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

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

#1
hat
Enthusiast
Intel forgot how to solder?
Posted on Reply
#3
Chloe Price
"hat said:
Intel forgot how to solder?
It's because the die is thicker if I understand it right. The die logic is on bottom of the die, and thicker die makes it harder to transfer the heat away.
Posted on Reply
#4
R-T-B
"hat said:
Intel forgot how to solder?
More like liquid metal always has outperformed solder, if only marginally.

They do appear to be having more quality control issues than I would like though...
Posted on Reply
#5
Zubasa
"R-T-B said:
More like liquid metal always has outperformed solder, if only marginally.

They do appear to be having more quality control issues than I would like though...
Thus all the BS about how solder is natually bad because micro-fractures etc.
Looks like there might already be air gaps in the solder in the first place.
Posted on Reply
#6
Xzibit
"Chloe Price said:
It's because the die is thicker if I understand it right. The die logic is on bottom of the die, and thicker die makes it harder to transfer the heat away.
Intel try'n to be Thicc

Roman is just body die shaming. -6c just by liposuction (0.2mm). -14c in total
Posted on Reply
#7
kastriot
Let me guess, you tried prime with avx instruction which is unrealistic so this nonsense with liquid tim is total waste of time except for milking people to do delid and -9C is marginal -15-20C would be acceptable so yeah.
Posted on Reply
#8
Zubasa
"kastriot said:
Let me guess, you tried prime with avx instruction which is unrealistic so this nonsense with liquid tim is total waste of time except for milking people to do delid and -9C is marginal -15-20C would be acceptable so yeah.
The thing is Intel themselves is the ones pushing hard on AVX instructions because it is fastest on their own chips.
So Intel definitely intends their chips to be able to run AVX instructions.
Posted on Reply
#9
ppn
Only need to lap the die or and delid for chasing record scores. For everyday use and resale value is counter productive to destroy the cpu. Now Up to 0.5mm is dummy layer. 178mm2 die area 37Mtr/mm2 so expect this to be shrinked to 66mm2 on 10nm and 100Mtr/mm2. Of course it doesn't translate to perfect scaling but the 10nm chip would be much much smaller. perhaps you didn't ask for it, dont flatter your self, the dummy layer is there for better cooling of 10nm. Remember this chip was not supposed to release on 14nm, AMD forced it.
Posted on Reply
#10
Tsukiyomi91
everyday use, soldered TIM does the job already. Don't need to waste your time & money in those delidding business unless one knows the risk of doing so in the pursuit of those silly world record.
Posted on Reply
#11
15th Warlock
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.
Everyone and their dog knows this, how is this front page news? :confused:
Posted on Reply
#12
Tsukiyomi91
@15th Warlock probably a "new thing" that Intel has been keeping that stuff in the dark & only brings it to light speficially for this new SKU??
Posted on Reply
#13
R0H1T
"15th Warlock said:
Everyone and their dog knows this, how is this front page news? :confused:
What's noteworthy is how the die & the PCB are extra thick, Roman speculated that it might be for the extra large (air) coolers that this chip obviously needs.
The extra thickness is what's probably causing excess heat buildup.
Posted on Reply
#15
Vayra86
Why 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...
Posted on Reply
#16
delshay
1 hour to grind down the die, far too long for me. where did I put my angle grinder. All jokes aside, is it worth it for around 6c from delidded to lowering the height of the die.
Posted on Reply
#17
trog100
genuine 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
Posted on Reply
#19
R0H1T
"HD64G said:
Since I see big differences in power draw and temps between reviews (the TPU one has resulted in the low end of both), I can only assume that the bios settings of the different motherboards tested have to do with it. Some power limiters are on or off probably and change the results drastically.

An example below that show that clearly.

https://www.hardocp.com/article/2018/10/19/intel_core_i99900k_9th_generation_cpu_review/5


There's also different workloads for power consumption numbers showing up on different review sites, the PL2 limit is exceeded in AVX heavy stress tests. That's a fact, so max power consumption, at it's worst, is 220W or thereabouts.
Posted on Reply
#20
Basard
I saw in the video that he put pressure down on the IHS as he was resoldering it. He probably should have resoldered without putting any pressure down. Perhaps that, combined with the glue scraped away, would have helped avoid the spiked temps on certain cores due to cracking solder.:confused:

I kinda saw this coming anyways. Didn't wanna be that guy crying about how hard its going to be to delid--seeing as we all wanted solder in the first place. They definitely could have done a better job of it though.
Posted on Reply
#21
lexluthermiester
"Basard said:
He probably should have resoldered without putting any pressure down.
Going to have to agree with this as solder does not like being pulled apart, especially when it's that thinly applied.
Posted on Reply
#22
Durvelle27
Maybe next batches will be much better
Posted on Reply
#23
ppn
"Vayra86 said:
Why 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.

Gosh, its almost a trend, isn't it...
They justify delidding and lapping the die for +200MHz, or 4% overclock is aimed at some enthusiasts. I guess For normal use, no small ftt, it will rarely heat above 65C. This guy takes absolutely the heaviest load, small ftt or else.
Posted on Reply
#24
qubit
Overclocked quantum bit
Intel 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.
Posted on Reply
#25
E-curbi
OH the IRONY is so thick on this one. :roll:

I've been laughing non-stop for 16hours. :roll:
Posted on Reply
Add your own comment