Intel should really be ashamed. 8086k delid temp reduced 20C+ under load

J

JNavy89GT

[H]ard|Gawd
Joined
Apr 25, 2001
Messages
1,800
In a moment of weakness, and mania, I bought a 8086k in the quest for 5ghz stable. My 8700k only did 4.8ghz at reasonable volts with my AIO cooler. That 8700K was delidded, ran at 4.8ghz 1.36v, and maxed in the low 70's C in general. Before delid, it would peak well over 85C and programs would shut down due to temp limit.

The 8086k on the same Artic 240AIO cooler couldn't do 5ghz like so many reviews and reports suggest. Temps would get out of hand above 4.8ghz and 1.25v set in bios. At least this chip would do 4.8ghz at MUCH lower voltage. Attempts at 5ghz needed 1.3v to start stress testing, but temps would immediately spike to 85-90C and programs would shut off due to temp protection.

Over the course of a weeks use, and back at 4.8ghz and 1.25v just to keep temps more normal, I started to notice my temps going much higher. At 4.8ghz just loading 3d Mark would spike 2-3 cores over 85C for a split second and at the end of the run, all cores had seen 80C+. I rechecked/applied TIM to the cpu/AIO block a couple of times w/out help.

I got around to delidding the 8086k last night. Popped it in tonight, same AIO cooler, and at 5ghz I'm maxing in the mid/high 60's right now. My rad hasn't totally heatsoaked yet, so I'm sure this will go up over time. But I'm 10 mins in and with the stock Intel paste/TIM it would immediately spike to >85C and shut down OCCT before I pulled it for delid.

Inspecting the TIM interface during the procedure, I could tell about 2/3 of the cpu had a very THICK application of the paste which I assume was causing the high temps and as temps seemed to be getting worse over the past week, I'm assuming the TIM was cooking and not as effective. Intel should really be ashamed to have pushed this cpu as some special edition. I know I'm a tool for spending so much, but what can I say. I wanted 5ghz, and thought it was my best chance when buying a wild cpu.
 
Last edited:
passed 1.5hours of OCCT test, 5ghz, 1.295v set in bios. Max temp on 2 cores was 72C, others were 70C. Average was more in the mid/low 60's. Crazy the difference. I've delidded 5-6 cpus now. This is by FAR the most drastic difference. For a premium cpu, Intel really needs to up their game and at least get a better spec on their parts. There was just too much gap between the cpu and IHS/TIM. I'm not sure if it was too much TIM, not enough pressure, or the Z height of the IHS. I did lap the base of the IHS but I didn't take much material off, just enough to get a nice even surface which did get it to all copper fwiw. Anyway, just ranting a bit. Don't mind me.
 
Have you tried testing in Prime yet? Mine is stable in 26.6 but immediately crashes in 29 AVX. Stable in every other test.
 
This makes me seriously consider delidding my 8600K. I'm all aircooled and temps are pretty good for air but I've got like 3x120 mm front intake and 3x120mm out at top and 1x140 mm at the back.

Temps usually 80~85C in Prime95 non AVX load (with AVX ~95C) and for example gaming usually max 75C (typically 70-75C after hr or two of gaming now in the warm summertime with my absolutely disgusting room temp with no AC, during Winter it was some 7-8C lower even but it's cold in Finland :p).

BUT it's such a dud overclocker tho, I'm currently running 4.75GHz 1.31v, I couldn't get 4.8GHz fully stable at 1.33v or so even and during summer I think ~1.32v is roughly the max limit aircooling is capable of running these chips anyway without delidding at the very least. I wonder if you think there's chances I'd see an improvement in overclock besides temperature though if delidding, I mean I'd be supper happy to get 4.9GHz stable but yea 4.8~4.85GHz thereabouts I guess should at least be possible from delid.
 
RPGWiZaRD said:
This makes me seriously consider delidding my 8600K. I'm all aircooled and temps are pretty good for air but I've got like 3x120 mm front intake and 3x120mm out at top and 1x140 mm at the back.

Temps usually 80~85C in Prime95 non AVX load (with AVX ~95C) and for example gaming usually max 75C (typically 70-75C after hr or two of gaming now in the warm summertime with my absolutely disgusting room temp with no AC, during Winter it was some 7-8C lower even but it's cold in Finland :p).

BUT it's such a dud overclocker tho, I'm currently running 4.75GHz 1.31v, I couldn't get 4.8GHz fully stable at 1.33v or so even and during summer I think ~1.32v is roughly the max limit aircooling is capable of running these chips anyway. I wonder if you think there's chances I'd see an improvement in overclock besides temperature though if delidding, I mean I'd be supper happy to get 4.9GHz stable but yea 4.8~4.85GHz thereabouts I guess should at least be possible from delid.
Click to expand...

I’m running delidded 6700k and my chip is quite a dud too when it comes to OC, running [email protected] and 4.6 not being stable on air even at 1.35, above which I don’t wanna go and such voltage difference isn’t worth it for 100 MHz. That said, delid helped to decrease remps, but not really the voltage requirements. It was pretty much running at the same voltage, but shaved off 15-18 degrees, which is nice.
 
I've had my 8086K running at 5.3Ghz after delidding for a handful of weeks now. I'm seeing temps in the mid 80's under full load. I'm ok with that. It's not a CPU I plan on keeping the rest of my life and If I burn it out, which I won't, the $100 plus dollars I have into it after selling my old 8700K for $300 is not going to make or break me.

I do plan on getting a Intel i9 9900K and delidding it as well. I'm hoping I can get 5Ghz across all 8 cores. That would be an absolute dream for me.
 
If you still think Intel needs to be ashamed of properly designed their CPU for something different than just a home user OC'er, then maybe you should realize you are not the world.
A lack of understanding why a choice was made does not mean the choice was wrong.

There is a lot that Intel should be ashamed of, but this is really not one.

But congrats on your temp drop. Now try in prime95 with AVX enabled :D OCCT not he most stressfull cpu load
 
SvenBent said:
If you still think Intel needs to be ashamed of properly designed their CPU for something different than just a home user OC'er, then maybe you should realize you are not the world.
A lack of understanding why a choice was made does not mean the choice was wrong.

There is a lot that Intel should be ashamed of, but this is really not one.

But congrats on your temp drop. Now try in prime95 with AVX enabled :D OCCT not he most stressfull cpu load
Click to expand...
Wasn't suggesting it was 24/7 stable. Just a reference point which would have failed hard before.

I also disagree that Intel SHOULD be ashamed of themselves for cheaping out on the TIM and application/assembly of these cpus. IF AMD would be pulling this the Intelphiles would be understandably all over that low hanging fruit. I have no allegiances to any brand. They don't pay me, I pay them. And IMO this is a slap in the face by intel to the enthusiast. Never said I was the world, but Intel markets the K series cpus to enthusiasts at an ADDED price/profit. Further, THIS particular cpu IMO was a defective application of TIM, given how temps were going up up up over the week. Granted that was oc'd at lower voltage, but still, not something Intel should be proud of.
 
RPGWiZaRD said:
This makes me seriously consider delidding my 8600K. I'm all aircooled and temps are pretty good for air but I've got like 3x120 mm front intake and 3x120mm out at top and 1x140 mm at the back.

Temps usually 80~85C in Prime95 non AVX load (with AVX ~95C) and for example gaming usually max 75C (typically 70-75C after hr or two of gaming now in the warm summertime with my absolutely disgusting room temp with no AC, during Winter it was some 7-8C lower even but it's cold in Finland :p).

BUT it's such a dud overclocker tho, I'm currently running 4.75GHz 1.31v, I couldn't get 4.8GHz fully stable at 1.33v or so even and during summer I think ~1.32v is roughly the max limit aircooling is capable of running these chips anyway without delidding at the very least. I wonder if you think there's chances I'd see an improvement in overclock besides temperature though if delidding, I mean I'd be supper happy to get 4.9GHz stable but yea 4.8~4.85GHz thereabouts I guess should at least be possible from delid.
Click to expand...

From the handful of cpus I've delided. I can usually tell with a rough guess how much temp drop I'll get by watching the SPIKE temps when first starting a stress program like OCCT/Prime. What I mean by spike temps, is the temps that the cpu will spike to briefly for a split second or so when the cpu hits a large/heavy load. For instance. If your idle temps are say mid 30's, OCCT loads up/and running and you get say 60's or low 70's temp, but then you may see a spike temp. 80-85C etc.... The larger the delta between the spike temp and baseline temp, the more difference I've seen with delidding. Also, watching individual core temps during OCCT can show if 1-2 cores are way higher than the other's, another sign a delid may benefit you.

That is assuming your cooling is ok, your TIM application between cpu and cooler is ok, you're not going insane on vcore etc..... One reason I like OCCT is I will see those spike temps faster than prime as prime loads are somewhat variable, some threads being more intensive than others. I'll generally see those spike temps on OCCT in a few minutes or less even. Also, if you get high spike temps like >80C within a minute or two, that's another sign I've seen that benefit a delid. At least in my experience. I will say only ONE of my delids did any better with OC'ing though. But I run an AIO and keep voltage generally under 1.37v so that's part of it too. FWIW anyway!
 
JNavy89GT said:
Wasn't suggesting it was 24/7 stable. Just a reference point which would have failed hard before.

I also disagree that Intel SHOULD be ashamed of themselves for cheaping out on the TIM and application/assembly of these cpus. IF AMD would be pulling this the Intelphiles would be understandably all over that low hanging fruit. I have no allegiances to any brand. They don't pay me, I pay them. And IMO this is a slap in the face by intel to the enthusiast. Never said I was the world, but Intel markets the K series cpus to enthusiasts at an ADDED price/profit. Further, THIS particular cpu IMO was a defective application of TIM, given how temps were going up up up over the week. Granted that was oc'd at lower voltage, but still, not something Intel should be proud of.
Click to expand...

Why? This is a limited-edition re-brand of the 8700k, not the new core revision with advertised new paste like Devil's Canyon. They created Devil's Canyon because the paste on Haswell was dropping overclock by 2-300Mhz, erasing the IPC advantage over Sandy.

On Coffee people are achieving within 100Mhz of what the core is capable of with a good water cooler. Delidding Coffee just gives you that last 100Mhz, and the comfort of lower core temps.

You knew what the fuck you were buying in your moment of weakness, and you already own an 8700k so you know how shitty the cooling is when you're shooting for orbit.

You want that last fucking 100 MHz, you're going to have to earn it cowboy.
 
Last edited:
If you don't mind me asking, what is usually the advised TIM alternative these days. Would for example Thermal Grizzly Conductonaut between IHS & CPU die + Thermal Grizzly Kryonaut be a decent working solution or would Conductonaut in between both provide significantly better result (3C?). I have used Coollaboratory Liquid Ultra with with tower heatsinks with mixed opinions (I find it loses quite a lot of performance over time and once I accidentally dropped slight bit of the liquid metal over the mobo parts and took ages to get all that cleaned up) so I'm a little so-so about using liquid metal between IHS and the heatsink if there's some non-liquid metal option with better longevity and less risky application and within 2~3C or so directly after application.
 
Last edited:
RPGWiZaRD said:
If you don't mind me asking, what is usually the advised TIM alternative these days. Would for example Thermal Grizzly Conductonaut between IHS & CPU die + Thermal Grizzly Kryonaut be a decent working solution or would Conductonaut in between both provide significantly better result (3C?). I have used Coollaboratory Liquid Ultra with with tower heatsinks with mixed opinions (I find it loses quite a lot of performance over time and once I accidentally dropped slight bit of the liquid metal over the mobo parts and took ages to get all that cleaned up) so I'm a little so-so about using liquid metal between IHS and the heatsink if there's some non-liquid metal option with better longevity and less risky application and within 2~3C or so directly after application.
Click to expand...

Will never use liquid metal between Ihs and heatsink. But why would there be any issues between ihs and die? It’s not gona move there unlike TIM and has greater attributes. There’s no risk in applying it either. Just use the tape around die and peel it off when done. Tons of videos on YouTube.
 
Raendor said:
Will never use liquid metal between Ihs and heatsink. But why would there be any issues between ihs and die? It’s not gona move there unlike TIM and has greater attributes. There’s no risk in applying it either. Just use the tape around die and peel it off when done. Tons of videos on YouTube.
Click to expand...
if theres no risk why do you need the tape? ;)
 
defaultluser said:
Why? This is a limited-edition re-brand of the 8700k, not the new core revision with advertised new paste like Devil's Canyon. They created Devil's Canyon because the paste on Haswell was dropping overclock by 2-300Mhz, erasing the IPC advantage over Sandy.

On Coffee people are achieving within 100Mhz of what the core is capable of with a good water cooler. Delidding Coffee just gives you that last 100Mhz, and the comfort of lower core temps.

You knew what the fuck you were buying in your moment of weakness, and you already own an 8700k so you know how shitty the cooling is when you're shooting for orbit.

You want that last fucking 100 MHz, you're going to have to earn it cowboy.
Click to expand...

Why the anger bro?

I see your point, but your missing mine. Yeah this is the same "core" as the 8700k. They should be ashamed of the TIM on all their K series CPU's is one of my points. Plus the 8086k is a "binned" version of the 8700k which should imply the silicon is cherry picked for better clocks/thermals etc.... I am sure intel did that. What they didn't do was a good job with the TIM application on this CPU, and if you read my post, you would see that temps were getting worse over time. IMO the application was too thick and the paste was cooking down and losing effectiveness. Or is that normal in your book? Or think I'm FOS. I was just venting with this thread. Wasn't trying to set the world on fire with some new topic. Just feel Intel has gotten really LAZY and needed to be called out for it a bit.
 
PS, whomever asked about prime95 AVX. Running it now. 40+ mins in and still going. I know that's not enough for the stable crowd, but it's what I got so far. Temps are peaking at 80,81,81,82,83, and 84C across the cores when the wattage reported through coretemp is 250+(I wanna say it was up to 274w). Temps are low 70's when reported cpu wattage is around 200, and in the low 60's when wattage <170. I think unless I get a custom loop or drag my old Prometia unit out, those are about as good as temps will get ;). Goal achieved of 5ghz, provided testing continues to go well.
 
Jnavy,
As an Intel shareholder, I thank you for purchasing the same processor twice (8700K and 8086K).
Much appreciate the extra 500 revenue for the 200 mhz gain you received.

;)
 
Well I am an Intel and AMD shareholder lol. So yeah! I fell yah.

But I’m the end it’s a compromise of time vs. $. $$$$ and lack of patience won this time
 
Do any of these mainstream chips come soldered? or are all of them just glued togather... any way to know?
 
While I agree with the idea that Intel should build its CPU's with the best heatspreader design, I think you guys are missing the point. Intel's cost cutting measures still allow their CPU's to perform as they should. Intel doesn't give two squirts of piss about overclocking nor do they care about the enthusiast. They throw us a bone occasionally but they are out of touch with the DIY and enthusiast markets.

This is why Intel throws skulls on server or workstation parts and calls them "gaming" parts. It's out of touch with its most vocal, yet smallest group of customers. It's as simple as that. Also keep in mind that Intel's CPU's are still vastly outclocking AMD's parts even with a shitty TIM. There is virtually no chance Intel is going to change its behavior on this issue.
 
Last edited:
It probably could have, though, for a special edition chip that is absolutely geared towards enthusiasts.

Even a shit parent still calls on Christmas.
 
I agree that pasted IHS temps are absurdly ridiculous. I'd be OK with a pasted IHS on clock-locked processors. But, using that shit on processors that are designed to be unlocked and overclocked? I think Intel missed a golden opportunity for more profit since Ivy Bridge: charge $5-10 more for $0.05 worth of solder.

When I held out for IB after *every* Bulldozer review results let me down, I kept my fingers crossed that the reduction in TDP for IB vs SB would make things fine and dandy with the move to a pasted TIM. When I got my system built and running, then applied a modest OC of 4.4 GHz, the load temps left me like:

FFFFFFF LOL.gif




Meanwhile, I'm still trucking along with an un-overclocked 3770K sporting a dual 140mm fan Megahalems on it to keep load temps comfortable, and each day is a step closer to ditching Intel and going back to AMD once Zen2 comes out, as long as AMD keeps using solder...
 
ZeqOBpf6 said:
It probably could have, though, for a special edition chip that is absolutely geared towards enthusiasts.

Even a shit parent still calls on Christmas.
Click to expand...

I would be willing to pay $10-$20 more for a CPU that's got a soldered heat spreader for OC'ing. I think "K" and "X" models should come this way as they tend to cost slightly more than their less enthusiast oriented counterparts.

As for the second comment, I disagree. The shit parents I know don't call on Christmas. (Mine certainly don't.)
 
So I opted not to delid the 8086k. With a Kracken X52 at 100% in an NCASE M1 case I got the chip 24/7 Prime 95 AVX and Blender stable at 1.33 Vcore at 5GHZ all cores with no AVX offset. I had to limit the TDP to 150 watts though due to temps.

Anything above that I need to Delid or get better cooling.

It’s an impressive chip, but if your serious about OCing it to the max delidding is a must.
 
Dan_D said:
I would be willing to pay $10-$20 more for a CPU that's got a soldered heat spreader for OC'ing. I think "K" and "X" models should come this way as they tend to cost slightly more than their less enthusiast oriented counterparts.

As for the second comment, I disagree. The shit parents I know don't call on Christmas. (Mine certainly don't.)
Click to expand...
It's really sort of amazing that intel still doesn't solder their high end chips when their competition is. I agree all K series should but how does Intel not do it on a $2000 i9-7980XE that draws a ton of power especially when overclocked? It's almost disrespectful to enthusiasts. Looking like that will change next generation at least.
 
Hell even if NOT soldered, they could at least get the TIM right. Mine was >20C difference after delid and WOULD overheat with a MILD overclock on an AIO cooler. Proof in my eyes their TIM application process doesn't really go through a QC process after application. I'm sure all binning etc.. is probably even before the die is applied to the cpu PCB muchless the IHS applied.
 
Do you need to reseal if your CPU is going to be sitting vertically? I'd install it with the case on its side.
 
evilpaul said:
Do you need to reseal if your CPU is going to be sitting vertically? I'd install it with the case on its side.
Click to expand...

Nah, mounting pressure is more than sufficient. The main reason to reseal is if you're going to be moving the CPU around unmounted, like for example if you're planning to ship it somewhere.
 
Sancus said:
Nah, mounting pressure is more than sufficient. The main reason to reseal is if you're going to be moving the CPU around unmounted, like for example if you're planning to ship it somewhere.
Click to expand...

But won't the liquid metal tend to flow due to gravity if the CPU is mounted vertically? I've always wondered about this myself.
 
You must log in or register to reply here.
Back
Top