i5-6600 delid temperature issues

[J3is]

Hi everyone, I'm new to the forum.

Two days ago I delidded my 6600 (non-K) hoping to get better temperatures.
My CPU is overclocked to 4.4GHz at 1.33V using the BCLK OC method (on a Z170 Gaming M3 from MSI), and it's cooled by a 1 year and 3 months old LEPA Aquachanger 240 AIO (with Corsair's SP120 fans on it).

I had this CPU running at a maximum of 85°C, which seemed to be a bit high for a liquid cooler, so I wanted to delid my CPU.
I used the 3D printed tool method with the vice, and it worked fine.
I used CLU between die and IHS, (and I don't think the application of it isn't right, since I removed it and put it back a few times trying to fix my problems, and there was a trace of it under the IHS, indicating that it touched it), and MasterGel Maker Nano between IHS and WB.
I also scraped off all the silicone between PCB and IHS (there are still some traces left, but they aren't making any extra thickness).

And now all my problems start.
I boot up the PC, fire up HWiNFO and Prime95 (26.6, without AVX instructions), and I see temperatures of 95-105(!!)°C.
I thought I ****ed up something, so I went on and replaced CLU.
Still the same.
At this point I started thinking that my CLU was somewhat fake or something, so I tried putting some MasterGel instead of it.
While doing this, I noticed that, 25 mins after I shut down the PC, my rad was cold, tubings were cold, wb was cold, but the part of the wb that makes contact with the CPU was still somehow hot.
I thought something was wrong with the AIO, so I went on and replaced it with the stock cooler I still had in the box.
Reboot the PC, and I see 40°C idle and up to 100 under Prime95 (we're still talking 4.4GHz 1.33V, and it's still 5°C less than the AIO), so I left it like that, thinking everything was ok.
Later that day I decide to perform a CMOS reset. Then I fire up Prime95 again and I see temps up to 91°C, at now stock frequencies and voltage, even though tests say that a non delidded 6600 should stay between 80 and 85°C.

I don't have any idea anymore, what could be causing this? Did I just break the laws of physics?

Thanks in advance.

 

[custom90gt]

You need to apply CLU to the lid and the die itself. You can't just put it on one layer and expect it to spread out.
There are lots of videos on youtube with proper application of it.

 

[J3is]

You need to apply CLU to the lid and the die itself. You can't just put it on one layer and expect it to spread out.
There are lots of videos on youtube with proper application of it.

Thanks custom.
I know that, and I used the brush included in the CLU package.
This was one of my many applications:

https://imgur.com/a/bi1b1

Even though it went a bit outside of the die I don't think it's not spread well. Also, the bubbles in the lower part of the die are due to the fact I had to pull to take it off.

What do you think?

 

[custom90gt]

You spread the paste on the die and IHS? What did you clean the stock stuff off with? How was your mount between the AIO and the IHS? I've never ran a stock heatsink on a 6600 so I can't comment on temps there, but 91C under prime95 seems reasonable for that small heatsink. I'm more concerned about why your AIO was getting so hot...

 

[J3is]

You spread the paste on the die and IHS? What did you clean the stock stuff off with? How was your mount between the AIO and the IHS? I've never ran a stock heatsink on a 6600 so I can't comment on temps there, but 91C under prime95 seems reasonable for that small heatsink. I'm more concerned about why your AIO was getting so hot...

Yes, I tried spreading it on both the die and IHS and only on the die, with no difference.
I removed the old paste with a towel similar to a coffee filter (so that it doesn't leave lint around) and some isoproyl alchohol (the first time I used the towel inside the CLU package, which is basically the same stuff).
I'm with you on the fact that 91°C seems resonable, but after delidding (and without AVX instructions) I thought it would've gone down even with the stock heatsink.
I'm pretty sure the mount between AIO and IHS was fine, since it never gave me this kind of problems and I had to take it off and put it back quite a few times, so it's unlikely that I mounted it wrong everytime.

 

[custom90gt]

Drops on the stock heatsink are likely non-existent since the stock heatsink probably gets saturated pretty quickly.

 

[trents]

And is your water cooler working properly? You say you put the stock cooler back on to test but I would expect the kind of temps you got with the stock cooler using the frequency and voltage of your overclock.

And how did you get your overclock to 4.4 using BCLK? I can't increase my BCLK more than 2-3 mhz on non K Intel chips before I get gross instability. Modded bios or something?

 

[J3is]

Drops on the stock heatsink are likely non-existent since the stock heatsink probably gets saturated pretty quickly.

Didn't know this. So you're saying that it's normal then? I'm a bit worried that temps are actually higher than official tests around the internet.

And is your water cooler working properly? You say you put the stock cooler back on to test but I would expect the kind of temps you got with the stock cooler using the frequency and voltage of your overclock.

And how did you get your overclock to 4.4 using BCLK? I can't increase my BCLK more than 2-3 mhz on non K Intel chips before I get gross instability. Modded bios or something?

On Skylake, FSB clock and PCI clock are separated from each other. So you can increase your BCLK as much as you want, and you'll be increasing only the CPU frequency. Obviously Intel doesn't allow this, but some older BIOSes on Z170 mobos do. And so I did.
There's a little problem here anyways. With the stock cooler, I'm touching 91°C at stock frequency, not OC, and so it seemed to me that it was a bit higher than the tests you find online.

Also, I'm not sure anymore my AIO is working, since it gets to 105°C straight away (even though it worked before delidding, even at 4.4GHz, staying at about 80-85°C), but I don't want to throw money on a new one before doing every test possible.
And none of my friends have a comparable cooler to test my CPU with, so I'm pretty much stuck.

 

[trents]

Didn't know this. So you're saying that it's normal then? I'm a bit worried that temps are actually higher than official tests around the internet.




On Skylake, FSB clock and PCI clock are separated from each other. So you can increase your BCLK as much as you want, and you'll be increasing only the CPU frequency. Obviously Intel doesn't allow this, but some older BIOSes on Z170 mobos do. And so I did.
There's a little problem here anyways. With the stock cooler, I'm touching 91°C at stock frequency, not OC, and so it seemed to me that it was a bit higher than the tests you find online.

Also, I'm not sure anymore my AIO is working, since it gets to 105°C straight away (even though it worked before delidding, even at 4.4GHz, staying at about 80-85°C), but I don't want to throw money on a new one before doing every test possible.
And none of my friends have a comparable cooler to test my CPU with, so I'm pretty much stuck.

Ha! I didn't know that. I do know that ASRock put out some motherboards that had a clock gen chip that accomplished this but they have disappeared. And because they incurred the wrath of Intel they issued bios updates that nixed that ability. What motherboard and bios version are you using? Please create a Signature and outline your components.

Also, is there a chance your warped the lid when removing it?

 

[J3is]

Ha! I didn't know that. I do know that ASRock put out some motherboards that had a clock gen chip that accomplished this but they have disappeared. And because they incurred the wrath of Intel they issued bios updates that nixed that ability. What motherboard and bios version are you using? Please create a Signature and outline your components.

Also, is there a chance your warped the lid when removing it?

I know that because of those being separated you can bump it up if your BIOS lets you. And yes, Intel nixed that by forcing manufacturers to push out BIOS updates (but nothing prevents you from using old BIOS versions, amirite? )
Huh, and it also disables some minor features, that's the price you have to pay for it.
I'm using a MSI Z170A Gaming M3, BIOS version A61.
I don't think any warp happened, since the lid stands flat on top of the PCB.
I created the signature, let's see if it appears here!

EDIT: yep it does!

 

[trents]

After applying the CLU and try to remove the lid do you get any sense of suction or resistance that would indicate good contact?

 

[J3is]

After applying the CLU and try to remove the lid do you get any sense of suction or resistance that would indicate good contact?

Yep, I do.
Also, in the photo I posted before I didn't apply any to the IHS, but it's there, so contact should be good.

 

[custom90gt]

With your AIO and it hitting 105C does it throttle the clocks? My guess is you may have a bad pump. I had a corsair H115i that the pump went bad on and it was hot instantly/would throttle the core speeds to nothing.

 

[J3is]

With your AIO and it hitting 105C does it throttle the clocks? My guess is you may have a bad pump. I had a corsair H115i that the pump went bad on and it was hot instantly/would throttle the core speeds to nothing.

(I may have just deleted my last post by mistake)
SOO, you may be right. As soon as I started up the PC (with the case open), the pump was really loud, then it slowly became quiet again.
I put the CPU at 3.3GHz 1.35V just to have a high heat output, rebooted the PC and fired up Prime95.
After some minutes in, package got to 115°C, and as soon as the cores neared 100°C, some of them started turning off, and the CPU throttled down to 170MHz, to the point where HWiNFO took a while to display the information after a refresh.
Ultimately, the PC shut off on me, and trying to reboot it (even after a while) results in a forced shut down during post.

I would post a screenshot if only I could boot the PC back up... Gonna put the stock heatsink back on.

 

[custom90gt]

I'd RMA your AIO, I think it's done.

 

[J3is]

Yep, definitely.

Thanks for the help!