Lapped 9980XE Kingpin IHS: Unexpected Results
Ойындар
We benchmarked the nearly perfectly lapped Intel i9-9980XE that KINGPIN helped us lap. Results are in one way unexpected, and in another way expected.
Ad: Buy be quiet!'s Straight Power 650W (geni.us/NZ8szyS on Amazon)
Watch part 1 here: • Perfect CPU Lapping w/...
Article is pending as we're traveling -- sorry!
Lapping the i9-9980XE IHS was made easy with Kingpin's lab setup. The biggest point of improvement is in surface contact between LN2 pots and the IHS for a flatter, thinner layer of paste with reduced risk of cracking in extreme cold. Additional benefits, like reduced thermals on ambient, were also tested in our lab. We further tried to plot any improvement in core-to-core deltas as a result of lapping the IHS, ultimately looking to determine if it's worth it to lap a CPU for "normal" use.
We have a new GN store: store.gamersnexus.net/
Like our content? Please consider becoming our Patron to support us: / gamersnexus
** Please like, comment, and subscribe for more! **
Follow us in these locations for more gaming and hardware updates:
t: / gamersnexus
f: / gamersnexus
w: www.gamersnexus.net/
Editorial: Steve Burke
Video: Andrew Coleman
Links to Amazon and Newegg are typically monetized on our channel (affiliate links) and may return a commission of sales to us from the retailer. This is unrelated to the product manufacturer. Any advertisements or sponsorships are disclosed within the video ("this video is brought to you by") and above the fold in the description. We do not ever produce paid content or "sponsored content" (meaning that the content is our idea and is not funded externally aside from whatever ad placement is in the beginning) and we do not ever charge manufacturers for coverage.
Пікірлер: 447
Watch part 1 here: kzread.info/dash/bejne/m4ecxalyYciyn8Y.html Article will have to wait as we are in Canada right now to roast Linus!
@allen-simpson
5 жыл бұрын
Oh I bet that'll be delicious.
@mountedpatrolman
5 жыл бұрын
Just delid the damn 9980XE
@darthkarl99
5 жыл бұрын
@@mountedpatrolman Der8auer wreaked a 9980XE that way, the way the components are mounted on the 9980XE the lid tends to rip of micro capacitors and other little things as it comes off.
@mountedpatrolman
5 жыл бұрын
@@darthkarl99I saw the video, He also did one successfully...
@MrDanielSolano007
5 жыл бұрын
CNC SURFACE GRIND that cpu. Off course not with your hands, it is impossible to put the right amount of pressure on that thing
Sorry to tell you this Steve, but I don't think you're the first to benchmark Kingpin's lap 😘
@GamersNexus
5 жыл бұрын
YOU should be at the roast!
@GamingMafia_id
5 жыл бұрын
I want Kingpin to benchmark my lap instead. Wait...
@taipeitaiwan10
5 жыл бұрын
Yes, it's rumored Paul sat there first.
@AdnanCucak
5 жыл бұрын
Savage
@steven-qk1sg
5 жыл бұрын
@Gaming Mafia lmao
reentry vehicle heat density: ~1MW/m² 4.6GHz 7980XE heat density: 0.808MW/m² we're almost there
@oreolamp5676
5 жыл бұрын
Plebestrian Lmfao From what kind of orbit is that re enter from
@plebestrian9323
5 жыл бұрын
pulled the number from a paper about reentry vehicle heat shields, don't think they specified
@SoylentGamer
4 жыл бұрын
Comet Lake will do it.
@Begleitkubus
4 жыл бұрын
@@oreolamp5676 www.faa.gov/about/office_org/headquarters_offices/avs/offices/aam/cami/library/online_libraries/aerospace_medicine/tutorial/media/iii.4.1.7_returning_from_space.pdf I think this is a paper similar to the one that @Plebestrian read. Scroll down to 4.1.7-322 and note the graph with reentry speeds in relation to heating rate. According to this the 484mm² die of the 18-core chips at 500W power draw equals the PEAK heat energy output of a ~9600m/s reentry speed to earth. This means we are already exceeding the heat energy density that the space shuttle surface had to endure. 9600m/s is ~21500mph, the typical space-shuttle reentry speed is stated at 17500mph.
@daedalus6433
3 жыл бұрын
@@Begleitkubus Silicon density, bruh
Good timing, just got finished watching der8auers video on lapping
@janaebert3059
5 жыл бұрын
What? You can fap 2 times? Wow
@-eMpTy-
5 жыл бұрын
@@nyunster Yea, I watch a lot of hardware KZreadrs =)
@tmi1234567
5 жыл бұрын
I just lapped my i5-4690k and dear god it was bad. It was terrible after 5 years of use. I had a Noctua NH-D15s and was hitting 90-100c under 4.3GHZ @ 1.35v (prime 95 small fft) and was not good. My new Ryzen 7 2700 running under a 140w load (small fft @4 GHz) runs at like 70c. sooo... the I5 got a lot lower temps after the lap. Lapping is not that hard. I went 120 (Dry rest are wet), 400, 800, 1500 then 3000 grits. Took me about an hour and a half to do properly. I am just glad that my cpu has good enough temps now.
@tmi1234567
5 жыл бұрын
I had mt 4690k on a hyper 212 evo and was overheating at 4.4GHz while gaming so I upgraded to a Noctua 120mm static focused fan with a small temp change. Got a Noctua NH-D15s and got a small temp difference. Took a 120mm Delta Electronics high rpm fan and dropped it on the d15 and still didn't see a huge temp change. Lapping gave me like a 20c drop on the Hyper 212 evo with the Noctua fan on it.
@Slavolko
5 жыл бұрын
@@tmi1234567 Good work! I had a 4670k and 4690k, so it's cool to see others pushing that hardware.
NZXT gives you pucks, cooler master gives you cups
I desperately wanted you to not mention the reeses and just casually eat one or two of them throughout the vid and act like nothing had changed.
@johnpainter8090
5 жыл бұрын
seriously man!!!!
@ICanDoThatToo2
5 жыл бұрын
Little known fact -- the mod mat has a place just for reeses.
Holy smokes that bromance
The blood gives the CPU a extra 200mhz. The ultimate cpu is it be part of the cooler, not a seperate peice.
@suit1337
5 жыл бұрын
Use virgin blood harvested at moonlight you get another 200 MHz
@soapa4279
5 жыл бұрын
The human eye can only see 144mhz of blood
@3Balala3
5 жыл бұрын
20 years ago there was no ihs on amd cpus
@NorAlienEuqcab
5 жыл бұрын
3Balala3 I still have a few of them 😁
@taiiat0
5 жыл бұрын
Actually, i saw not too long ago that integrating cooling into the Silicon is being heavily researched and has prototypes, just about ready to be possible to be used in Processors. I do wonder what pressure liquid channels that small are gonna have to run at, though.
I expected a bigger difference from the lapping, good to know it's less than I expected. Thanks for making the video!
I came for the cat. Stayed for the test numbers; and the cat.
man i was looking for this video after watching the first one where u guys were sanding down the CPU. i couldn't find it but suddenly its recommended to me nice!!
You've got my vote on this one. I'd love to see you dig deeper and find the real issue with this chip. To me, this is interesting, I love troubleshooting and I love seeing pros troubleshoot stuff I'll never get a chance to.
0:10 *experienced lapping inspector doing quality control on Steve's work*
These rabbit holes are interesting. I'm not into XOC, but I do find these minutiae details fascinating.
Shaving off 3°C is huge... imagine what effort you have to do with anything else to get this result.
Been waiting for this! Nice.
Best review ever, the cat definitely increased the cooling efficiency with unseen blackmeowgic
i love how steve has so much access to great hardware that he now refers to liquid cooling as air cooling, because the concept of a heatsink and a fan as the only cooling is simply no longer on the table. like... at all. from a steve, good work as always.
I love watching these mad geniuses doing this cutting edge over clocking!.....For Science!
The holy brofecta: tech jesus, der8auer, and kinpin
@givemeajackson
5 жыл бұрын
you forgot main character hair buildzoid
@Kaelygon
5 жыл бұрын
the holy trinity*
@richmeisterradio
5 жыл бұрын
@@givemeajackson buildzoid is like their holy prohpet that explains the inner workings of things.
i have no idea what he is talking about half the time (well most of the time) but he does make it interesting. In Steve we continue to trust!
Great work as usual. I think the lapping of 7980XE is a must to conclude this comparison n study. Would be interesting to see final result. I lapped and delidded my CPU to get best result but this may influence future decisions
Dropped my 10980xe about 8 degrees.... Vince is the goat. I wish they made a new xe. Like a 13980xs
I need those Reese's Cups
Keep looking into it!
we had a polishing station at our university lab for taking a look at metal structure (granular structure) under the microscope. basically a large LP discplayer and an arm which held the metal puck/tab down with a weight. as grinding material there was a diamond powder liquid poured onto a silklike disc. i built a polishing station for rings and stuff out of an old (vinyl) discplayer with 1000grid wet sandpaper.
The flatness to avoid cracking is simple in that any thermal shrinkage or expansion is also more uniform. This thermal expansion and shrinking produces stress in the material. While in a liquid state this is not so much an issue other than potential low level pressurization, but as a solid having different section properties (i.e. thickness) the thermal stress in the thicker sections will increase the stress upon the thinner sections causing cracks within the thinner sections or somewhere in the interface between the two. This same thing happens in concrete a lot and has to be accounted for in design.
I care! I like that it shows new isn't always better. also that following things with closer eye on how they are made should be apart of your deciding process. you guys doing the testing no one else does is what makes you guys you!! its the reason that group of us that cares watches even while watching other channels because you guys review that tech stuff the others don't think about. I mean you got the freaken kingpin in the mix. that shows the quality you care about bringing us. it was you guys ( don't care what anyone else says) that made aire flow important again because people were starting to forget about it.
I enjoy the methodological thinking of potential causality and variables that go along with these videos. Not like I would ever delid/lap my 8600K myself for my basic bitch Scyth Mugen 5 PCGH cooler...
Interesting stuff for sure! It's not a mainstream issue we are solving here, but the reward of modding your own components to yield better results is so.. so.. rewarding. I understand if you have your eyes on LN2 solutions at the moment, as it might be your "next step" in overclocking. Most of us, I believe, consider LN2 cooling videos as entertainment - whereas these types of "hacks" are the real value that might take us (the viewers) to our next level. Maybe this works great on some CPUs and little worse on others etc? Who knows, but you mentioned there are degrees to gain. I am prepared to try just to feed the 12-year old g33k inside of me!
You make such simple concepts sound so overcomplicated.
I think the blessing of Snowflake has had a bigger impact on temperatures than the lap
0:26 smiley ihs
Maybe you can add direct-die tests ? Would be interesting to see differences in this case too :)
Wait so whats the story with Cooler Master and the candy?
@hasinzafir3906
5 жыл бұрын
they sent it to steve as an april fools joke
@USSMariner
5 жыл бұрын
The only way it could have been better if it was the dark chocolate version
G'day Steve, I will never own any of these Mega expensive CPU's but still find this content really interesting, especially to see that the per core delta didn't change with the lapping on the 9980XE, I have also taken more of an interest in the thermals of my own PC's since watching your channel & even had a bit of a play mildly overclocking my old Core 2 Quad Q9650 but need a better cooler (stock Intel) & case (only has a 92mm exhaust) to get some better airflow
@radicalthunder5740
Жыл бұрын
Big whoop
A perfect lap and gold foil, very thin gold foil btw. You should try it once.
Perfect timing for your sponsored add. just under 30 sec.
i use 1200 grit wet with denatured alcohol taped to a small mirror on a flat surface for the lid and whatever heat sink surface i am using, lapping in a figure 8 motion by hand to a mirror shine, i have been able to reduced oc temps by about 5 degrees with some aio liquid coolers
Would any of the imperfections on the inside surface of the IHS make any difference on the 9980XE? Like would more solder between the die and IHS in certain spots possibly create the Core to core deltas you're seeing?
ohh a late night upload. NICE!
We do care, please follow down this path!
NGL I thought this was about Snowflake sitting on Kingpin's lap and reviewing its comfort.
@tiffanyparrott6799
5 жыл бұрын
Well now I need this to be a thing
I wonder how well Reese's chocolate will do as a thermal paste. I hear Gamers Nexus had some stock come in recently.
Please give us more... I'd definitely like to see what the "best" thermal mod for running an 9980XE on water would be.
EVGA RTX 2080 TI Kingpin edition PCB on AMD's display for their Vega (7?) GPU... Could it be?
I remember lapping my pentium 4 EE socket 478 and I got a 7 to 9 Celsius temp drop on air cooled zalman heat sink.
I was wondering what Senior AMD Analyst (aka The Real Editor-in-Chief) Snowflake thought of this. Thank you, Snowflake, for providing the raw, real journalism this industry needs.
Really interesting stuff as always. Maybe you've worked enough on the 7980xe, I would just wait and focus on the next Ryzen 3k and Intel 10nm.
Steve, that huge delta is retained even when using a lapped stock (or an aftermarket copper IHS) on my delidded LM 7980XE. I still have cores that are a minimum of 20C higher than the coolest cores. One of my HWBot teammates is using direct-die on his 7960X and I believe that is the only way to lower (as you've stated) the delta between cores.
Perhaps you guys could explore the idea of using the CPU's heat spreader as the base of the LN2 pot - ie you just have a tube attached to the top of it, removing the thermal paste and the need for lapping. The low temperatures wouild make materials used for seals a bit brittle, but I think it could be done with precision machining of the heat spreader and pot walls, perhaps on a lathe for example.
I care! Great content! What CPU cooler do you use? Could that be a variable in providing/not providing an adequately flat mating surface? I remember back in the Q6600 days people lapping the thermalright ultra 120 extreme as it was convex from the factory.
it's here!
I haven't lapped anything since around 3000 series era. I remember back then getting pretty decent improvements and more stable temps on both Intel and AMD. I also remember fairly well that those earlier chips domed up pretty bad instead of concaving in which could lead to a cooler being mounted at a very small cant (2-5⁰)
Steve, I seem to remember that Roland did delid a 99xxX after heating it to 250+ C in order to melt the solder so he didn't kill the CPU (like when he tried to delid without heating the CPU) and he found that the substrate was thicker on the 99xx CPUs and he had to take off about 15 microns of substrate in order to get good temps (the substrate was acting as an insulator).
I haven't really eaten candy in years but you've really made me want some Reese's Cups.
8:42 the confidence
@vffa
3 жыл бұрын
Right?
Maybe revisit the topic of heat spreader flatness with your fancy flatness test equipment.
I would love to see the i9-99890XE delid and then test to see if you use LM if the core to core deltas would shrink.
You dont need to lap the cooler as well to see an improvement for flattening the cpu?
The IHS looks like it has a smiley face after the sanding... ... this is seriously the funniest Gamers Nexus video by far.
Were the Reece's cups in the case a rip on the Gigabyte Aorus giveaways from over the weekend?
What was the grease used to block the vent? And how did they remove it afterwards? I've hand lapped every CPU I've ever owned with significant improvement, but none have ever had a vent. I'll be going with a 5950x shortly and this may have a vent port so. Thanks for the help
Hey In Cooler master Q500L review please check what gpu coolers (msi gaming x for example :P ) fits with psu mounted close to the bottom Thanks in advance :)
is it possible that the core to core deltas are being created by the cores positions in the die? e.g. cores in the middle of the die hotter than cores towards the edge. Or maybe something to do with proximity to memory controllers, north bridge, etc...
for the fear of voiding my cpu warranty i actually lapped my cpu block as it had an extreme convex surface. managed to drop 10 deg c from my average temps and went from 87c to 65c under load (3 hours of gaming) needless to say i did not feel the need to go onto the cpu... perhaps this could be a video topic for the future; cpu heatsink perfection...
When you say the core to core delta goes down with direct die is that at a similar delta from ambient? Since you are cooling better all around the core to core delta could go down just because you are cooling everything better. I honestly feel like at least a fair portion of the delta has to be just due to the actual silicon and also the layout of the cores. Also it would be interesting to see if any of these things change which core is the hottest. I suspect the hot cores are staying the same through delidding and lapping, but you didn't really say.
Very confused... 17:50 about 5 seconds from here is when I expected Steve to explain what disease that case has, and how to avoid it!
@k3salieri
5 жыл бұрын
Have you or a loved one died from Reeses Case Disease? You may be entitled to compensation.
What if you were to sand the bottom of the ihs? That would reduce the z hieght gap between the die and underside of the hear spreader would it not? Im not sure how much of a tolerance there is between the two but would that not do more in theory?
That easter pc full of chocolate what a nice idee Steve !
Have a request here, review the old school thermaltake big typhoon air cooler from back in the mid 2000's would love to see how it would do on modern cpus I suspect it was way ahead of its time, just a thought :-)
What about lapping the cooler plate as well? Wouldn’t theoretically lapping them together be the best? (As long as you keep the same orientation)
Despite the minimal gains from lapping IHS, I am still considering doing it for my 9900k even though it wont be put under LN2. I do care about this content so please do more testing and send CPU to K|ngp|n. Thanks
One thing I realized on my 7900X is that the hottest cores were the first, last, and middle 2 (4 out of 10). They were consistent, with different cooling solutions, water blocks, thermal paste, and even with the stock IHS vs copper IHS. Now, I don't know if those are "accurate" representations of the physical layout, but it almost seems to me that, if consistent, they could be "center cores", or at least cores in more "dense areas". I mean, this is all speculation, as I've tried getting core layout info from Intel and it doesn't seem to be publicly available. Just food for thought… :)
I watched both your adds today. I never do that but you're awesome so... you know how it is hehe
I think we're all eager to find out the delta-T over ambient of K|ngp|n's lap.
What if you liquid cool with a home ac chilling the coolant like LTT would that be cold enough to cause "cracking" in the thermal paste? I have a 9900ks with a copper IHS that I got that way was thinking of lapping it when I get all my "liquid cooling" parts but was curious if the risk was worth it. And and wanted to know what I could/might expect
might be interesting to find out if lapping to the finish of a gauge block and wringing the pot and the IHS might eliminate the need for thermal paste
In der8auer's newest video he explains how IHS' warp as the solder cools and sucks the center down leaving the corners high.
You kind of answered my question from the last video of why the surfaces aren't polished here to create traction to thermal paste. Still curious though if maybe a direct copper to copper contact wouldn't be better. With a perfect polished surface you wouldn't need an interface material. There is a company that makes a laser dial indicator as an add-on for their laser calibration system. It could be used to make sure there are no variations or any concave or convex surfaces. Obviously not practical for the home overclocker but for a person that is going all out I think it would be the thing to do. With an instrument like that you could even test for variations coming from contraction under the nitrogen. Due to differences in thickness of the contact materials, the rim thickness of the spreader/block vs the center thickness could cause the metal to shrink under nitrogen temps creating a gap.
Maybe I missed it. But what was the practical joke from Cooler Master?
Uhhhhh....the candy?!?!?!?! The whole reason I watched the damn video!
Would you consider lapping the CPU die to improve core to core temperature delta's? I did this very lightly on a 8086K and saw core temperature delta's drop from 10c to around 5-6c. The 8086K die's are small and with such a smaller core count compared to the 9980XE for eg i'd imagine there would be much more to gain from doing this with a CPU with a larger die and many more cores.
The core- core temperature difference isn't just from the layout of the cores physically? Like, aren't they in a grid, so some are on a corner and some are in the middle surrounded by other cores, I'd assume the corner ones are a bit cooler.
Interest registered. I would also be curious to see a lapped cooler as well just to make sure everything is making the best contact.
Hey man I think you got a reese's peanutbutter cup in your PC fan! I can almost smell the yum.
Since you have the 7980xe de-lidded I'd be interested in more testing with the heat spreader. Not only lapping but testing the inside for consistent thickness. I'd also be interested in seeing if different thicknesses would affect the performance and if different profiles would have any impact. For example, feathering the heat spreader thickness towards the edge both thinner and thicker. Also, once lapped and of uniform thickness, gradually remove a layer of a fixed percentage of the starting thickness. So, if the lapped and even heat spreader is 2mm then remove 5% or 0.1mm. Though typing that out it seems like it would require multiple heat spreaders and a ton of work.
@VeritasEtAequitas
6 ай бұрын
In short, no.
I know this might sound crazy to you all, but what do you all think about using a 3D printer to build some kind of multi-layered multi CPU socket buss so multiple MPUs could be simultaneously cooled and run in parallel to share the current like use to be done with audio amps with a S-load of power transistors back before MOSFETs took over?
I am wondering, why LN cooling teams use all this sponge, tape and so on to cover entire motherboard (to prevent condensation) - instead of using just a closed case filled with a completely dry gas, like pure nitrogen? No humidity, no condensation. You can make such tank with acrylic sheets or smth like this. It even does not have to be pressure tight, just rubber around all cables, rubber seal around the LN pot to fit it in the top cover. You get rid of humid air, you do not risk condensation on spots of the MB that you missed. You can even provide some kind of a loop of cool nitrogen gas, to provide additional cooling to VRMs and RAM. Is there some kind of a trick that would make it fail?
How do you think the IHS gets a curve to it? Certainly they are milled flat initially on both the underside and top-side. I would bet that the interface is not an issue as much as the underside of the IHS being curved to match the top, warped during nickel plating or uneven deposition of material I would guess. It may be possible to lap both the underside of the IHS and the die itself to reduce core-core deltas. Lastly I am surprised that kingpin does not have a jig to hold the IHS to lap perfectly flat rather lapping it by hand on his machine.
Guess someone with an accurate CNC could cut off a fraction of a millimetre to even the surface instead of grinding.
Silicone cupcake pan and brownie mix, push Reese's into center after baking, demold after refrigerating pan over night.
@PixelGod240
5 жыл бұрын
Don’t for get the green butter ;)
So I have a set of high end ceramic stones for sharpening knives and equipment to keep them flat and I'm wondering... Can I just use these and make sure I keep them super flat? I have leather strops with different compunds I use also that will put a literal mirror polish on about anything. Kinda wanna try it on my 9900k but it costed me more than I really should have spent anyways lol.
@prescott231233
5 жыл бұрын
It should work, but don’t wet stone it, unless you use very high percentage isopropyl alcohol. Also, I’m not sure what grit your tools go to, but the higher grit won’t work as well and will just polish the lid. You know that though. It would be cool to see a super shiny cpu though!
It might be the Kraken X62 mounting, not getting good pressure. With good watercooling solution with good spring mounting I think the results will be better, but it really might be the solder underneath being what it is. And come on GN, who the heck will use 9980XE with Asetek mainstream AIO, at least put the 360 MLC Phoenix.
more importantly, whatchu gonna do with all those reese's?
@tiffanyparrott6799
5 жыл бұрын
Somebody suggested Steve eating all of them while keeping eye contact with the camera and not saying a word the entire time
@adamblomberg
5 жыл бұрын
Selling them to the Chinese
@mjc0961
5 жыл бұрын
@@piers389 The bin is the only valid place for you and your comment to go.
@piers389
5 жыл бұрын
@@mjc0961 You're wrong - it's the only place for your country to go.
@user-yv2cz8oj1k
4 жыл бұрын
What is the lethal dose of peanuts? 🤣
Nice info update on lapping. Btw. Did you receive a copywrite infringement using Oda Mae and Molly scene? 😂
Did y’all ever finish the coolermaster snacks?
did he ever explain the reese's all over the desk yet? i keep missing it
I would assume the high core to core delta is simply because of the number and the thermal density and number of cores. Where a higher overclock makes it harder to cool the center cores compared to the outer ones. This is not as much an issue on the 8 core parts ore when running at stock. Ore are other people seeing smaller delta's at the same voltage / frequency / workload?
Lapping isn't the be and end all for mating the surfaces - just because they're flat doesn't mean they're parallel, etc. If you want much better contact for LN2 then instead of grinding them both on stones you should be mounting the cooler and CPU, bluing them and then scraping the two surfaces into each other. That will take actual physical contact up from single figure digits into the low teens.