Friday, October 19th 2018

Intel Core i9-9900K De-lidded, Soldered TIM Outperformed by Liquid Metal

We kept seeing hints regarding Intel's 9000-series processors running hot, including from their own board partners. As it turned out, the actual results are a mixed bag with some running very hot and most others ending up being power-limited more so than temperature-limited. Our own review sample showed overall better load temperatures relative to the predecessor 8000-series processors thanks to the soldered TIM (sTIM) used here, to give you some context. But that did not stop overclocker extraordinaire Roman "Der8auer" Hartung from de-lidding the processor to see why they were not generally better as expected.

As it turns out, there are a few things involved here. For one, replacing sTIM with Thermal Grizzly Conductonaut (Der8auer has a financial interest in the company, but he does disclose it publicly) alone improves p95 average load temperatures across all eight cores by ~9 °C. This is to be expected given that the liquid metal has a vastly higher thermal conductivity than the various sTIM compositions used in the industry. Of more interest, however, is that both the PCB and the die are thicker with the Core i9-9900K compared to the Core i7-8700K, and lapping the die to reduce thickness by a few microns also does a lot to lower the CPU temperatures relatively. Overall, Intel have still done a good job using sTIM- especially compared to how it was before- but the current state of things means that we have a slightly better stock product with little scope for improvement within easy means to the consumer.

Add your own comment

75 Comments on Intel Core i9-9900K De-lidded, Soldered TIM Outperformed by Liquid Metal

#26
Vya Domus
qubitThe savings they make from not using the best are miniscule
I doubt they care about that. At this point I am convinced they are doing on purpose to artificially limit their CPUs for one reason or another.
Posted on Reply
#27
Vayra86
qubitIntel spend millions on developing these processors, yet they can't give them the best thermal solution possible.

The savings they make from not using the best are miniscule and the processors are quite expensive, so cost can't be a factor. I just don't get it.
And yet... that is not what I am seeing.

I will not say TIM or soldered TIM is the 'best' solution, but both solutions are fine. We have virtually NO stories of Intel CPUs getting fried. Thát is a metric Intel uses. Not the wishes of some wannabe CPU experts on a forum.

What I ám seeing: Intel needs to push Core harder and harder to keep up on the current node. The thermal issues with these CPUs only started with Kaby Lake's 7700K. And only with a ton of vCore on it. Kaby Lake is also the first gen that comes with notably higher base and turbo clocks for these CPUs. There is only one possible outcome: these CPUs will be running hotter

Then came Coffee Lake, with increased core counts ánd increased base and turbo clocks. Now, Intel clocks them almost to cap right out of the box. The headroom is shrinking rapidly, and again, the only possible outcome is a hotter CPU. And then, 9th adds another few hundred mhz and 2 cores. Boy, I wonder what'll happen.

I don't see artifical limitations from Intel. I'm seeing an architecture that is stretched to the max, with higher quality (core count) parts moving lower down the stack and clocked higher out of the box. Intel needs bigger, more functional CPUs from a wafer and the requirements for quality go up. The variance is as much in the thermal solution as it is in the quality of the CPU itself, simply because all parts are stressed.
Posted on Reply
#28
E-curbi
Vayra86And yet... that is not what I am seeing.

I will not say TIM or soldered TIM is the 'best' solution, but both solutions are fine. We have virtually NO stories of Intel CPUs getting fried. Thát is a metric Intel uses. Not the wishes of some wannabe CPU experts on a forum.

What I ám seeing: Intel needs to push Core harder and harder to keep up on the current node. The thermal issues with these CPUs only started with Kaby Lake's 7700K. And only with a ton of vCore on it. Kaby Lake is also the first gen that comes with notably higher base and turbo clocks for these CPUs. There is only one possible outcome: these CPUs will be running hotter

Then came Coffee Lake, with increased core counts ánd increased base and turbo clocks. Now, Intel clocks them almost to cap right out of the box. The headroom is shrinking rapidly, and again, the only possible outcome is a hotter CPU. And then, 9th adds another few hundred mhz and 2 cores. Boy, I wonder what'll happen.

I don't see artifical limitations from Intel. I'm seeing an architecture that is stretched to the max, with higher quality (core count) parts moving lower down the stack and clocked higher out of the box. Intel needs bigger, more functional CPUs from a wafer and the requirements for quality go up. The variance is as much in the thermal solution as it is in the quality of the CPU itself, simply because all parts are stressed.
Dood, you can't get your 8700K to stabilize any higher than 4.8Ghz? Are you only setting volts and Multiplier? I can help if you want. Have an 8700K overclocking bios configuration in my build log (signature below). Copy and Paste it and try it out - works pretty well.

Sorry, I like checking out specs - seeing what others are running. :)

There's a new historical statistics chart from SL went up like two days ago.

Here's the link:

siliconlottery.com/pages/statistics
Posted on Reply
#29
Vayra86
E-curbiDood, you can't get your 8700K to stabilize any higher than 4.8Ghz?

Sorry, I like checking out specs - seeing what others are running. :)

Here's a new historical statistics chart from SL. Put's things into perspective.

siliconlottery.com/pages/statistics
Cute. 4.8 Ghz @ 1.31V. Not the 1.375V and up that you find under your link.

I thought you liked reading, you suck at it, apparently :D

Which ironically is also exactly what I am seeing when it comes to these topics. Very selective cognitive abilities. Unable to process that 1.4V across 6 cores will suffer from heat. Unable to realize that this heat is happening on ever smaller spaces as the nodes get smaller... and so on.
Posted on Reply
#30
E-curbi
Vayra86Cute. 4.8 Ghz @ 1.31V. Not the 1.375V and up that you find under your link.

I thought you liked reading, you suck at it, apparently :D

Which ironically is also exactly what I am seeing when it comes to these topics. Very selective cognitive abilities. Unable to process that 1.4V across 6 cores will suffer from heat. Unable to realize that this heat is happening on ever smaller spaces as the nodes get smaller... and so on.
Sorry, I thought you were trying to find out what your CPU was capable of.

Not getting into a "your volts are too high" argument. lol

Never question a member of the forum tribal elders. Reverse and cyclical logic will always prove him correct. Pardon me. :p
Posted on Reply
#31
trog100
E-curbiSorry, I thought you were trying to find out what your CPU was capable of.

Not getting into a "your volts are too high" argument. lol

Never question a member of the forum tribal elders. Reverse and cyclical logic will always prove him correct. Pardon me. :p
my 8700K runs fine at 5 g maybe higher i havnt tried.. but it just runs too hot.. right on the edge of throttling.. a re-lid with liquid metal would bring it down so i run it at 4.8 g one day if i get really bored i might do it but at the moment i just cant be arsed..

but to me soldered lid or not the 9900K is doing exactly what i would expect it to.. running f-cking hot when all 8 cores are working.. as for the soldered lid.. intel did this because they had to.. this chip is at its limits heat wise.. just like my 8700K chip is.. end of the line.. he he..

it also got bugger all to do with bad soldering those that think it is are just away with the fairies.. :)

trog
Posted on Reply
#32
ArbitraryAffection
Vayra86Why is the 9900K so hot? - der8auer

Because his silicon cash cow is rapidly drying up now with soldered solutions that really are pointless to improve upon.

Wake up people.


Little scope for improvement for the consumer. Let's see how overclocking is doing in 2018 besides Intel, shall we.

- Nvidia GPU: clocks right to cap out of the box. BIOS locked down tight, temperature limited.
- AMD CPU: overclocking can be detrimental to performance, and barely nets gains in any case.
- AMD GPU: they can still overclock, but don't have much more than 10% left in the can. High end GPUs need to be downvolted instead.
- Intel CPU 8th Gen and previous: being clocked closer to their cap and straight into 'our' OC headroom. Not just on K CPUs either. Delid on 8th gen and earlier would provide lower temps, but barely improved clock potential.

Gosh, its almost a trend, isn't it...
Overclocking my 2600 to 4.1 from 3.5 all core made a significant improvement in performance.

Truth is, 9900K is once again a poorly made product. And people are paying a premium of twice as much for that 16%~ higher FPS at 720p and a Processor that can double as a heating element for a blast furnace^^
Posted on Reply
#33
R-T-B
ArbitraryAffectionTruth is, 9900K is once again a poorly made product.
Even the article states that the sTIM isn't anything subpar, so I wouldn't go that far.
Posted on Reply
#34
Mescalamba
Issue isnt solder, but really thick layer (of copper?) between actual silicon and TIM (and heatspreader). Most likely to avoid any crumbling of CPU itself. Best solution atm is probably to delid, sand it down, cut heatspreader and keep only frame of it (slightly thicker than CPU itself), apply some TIM on CPU directly and carefully assemble with cooler directly in contact with CPU. That layer of copper applied on CPU is thick enough to act as normal heatspreader anyway.

Only issue is fragility (maybe, since PCB and copper is thick it might be actually fine, within reason).
Posted on Reply
#35
Vayra86
E-curbiSorry, I thought you were trying to find out what your CPU was capable of.

Not getting into a "your volts are too high" argument. lol

Never question a member of the forum tribal elders. Reverse and cyclical logic will always prove him correct. Pardon me. :p
Nah, your post was a clear stab at a previous comment without providing any substance. And this post here is another one. Get on topic, I'd say.
ArbitraryAffectionOverclocking my 2600 to 4.1 from 3.5 all core made a significant improvement in performance.

Truth is, 9900K is once again a poorly made product. And people are paying a premium of twice as much for that 16%~ higher FPS at 720p and a Processor that can double as a heating element for a blast furnace^^
You can debate whether or not Intel should clock these CPUs so high out of the box, but the facts remain, they are in safe ranges even under a nice OC. Are they temperature limited before voltage limited? Certainly. But there is no competitor that is doing a better job at either overclocking or clocking. So we can complain all day about 'bad product' but no one has invented anything better.
Posted on Reply
#36
E-curbi
Vayra86Nah, your post was a clear stab at a previous comment without providing any substance. And this post here is another one. Get on topic, I'd say.
Honestly, it wasn’t a stab at any previous post. I don’t spend enough time here to analyze posts or trends or personas.

I was only jumping in and trying to help.

Making the assumption (maybe I'm wrong) that no one would voluntarily run an 8700K at 4.8Ghz as a daily – unless they were stuck and needed a simple overclocking guide to move forward.

I only thought these 8700K overclocking steps below might get you to 5.0Ghz or beyond, depending on your chip.

Peace brother! :)


1-Ai Overclock Tuner to XMP, then select NO to all core enhancement in the Notice window. Apply by saving out of bios and reboot back into bios and make sure your memory is running at your individual memory's XMP settings, in Asus bios that information is located at the right side information bar

2-Asus Multicore Enhancement to Disable

3-AVX Instruction Core Ratio Negative Offset to 3

4-CPU Core Ratio to Sync All Cores and 1-Core Ratio Limit Multiplier to 50 (for the first 5.0GHz overclock attempt). Scroll down and all 6 Cores Ratio Limit will change to 50

5-CPU SVID Support to Disable

6-Drop into External Digi+ Power Control (opens a new window) Set CPU Load-line Calibration to Der8auer says 5 or 6, I used 5. (5 seemed like a more conservative number than 6 LOL)

7-ESC out of Ext Digi+ Pwr Cntl back to the Extreme Tweaker window, select Internal Digi+ Power Control this time (opens a new window) and set Long Duration Power Limit to 55555 (4095) and Short Duration Power Limit to 55555 (4095)

8-ESC out of Int Digi+ Pwr Cntl back into the Extreme Tweaker window and set CPU Core/Cache Current Limit Max. to 55555 (255.50)

9-Min CPU Cache Ratio to 42 and Max CPU Cache Ratio to 42

10-BCLK Aware Adaptive Voltage set to Disable

11-CPU Core/Cache Voltage to Manual, and for CPU Core Voltage Overide type in the value 1.350volts for your 8700K/8086K that should be good for the first 5.0Ghz overclocking run.

12-Go into Asus Overclocking Profile and save your settings in Profile 1 (if you wish). Then APPLY all these settings SAVE OUT of bios and boot into Windows
Posted on Reply
#38
E-curbi
Rahmat Sofyanintel should hire Roman ..
Intel cannot afford to pay - The Farmer. :p

Der8auer translated to English.
Posted on Reply
#39
TheinsanegamerN
trog100genuine competition means that both parties now run their chips at the speeds they are capable of as opposed to one team cruising with plenty of overclocking headroom and the other team struggling to keep up..

also adding two more cores plus an increase in clock speed isnt easy its bound to cause heat problems when all cores are working hard..

soldering the lid on has helped a bit but not by much.. the bottom line being that the days of good problem free overclocks are now over on the top end chips.. maybe that is how things should be..

the 9600K is a better option for those that want to save some dosh and have all (six) cores at 5 gig or a tad over.. two cores less will make all the difference..

the fact these things only boost fully on one or two cores tells the true story and has done for a while..

trog
But why is it that intel's 8 core can hit 225-250 watts of power usage on an 8 core chip, sometimes pushing 300, when AMD's 8 core at 4.3 is only pushing 140 watts?

I know that power consumption is exponential past a certain point, but given intel's ability to clock in the past, such high power consumption is really off, and that high usage would explain the high temps. 250 watt from such a small die is going to be a PITA to cool no matter what.
Posted on Reply
#40
oxidized
R-T-BMore like liquid metal always has outperformed solder
This.
Liquid metal obviously conducts better than solid metal, in every situation
Posted on Reply
#41
hat
Enthusiast
TheinsanegamerNBut why is it that intel's 8 core can hit 225-250 watts of power usage on an 8 core chip, sometimes pushing 300, when AMD's 8 core at 4.3 is only pushing 140 watts?

I know that power consumption is exponential past a certain point, but given intel's ability to clock in the past, such high power consumption is really off, and that high usage would explain the high temps. 250 watt from such a small die is going to be a PITA to cool no matter what.
It likely has to do with Intel's monolithic design vs AMD's CCX design, as well as the processes used to make them. IIRC, AMD is actually using a process designed for low power, efficient chips. It's not really something you would want to make strong desktop chips out of, but it works well enough. Intel is likely using the appropriate high performance (but high power) process for these chips.
Posted on Reply
#42
trog100
TheinsanegamerNBut why is it that intel's 8 core can hit 225-250 watts of power usage on an 8 core chip, sometimes pushing 300, when AMD's 8 core at 4.3 is only pushing 140 watts?

I know that power consumption is exponential past a certain point, but given intel's ability to clock in the past, such high power consumption is really off, and that high usage would explain the high temps. 250 watt from such a small die is going to be a PITA to cool no matter what.
it all comes down to where you start from.. cruising speed or near flat out speed.. plus i dont think amd chips clock much higher no matter what the voltage gets put through them they just trip over their own feet long before they get to super high watts..

intel chips will clock higher they just hit a heat barrier.. its end of the line..

trog
Posted on Reply
#43
jmcosta
oxidizedThis.
Liquid metal obviously conducts better than solid metal, in every situation
yeah, its the best for performance but they have disadvantages, for having a short longevity, being conductive and corrosive,
for example the Liquid Ultra is gallium based and will destroy aluminum surfaces but most of these disadvantages are no problem for entusiasts .
Posted on Reply
#44
R-T-B
jmcostafor having a short longevity
Your other points are valid but there is no shortness of longevity on liquid metal. It's a metallic compound, it will literally last decades.
Posted on Reply
#45
hat
Enthusiast
oxidizedThis.
Liquid metal obviously conducts better than solid metal, in every situation
Uhh... isn't solder also "liquid metal" for a brief time, until it cools? Or does the physical state of matter actually... matter?
Posted on Reply
#46
R-T-B
hatUhh... isn't solder also "liquid metal" for a brief time, until it cools? Or does the physical state of matter actually... matter?
Well part of the definition of a liquid is it conforms to the shape of it's container. So naturally better contact is to be expected.

That said, yes, you could make the solder liquid too... at a very hot temp that would wreck your cpu...
Posted on Reply
#47
Vayra86
E-curbiHonestly, it wasn’t a stab at any previous post. I don’t spend enough time here to analyze posts or trends or personas.

I was only jumping in and trying to help.

Making the assumption (maybe I'm wrong) that no one would voluntarily run an 8700K at 4.8Ghz as a daily – unless they were stuck and needed a simple overclocking guide to move forward.

I only thought these 8700K overclocking steps below might get you to 5.0Ghz or beyond, depending on your chip.

Peace brother! :)


1-Ai Overclock Tuner to XMP, then select NO to all core enhancement in the Notice window. Apply by saving out of bios and reboot back into bios and make sure your memory is running at your individual memory's XMP settings, in Asus bios that information is located at the right side information bar

2-Asus Multicore Enhancement to Disable

3-AVX Instruction Core Ratio Negative Offset to 3

4-CPU Core Ratio to Sync All Cores and 1-Core Ratio Limit Multiplier to 50 (for the first 5.0GHz overclock attempt). Scroll down and all 6 Cores Ratio Limit will change to 50

5-CPU SVID Support to Disable

6-Drop into External Digi+ Power Control (opens a new window) Set CPU Load-line Calibration to Der8auer says 5 or 6, I used 5. (5 seemed like a more conservative number than 6 LOL)

7-ESC out of Ext Digi+ Pwr Cntl back to the Extreme Tweaker window, select Internal Digi+ Power Control this time (opens a new window) and set Long Duration Power Limit to 55555 (4095) and Short Duration Power Limit to 55555 (4095)

8-ESC out of Int Digi+ Pwr Cntl back into the Extreme Tweaker window and set CPU Core/Cache Current Limit Max. to 55555 (255.50)

9-Min CPU Cache Ratio to 42 and Max CPU Cache Ratio to 42

10-BCLK Aware Adaptive Voltage set to Disable

11-CPU Core/Cache Voltage to Manual, and for CPU Core Voltage Overide type in the value 1.350volts for your 8700K/8086K that should be good for the first 5.0Ghz overclocking run.

12-Go into Asus Overclocking Profile and save your settings in Profile 1 (if you wish). Then APPLY all these settings SAVE OUT of bios and boot into Windows
Thanks, I'll try finding those Asus settings on my AsRock board. Strike three of not reading... Maybe you should just stop. Every assumption you make is a major misfire. If you can't or don't want to take time to read, don't take time to reply either.
trog100it all comes down to where you start from.. cruising speed or near flat out speed.. plus i dont think amd chips clock much higher no matter what the voltage gets put through them they just trip over their own feet long before they get to super high watts..

intel chips will clock higher they just hit a heat barrier.. its end of the line..

trog
Exactly. It used to be common practice to take 'one step back' after finding your max OC. Now we complain that we need to do so, and if we don't we complain about exponentially growing requirements... Pretty weird to me.

And it gets even more strange when you consider 'we'd rather not use AVX' because that'd cost us 100 mhz as well. So fuck stability, 5 Ghz or teary eyes. The entitlement generation in a nutshell - oblivious to circumstance and 'must have what others have'. And when that somehow doesn't work out, the world is to blame. No need to look at your own OC, blame Intel and everyone else :D
Posted on Reply
#48
trog100
mass produced things always have a built in safety margin.. that is your "step back".. a good figure to avoid too many returns and give good reliability would be 10%..

but in the strange world of the red and green teams the 10% does not apply.. the winning team artificially clocks its products down to just keep in front of the team that is struggling to keep up.. here we get the good overclock which isnt really an overclock its just running the product at a speed its really capable of..

amd are still struggling to keep up but on the more core front they are in front.. more slower speed cores at a good price.. they have rattled intels cage enough to push them into the more cores race.. the snag is adding more cores whilst still keeping the higher core speed simply generates more heat.. they cheat by only boosting one or two cores to the claimed boost speed but over clockers want all cores running at the same higher speed.. the end result is pretty obvious the f-cking things get too hot..

all we are seeing here is intel are no longer clocking down (cruising) just to stay in front they are now having to try hard to stay in front.. the days of easy 30% overlcocks on intel chips are well and truly over.. it kind of reminds me of the old P4 days.. :)

there are pros and cons.. on the one hand we are getting an honest performing product out of the box but on the other hand there is no false overclock on the top end chips.. unless one buys the 9600K.. which isnt the top end chip but for gamers and overclockers at the new higher price levels its the one to buy.. he he..

for gamers and overclockers the 9900K is a fail..

trog

ps.. the winning team under clocking their products also has the advantage of not making the losing team too silly.. making them look to silly makes them try harder which in the end ups the anti for both teams.. nether team wants that.. its all a bit of a con and always has been..
Posted on Reply
#49
Fabio Bologna
I would be curious to see how it performs with something like that NudeCNC Direct die water cooler...
Posted on Reply
#50
jmcosta
R-T-BYour other points are valid but there is no shortness of longevity on liquid metal. It's a metallic compound, it will literally last decades.
while that is true, from my experience i have been forced to reapply triennially my cpus.
This is because the liquid damages any surface, it is less reactive against nickel plating but it does degrade over time, the same for copper that absorbs it.
but again for enthusiasts, is no big deal since people change their hardware every 2-3years.
Posted on Reply
Add your own comment
Nov 21st, 2024 13:38 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts