Core#2 - 39*c
Core#3 - 30*c

Why is Core#3 so low?

Full Load Temps

Core#0 - 78*c
Core#1 - 89*c
Core#2 - 88*c
Core#3 - 75*c

Now it seems under load Cores#1 and #2 are on average 10*c higher than #0 and #3. How comes? Anything to worry about?

It averages out to about 90-91*c full load after about an hour of Prime95 small FFT's.

Of course I cleaned both CPU and my Cooler Master 212x with TIM Clean and applied some new Thermal Grizzly. HS not seated correctly?

Any help and advice very much appreciated! Thank you!

Edit: Just an update. I have got the voltage down to 1.164 under load, as of now everything is fine. Temps have dropped to a max of 85*c average. A worthy note is though, all 4 cores are now with in 2-3 degrees of each other?

Those temps are high for such a low voltage.

It's the cooler. You would need a beefier one to get better temps.

I had my 3570k at 4.4Ghz using 1.31v and was getting max 80c when running prime small fft for about 8hrs.

---

Then again normal usage wouldn't push it so hard so you might be OK.

Devrij · 15 Apr 2017 at 22:55

A better cpu cooler would help here. It looks like you have a nice chip though! I get 4.5GHz on my 3770K at 1.332V (as shown in cpuz). Mine is delidded with liquid metal instead of TIM, so my temps are significantly lower (max about 68C in a 20C room), but with your lower volts there's no reason you can't go further.

C6ckneyGeezer · 16 Apr 2017 at 14:35

Yeah I think I may be pushing the limits of my CM 212. The problem I have is none of the bigger coolers will fit in my case. I believe the max I can fit is 160mm height wise, the 212 missed the side of the case by a few mm.

I currently have 2x 140mm fans as an intake at the front, 1x 120mm at the back as an exhaust, and then 2x 120mm fans on my 212 flowing onto the exhaust fan. So airflow is really good. Like you both say, I guess it's the limitation of the cooler.

I did take the cooler off, cleaned the CPU and cooler base and then lapped the heatsink base too! That seemed to drop my temps by a few degrees.

The option of delidding seems appealing. It seems it drops temps by 15-20 degrees for most? However atm I am sitting at 4.4Ghz @ 1.178v. Totally stable. So it seems it's quite a good clocker? I get a max temp of about 82-84*c under prime and a max temp of around 70-75*c while gaming. So is it worth delidding and chasing numbers for a little bit more gain in performance, if at all?

C6ckneyGeezer · 16 Apr 2017 at 17:51

Well, after sitting here for 20 mins checking through delidding tutorials, I took a blade to my 3770k and delidded it!

After taking the CPU out and cleaning everything up really well with TIM Clean, I broke out the blade. I went for each corner first, then slowly moved down each side. Within about 10 mins the lid was off! As everybody knows, the generic paste Intel used was all hard and flakey. It will got cleaned up, with ease too tbh. I used TIM Clean, cotton buds and a credit card to remove and silicon. If you have a sharp blade and are confident, it's a really simple task. Just need to take time and care, remembering which why the die is running.

I then applied a layer Thermal Grizzly Kryonaut, as that's what I had at hand. Lined the lid up and install the CPU again. Cleaned and then applied some past to the lid too. Now all back up and running.

At this time Im seeing these temps in HW Monitor, using prime small fft's for about an hour.

Core#0 - 60*c
Core#1 - 64*c
Core#2 - 69*c
Core#3 - 59*c

Which is an average of 70*c max temp! So that means I have dropped about 15*c on average overall!

Over the moon with the result. Now to push more voltage and higher clocks, see where she tops out... If I am at 4.4Ghz now with only 1.176v, reckon the 5Ghz is in reach?

Devrij · 16 Apr 2017 at 21:13

C6ckneyGeezer said:
Well, after sitting here for 20 mins checking through delidding tutorials, I took a blade to my 3770k and delidded it!

After taking the CPU out and cleaning everything up really well with TIM Clean, I broke out the blade. I went for each corner first, then slowly moved down each side. Within about 10 mins the lid was off! As everybody knows, the generic paste Intel used was all hard and flakey. It will got cleaned up, with ease too tbh. I used TIM Clean, cotton buds and a credit card to remove and silicon. If you have a sharp blade and are confident, it's a really simple task. Just need to take time and care, remembering which why the die is running.

I then applied a layer Thermal Grizzly Kryonaut, as that's what I had at hand. Lined the lid up and install the CPU again. Cleaned and then applied some past to the lid too. Now all back up and running.

At this time Im seeing these temps in HW Monitor, using prime small fft's for about an hour.

Core#0 - 60*c
Core#1 - 64*c
Core#2 - 69*c
Core#3 - 59*c

Which is an average of 70*c max temp! So that means I have dropped about 15*c on average overall!

Over the moon with the result. Now to push more voltage and higher clocks, see where she tops out... If I am at 4.4Ghz now with only 1.176v, reckon the 5Ghz is in reach?

Epic result, that's similar to what I'm getting at much higher volts and now you have some temp headroom to push further. Remember that these things are a case of diminishing returns, so 5GHz is probably unlikely, but 4.7GHz sounds like a reasonable goal. Let us know how you get on!

C6ckneyGeezer · 16 Apr 2017 at 21:58

Devrij said:
Epic result, that's similar to what I'm getting at much higher volts and now you have some temp headroom to push further. Remember that these things are a case of diminishing returns, so 5GHz is probably unlikely, but 4.7GHz sounds like a reasonable goal. Let us know how you get on!

Indeed, I was amazed by the results. I mean you could take the route of a bigger and better cooler, for the price of what? Around £100. But the cost of delidding it was nothing, and dropped temps great!

I am currently testing 4.6Ghz @ 1.240v and LLC at High. This gives me a constant 1.248v idle and under load, so no drops or spikes, good solid and stable vcore. Temps are peaking at 79*c under full load. No WHEA errors or anything of the sort at the moment.

So if that's the case I can't see me getting to the realms of 1.3v with this cooler. So I would think you are correct, 4.6/4.7Ghz is where the limit is for me. And tbh, I am well happy with 4.6Ghz, I don't see the need to push it further, would rather keep the temps down. Under 80*c and running 4.6Ghz for a budget cooler is fantastic!

Devrij · 17 Apr 2017 at 09:16

Out of curiosity, did you glue the lid back on our is it held on by the bracket?

C6ckneyGeezer · 17 Apr 2017 at 13:09

Devrij said:
Out of curiosity, did you glue the lid back on our is it held on by the bracket?

I was debating about it. However as I used paste and not that liquid metal stuff, so I thought I would not silicon it back it back due to renewing the paste in the future. It was simple to align when you clamp the CPU back in, just hold it in place. No real need to silicon it back tbh, no dirt or dust can get in under the lid.

I have been doing some more testing. It seems to get 4.7Ghz I wants 1.3v+, so that the point where the vcore jump is. I got to 1.296v under load and got some WHEA errors. Then it all went a bit flakey and I think the BIOS got somewhat corrupted. So pulled the CMOS battery out and set it all up again for 4.6Ghz.

So now I have dropped back down to 4.6Ghz and just gave it a little bit more voltage, now at 1.250v and High LLC, which gives me 1.260v consistently idle and load. Fingers crossed that's the sweet spot!

mattyfez · 17 Apr 2017 at 13:44

Nice. Tempted to do this on my chip as it's quite volt hungry, but temps creep up over 4.6ghz to levels I wouldn't want to run 24/7.

For the price of a craft knife and a tube of paste.. What's to lose?

Is that thermal grizzly a non conducive paste?

C6ckneyGeezer · 17 Apr 2017 at 15:44

mattyfez said:
Nice. Tempted to do this on my chip as it's quite volt hungry, but temps creep up over 4.6ghz to levels I wouldn't want to run 24/7.

For the price of a craft knife and a tube of paste.. What's to lose?

Is that thermal grizzly a non conducive paste?

Yep I used the Kryonaut stuff which in non conductive! It's great paste, not cheap but worth it.

Only tips I have is take it slow and angle the blade up towards the lid, not the chip! For a 15-20*c drop in temps, can't fault it!

mattyfez · 17 Apr 2017 at 15:51

Sod it, I'm gonna go for it, could be a £250 accident waiting to happen, but i used to modify and paint warhammer figures back in my youth so I'm quite handy with metal fatigue and scalpels lol!

So kryonaght is the stuff to go for? I don't want an electricaly conducive past in case of an accident.

Thanks

mattyfez · 17 Apr 2017 at 16:27

Crikey, it's a bit pricey
https://www.overclockers.co.uk/ther...mance-thermal-paste-11.1g-3-ml-th-000-tg.html

Do they sell it in smaller packets, wasn't wanting to spend more than a tenner for a 4g syringe

mattyfez · 17 Apr 2017 at 16:29

Is this the same stuff?
https://www.overclockers.co.uk/ther...formance-thermal-paste-1-gramm-th-020-tg.html

BriT · 17 Apr 2017 at 16:30

They do 5.5g for 13 pounds

mattyfez · 17 Apr 2017 at 16:40

I suppose if I buy the 5.5g I could repaste my gpu at the same time.. As that runs hot as hell.

I can see a plan formulating!

mattyfez · 17 Apr 2017 at 17:15

The one gram tube should do the job though, or is that a different compound?

mattyfez · 17 Apr 2017 at 17:29

Can you prise it off with a blade when warming it up with a hair dryer? All the videos on utube have some kinda brace that you put the whole chip in and then tapp it off.. Doesn't seem right..

BriT · 17 Apr 2017 at 17:34

Yeah you can use the 1g as its the same. I've been waiting for this tool to come in stock ages. https://www.overclockers.co.uk/der8auer-delid-die-mate-2-hs-003-dr.html Don't trust my self with a razor lol

mattyfez · 17 Apr 2017 at 18:10

Cool thanks, the chip itself seems to be in the middle of the PCB, with a thin line of gunk holding the lid on, so I reckon if I warm the chip up and use a craft knife carefully I should be able not ruin it haha!