AMD Threadripper CPU Lapping & LN2 Prep, ft. Bearded Hardware & 3970X

We're lapping an AMD Threadripper CPU, the 3970X, with Joe from Bearded Hardware. We'll be streaming a 3990X liquid nitrogen overclock on Sunday, 2/9 at 1PM EST.
Sponsor: Buy Gigabyte's X570 Master (Amazon - geni.us/WJgNK)
The AMD Threadripper 3970X needs to have its nickel plating removed for better contact with overclocking, but we also need to prepare for our 3990X XOC livestream (and later AMD Threadripper 3990X review). In this video, we're working with Joe Stepongzi to learn about AMD TR CPU lapping and to sand down our Hyper Beast LN2 container.
Subscribe to Joe here: / @beardedhardware
The best way to support our work is through our store: store.gamersnexus.net/
Like our content? Please consider becoming our Patron to support us: / gamersnexus
** Please like, comment, and subscribe for more! **
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.
Follow us in these locations for more gaming and hardware updates:
t: / gamersnexus
f: / gamersnexus
w: www.gamersnexus.net/
Host: Steve Burke
Guest: Joe Stepongzi
Video: Andrew Coleman

[@GamersNexus]

Who do you think sent us the 3990X? To be super clear, this was a 3970X CPU (for other streams). We just received a loaner Threadripper 3990X in the mail from another KZbinr and will be prepping today. The 3990X stream will be SUNDAY (tomorrow) at 1PM EST, so that's about 23 hours from time of this comment! We will stream on YT. Find Bearded Hardware here: kzbin.info/door/HcGSnu7Izq231nAS5zFmsw

[@b_0rk]

How nice of Wendell

[@einze2085]

My guess is Hardware unboxed. Love snowflake and your content as always.

[@luckythewolf2856]

Who else but Linus

[@Code7Unltd]

I take it that's Linus?

[@JhanSolo332]

Could be Jay? He hasn't done a review yet

[@eyetvideos467]

If you combine Joe and Steve you get Hagrid from Harry Potter 😂

[@robmaertens]

Bahahhaaha!

[@Dr01nE]

Tech Jesus + Tech Odin = Tech Hagrid 🤔

[@sambobsam]

100% true lol

[@raiyananindo1371]

Gosh you're so damn right 😂

[@chownzi2461]

"Everything looks tiny in my hand" I died of laughter

[@chrish1850]

The rotary polisher is not a good way to lap anything unless you have a jig that will hold the chip completely parallel to the surface. As trying to hold anything by hand you will apply uneven pressure and will inevitably put facets in your chip surface. You also need a fine enough abrasive to provide a smooth/flat surface.. I work in an industry where companies are lapping silicon wafer chips to witihin thousands of a micron... Send me one of your chips and ill lap it down to 50 nanometers (0.05 micron) across the surface (no charge) and will send it back for you to test. The surface will be so smooth it will look like a mirror when you look into it.

[@imacrazyguy5831]

@@davidhoward2237 Dont need a granite plate for lapping.. just a sheet of tempered glass will be just fine as a surface.

[@David-li1uc]

@GamersNexus at least send an old cpu to this guy to test it out , Bench it before with all your technique and after his Mirror finish, could be very interesting for us viewers!

[@morbid1.]

oh man I like this challenge :D

[@Rizon1985]

@@David-li1uc Flat is flat even if it's not parallel to the PCB. Heatsinks are screwed in on all corners so it would still be flat against flat There's also absolutely no way to test that difference because reseating is going to change the difference by multitudes more than a different lapping method would (unless you did your best to lap a convex surface on the IHS). Software also has too few digits to see the difference and an external meter would be placed too imprecise to get consistent tests.

[@David-li1uc]

@@Rizon1985 Well as flat as can be he pretend he can flaten to a whole new level where figure8 sanding can still introduce 'due to hand sanding' a convex surface " i have no measurements to compare Steve technique to His but that could be interesting even the measuring" , i do agree on the rad mount that it can tilt swivel and follow the fact that the sanding migth not be paralel to the socket it might achieve a truer flat , it should be nice to check direct die sanding and most importantly Rad sanding . i hope to see this soon :)

[@Minitomate]

"You can have a fun project with this, even if you void your warranty." The fun begins by being able to afford it.

[@GamersNexus]

I mean, sand paper is like $3. You can lap any CPU with an IHS.

[@Minitomate]

@@GamersNexus I was referring to the CPU, but even tho it sounds like a good experiment to do.

[@cm01]

When I upgrade from an i3-2100 to a 3000G soon, I'll both delid and lap the i3-2100. Just for practice. Then I'll do the same to the 3000G. By the time I can afford a nice processor I'll know what I'm doing and it'll cost almost nothing.

[@crazyjak56]

@@cm01 You can't get a used 2200g?

[@cm01]

@@crazyjak56 meh, sure I could

[@drwho900]

It's like having a teaser before the big one tomorrow!

[@DaveyKTech]

I just enjoy the chemistry you two have together. The chemistry is very evident and that makes these videos even more fun to watch.

[@bobmellon]

Dude I love your grasp of the word delta, your vids are awesome

[@ahorseinahospital]

Joe is such a cool guy, always a pleasure to watch your collab vids/streams

[@HMarcBower]

Always enjoy these collaboration events between you two. :)

[@MarioCRO]

Love the two of you talking it out, I can only imagine the outtakes, they must bi hillarious. Great vid and as always, really interesting and educational.

[@DrakkarCalethiel]

Can't wait for the LN2 OC stream!

[@neon_necromunda]

These two have such good banter between each other that id watch a once a month special on a regular or more. I like a bit of banter.

[@Leo9ine]

Notification gang! You're the best, Steve. And Patrick and everyone else.

[@witnesszer0]

Lap in a figure 8 motion you know to add more infinity fabric

[@vicviper319]

Seeing Joe back at GN HQ made my day! Looking forward to the stream :)

[@or2kr]

I didn't know how much I missed this lol

[@SateenDuraLuxe]

If you use a lapping plate that's flat within a few millionths of an inch, +/- .000003", technically you wouldn't need any thermal paste between your cpu and cooler. Gage blocks are some of the most perfectly flat objects you can get and if you slide them against each other they'll stick. kzbin.info/www/bejne/aJ2ygJ9pi92Cqqc

[@Outland9000]

Yes... I remember when I first got my hands on a set of gauge blocks... I was like _"What is this voodoo?"_

[@XavierAncarno]

Indeed they will stick to each other like magnet. Perhaps they should invest into a calibrated granite table

[@sinuosalice-1434]

True.... But for this purpose the actual difference between an insanely smooth and very smooth is neglectable probably, you might get more results by eliminating the shield and going die contact.... Or maybe they can lap the die too eheh

[@LunaStarFire]

@@XavierAncarno please for the love of god do not do this to gauge blocks,, basically the surfaces are so polished that they cold weld to each other so you pressing them into each other and then breaking them off leaves micro splinters on the surface.

[@gcutler15]

@@LunaStarFire this is flat out wrong because the wringing effect that gauge blocks can exhibit still happens with non metallic blocks, ceramic, etc. The blocks don't need to be metal to demonstrate this effect. All you need is a perfectly flat, perfectly smooth, hard surface.

[@Tech-NO-City]

Oh that make cents you guys are funny bunny. Congratulations on the one year anniversary.

[@a59x]

"Yes, We got one"

[@seymoronion8371]

kzbin.info/www/bejne/fImwlJWeo5uBppI

[@a59x]

@@seymoronion8371 kzbin.info/www/bejne/oJTMe2Zvmc-reZY

[@seymoronion8371]

Fair enough XD

[@honeyblade5243]

@@seymoronion8371 I like your response also lol!

[@RonI-qz2tz]

Interesting way to lap. I have always done it with a lapping block and paste. But I have never lapped a CPU. So that was way cool.

[@hughw.]

Fantastic Steve, Joe and Gamers Nexus Team!

[@kennyj4366]

Looking forward to watching it.

[@AlanDike]

I appreciate this.. I just lapped my 1950x and heatkiller iv pro. My cpu was pretty flat.. didn't even make it through the plating everywhere before the whole thing started getting polished on the full surface. The cold plate was really high centered though... I'm on water and I am happy to say I dropped a few degrees at idle... but I feel it's better contact and running even a few degrees at full load. If it wasn't for this video, I'd still be hemming and hawing over doing it.. My copper showed up kind of in the middle first.. and when I was seeing good polishing on the nickle everywhere I stopped...

[@ebsmith1102]

Sanding machine of glory

[@middleclassthrash]

I used to re-manufacture big rig fuel injectors and it's funny to see all these tools of the trade like lapping machines, crc spray, the blue lint free cloths, etc. after years. What a strange overlap.

[@khalidmkhan]

Good chemistry. Invite the bearded one again. Thanks for uploading.

[@MPcustomized]

Always check the lapping finish with shades, great move! 👌😎🤘

[@gazzacroy]

that was fun to watch :)

[@wyattkane]

The nerdiness of this channel pleases me greatly.

[@ramshaka]

Figured out a great solution to procuring an extremely flat surface when I did my lap. Had a piece of sample marble tile custom cut for me at the Home Depot... Works extremely well, and you can get it as small or large as you like, up to 1'x1'...

[@cliftonsr]

This is really cool and hopefully the CPU stays that way

[@Qalibrated]

Glorious hair meets glorious beard. Tech Jesus vs. Tech Odin. I like it! Good stuff

[@ramshaka]

Suggestion for taping your sandpaper down: First lay down a piece of tape on glass, then super glue the paper down to the tape. Maintains your flat surface, holds super strong, and comes back up relatively easily. Also, for holding parts, or anything else, you can use tape on the surface, and the part, with superglue in the middle for an ultra-thin two sided tape...

[@DiscoParachute]

This is my new favorite GN video

[@IMIDelorios]

How nice of Wendell to send you the chip! :)

[@ShokkuZaMonki]

The "Wow" at the beginning is so fucking good. It sounds so sincere it's hilarious.

[@isbestlizard]

I LOVE laser etched text!

[@krugec23]

my brother lapped his fx 6300 years ago,still works today

[@PowerPc94]

I wanted a 3990x on L2N so bad! Thank you, tech Jesus !

[@nanonxpc9677]

freaking hype cant wait to see it under LN2

[@jacobwutmanov198]

Nice Broski!

[@SLEEPYJK]

Joe "everything looks tiny in my hands" Stepongzi

[@rocnroll0065]

There's that slacker Joe!!!!

[@ASHBOT-OLDGUY]

Good to see the two of you laying well together lol WHAT NO TATER TOTS?...

[@BeardedHardware]

Adam Shumann Sr slacker!!

[@shaneeslick]

G'day Steve & Joe, Really Excited for the stream at 5am Monday, Nowra NSW time, I don't know if I will be able to sleep, P.s. I'm enjoying the lovely weather, It is 20ºC (68ºF) & has been 🌧🌧🌧Raining🌧🌧🌧 here for the last 3 days, It is really Awesome for all the wildlife & Fire Fighters after Drought & the Horrendous Bushfires

[@mosamaster]

Waiting with popcorn bag for the Sunday's 3990X Frozen Meltdown. 😁

[@Scouse29K]

I've lapped once before, plan to do it again soon to a 9900K, but I let the CPUs own weight do the work and just try and figure 8 it around the sandpaper with applying little to no pressure, just use my fingers to try and guide it around without pushing down on it at all. Looking forward to the live stream! Will have more ln2 myself Monday morning :)

[@Yoshikaable]

I really like this guy. the linus face just was amazing.

[@TheFirstExelion]

"You can spit on it" -Odin, 2020

[@BeardedHardware]

Exelion you’re welcome hahaha

[@SickPrid3]

tip from working in metallurgy - it doesn't matter which way you move it when sanding, even pressure is the key if you don't have good feeling of how much pressure you apply put some weight on top of sanded surface and just pull/push it around the sanding surface

[@markusstrobl1067]

Yep

[@Archer_Legend]

Heair to bear ratio is excellent!

[@workinprogress5936]

Watching this while playing farming simulator on xbox... life is good

[@HazmanTheGreat]

From a distance, Joe looks like a long bearded Alex Jones in the thumbnail LOL.🤣 Thanks for another great video Steve and Joe.👍

[@startedtech]

He looks nothing like Alex Jones though....

[@MandrakeFernflower]

Big Clive looks different today

[@meghankrajchi4913]

Fantastic! Did you bleed for this one?

[@SilentMechanic01]

Nice to see Joe is still alive 🤣

[@Malignus68]

Why does everyone assume that the nickel plating is the same thickness across the IHS? Is that proven? How do we know that the nickel plating isn't the way that they put a perfectly flat surface on top of the slightly imperfect copper base? It seems to me that, unless the process of nickel plating invariably produces uniform thickness, the fact that copper begins to show through unevenly doesn't necessarily prove that the nickel surface wasn't flat.

[@siclucealucks]

AFAIK they use some galvanic process (same as you would apply chromium on car parts) this process usually has a very even outcome - especially on such flat surfaces.

[@Rizon1985]

​@ISOHaven It does when nickel plating something. It's not like dipping it in paint or something. And they use electroless plating and not the electroplating people do at home. The variance is to the size of an ion because they are always attracted where the least are positioned which for nickel means 0.00000007 mm. The warping happens when placing the nickel plated IHS on the CPU and is often just by design because of accepted tolerances. They could make and sell CPUs much flatter than you'll ever get with pathetic sandpaper (grid 2000 sandpaper only does 0.01mm particles.) but there is no reason to. They absolutely do not give a shit that you buy their CPU because Dell buys 10 000 every month to put in a server rack with 6000 rpm fans blowing on the heatsink and they wouldn't if they added an unproportionate amount of money to get the temperature 1°C down.

[@siclucealucks]

​@ISOHaven 0,1 mm this is 3times as thick as a copper layer on a pcb ..the coating which we are talking about is in the range of a few micrometers - a veriation of 15um to 20um so +/2.5um is way below the flatness/parallelity of the IHS. for further read i can recommend you this www.nickelinstitute.org/media/2323/nph_141015.pdf edit i meant this advancedplatingtech.com/electroless-nickel/electroless-nickel-plating/ ...but the otherone is also good if we are talking about conventional electroplating

[@jefflavenau6805]

if the nickel was perfectly flat, it would wear evenly when abraided on a deadflat surface--they demonstrated here again that it doesnt

[@Rizon1985]

​@@jefflavenau6805 The whole IHS is deformed and that's why you see it disappearing over different spots. If you ever lapped a CPU you would see you're taking of nickel first and afterwards a mix of copper and nickel which wouldn't happen if the copper plate was flat. The deformity happens when placing the IHS on the CPU. The nickel plating happened months before that in a whole other factory. They place an order for a few ton of nickel plated copper sheets and then work with it from supply. And again this is electroless nickel plating where you have a variance of a few tens of picometers. Not the amateur electroplating people do in a tupperware bin.

[@Apollo-Computers]

makes me so sad that i dont have a business address to get ln2 delivered to when i see it flowing on the channel, but excited for the stream at the same time.

[@michaelholub5027]

If you've done this before you'll also note that the flatter the top of the chip is, the more prone it is to hydro-locking on the heatsync. I do on-site OEM Certified repair and there's a big difference from one chip to another. Older "dryer" thermal compound can also make a difference, but often the hydro-lock will force a pinned CPU out of it's socket.

[@ihateevilbill]

Sitting there looking for selotape while the sandpaper has sticky 3M backing. Genius XD

[@luckythewolf2856]

Not gonna lie I thought this was the review for the 3990x when I got the notification.

[@GamersNexus]

We're working on one of those too. The loaner from our YTer friend is making that possible. That'll be a little while, though. Now that embargo has lifted, we're just going to take the time to really work on it and put together some good content rather than rush.

[@luckythewolf2856]

Gamers Nexus your reviews are well worth the wait so I’m sure none of us mind

[@jasongooden917]

Me too

[@Kobrar44]

I'm sure everyone interested is actually in the target market for 3990X.

[@nc0gneto]

@@GamersNexus The little extra Luke over at Kitguru did with the Ice giant cooler was really interesting, I hope you go down that road.

[@pprietojuez]

That slacker viking is awesome!

[@PokemasterSD]

Steve's lushes locs and Joe's wicked beard. Combine both and you have Smooth McGroove.

[@VeniuEN]

I recommend using a dentists articulating paper to check for inaccuracies safely and without a need to sand it off, it can be wiped with alcohol and it highlights all points of contact, it's pretty precise.

[@jong2359]

I used to lap all of my heatsinks when I was younger and cared about maximum cooling performance. The difference was generally measurable from around 2-5ºF reduction under load. The CPU's I worked on were open die (Athlon XP era) so I never really thought of inverting the process and lapping a CPU lid.

[@rogueanuerz]

not delidding ?

[@jong2359]

@@rogueanuerz believe it or not, CPU's didn't have lids back in the day.

[@maxair95]

Steve and Joe looked like 2 metal band dudes doing Pc stuff....

[@Falkor82]

I saw that beautiful Excelsior class starship in the background. You should do a starship PC build.

[@DanielMosey]

I lapped my 2500k, then later my 2700k. It helped lower my over all temp by 2-3c on water. However it helped most by levelling all my core temps. Before there was around 8c difference between the hottest (core 2) and coldest (core 3), now it's a 1c difference.

[@connor040606]

Live stream colabs are fuckin epic!

[@tsunamidelta1351]

You may want to invest in a granite surface plate to do this on the regular. Tempered glass and non-tempered tends to vary in thickness and flatness by .001" to .004" at times. When I was running a wafer saw, we discovered that the small variation was a problem in the cut.

[@jamiermathlin]

If you increased the git on the emery paper, through a number of steps ending in 2000 or even 3000 grit you can achieve an almost perfect surface without visual scratches, this improves the contact still further, its worth a try.

[@nathanzylla4961]

The beard and the hair unite!

[@namizmo]

I was today years old when i realized MSI AM4 motherboards, even ones I couldn't see using it for, have LN2 bios settings to help the CPU boot.

[@LawrenceTimme]

I just stuck mine in the milling machine at work and went over it with an end mill took 0.2mm off 👍

[@Azlehria]

Fixtured on the IHS?

[@JMUDoc]

Somebody needs to do a video where the IHS and heatsink/waterblock are so flat and smooth that they can be wrung together without any paste in between. I don't think that's ever been done and I'd love to see what - if any - difference it makes (especially with sub-zero, as the TIM is the weakest link).

[@crazylarryjr]

Also removing the anodizing, removes a layer of the many layers between the silicon and the pot. Might not be noticable to many, but when dealing with high performance OC's, every degree counts

[@I-HAVE-A-BOMB]

Very good point I didn't think about that, would lead to better heat transfer but itll heat up far faster too.

[@brianterrill9587]

Ben Crow at Crimson Guitar(here) uses a 2 tape method and super glues the tapes together basically creating a 2 sided tape.

[@PvPbomber009]

You guys complement each other soo well with the long hair and beard

[@ares035]

If you combine Steve`s hair and Joe`s beard you got cousin IT!!!! :D

[@aaaaaaaaaassssssssdf]

i think i need to lap my ekwb, the aluminum series is kinda warped from the screws bending it...

[@masooncap9769]

Thank you Linus Tech Tips for loaning GN the 3990X so we can see another great review of that

[@ezg8448]

It's hardware Jesus and hardware Santa working together!

[@smithjones1906]

I watch a lot of machining videos so when they were talking about needing a really flat surface and wanted to use glass, I thought, "why not get a certified granite surface plate?" Then I remembered that the good ones are expensive, frustrating to ship, and have tolerances they probably have zero need for. But it was interesting to think, "how else might they do this job?" and weighing the pros and cons of each method. I think Kingpin's method is pretty damn good, I just wish it didn't include him actually holding the CPU with his hand.

[@centralintelligenceagency9003]

Today, Santa taught me how to lap a CPU.

[@thomaswilliams]

Tech Jesus and Gandalf .... This needs to be the title!

[@cookiemcwilliam5300]

The two surfaces get "rung" together. That means that all or most of the air gets pushed out from between those surfaces. So atmospheric pressure what's to hold them together.

[@666nacirema666]

used to lap all of my cpus and often the coolers.

[@dabj9546]

Is the die actually higher on the edge or is is just a very slight rocking motion that removes the edges faster than the middle?

[@t.k.tronix0815]

I always support it when the two homeless guys are integrated into society

[@vipervidsgamingplus5723]

T.K. Tronix good one

[@DukenukemX]

I actually polish my lapping to get it as mirror like as possible. My idea is that the less imperfections you have the less imperfections the thermal paste will have to fill.

[@jotstony]

You skipped over the part of when you change the sandpaper grit. So, when do you switch and can you sand too much? Is thinner better for the plate? Thanks

[@heyarno]

Ideally you would mate the cpu and cooler together. Just take a look, how telescope mirrors are made. (The good old Dobson video is the easiest to understand in my opinion.) And use the cpu, like the tool glass.

[@personaldronerepair6141]

Hey Steve, do you believe that the H115i Platinum would still have the old style concave cold plate characteristics? I only ask due to the new pump design. Just wondering if the new cold plate would be redesigned to be simply flat. Thanks. You two are having too much fun btw.

[@rounakr94]

Tech Jesus ftw 🤘🏻

[@clickykbd]

The make blue dye specifically for marking or tracking progress in metal work like this. Search for "Dykem Layout Fluid". Will easily show where you have made progress and what is untouched (still blue). More "pro" than sharpie method.

[@BeardedHardware]

clickykbd hahaha nice

[@Azlehria]

The machinist I learned from liked using jumbo dry-erase markers to "blue" parts. Quick, easy, and surprisingly even.

[@pa-pyro2804]

You should try a diamond plate. I use them on my very fine planer blades and are perfectly flat and very fine grit

[@Bibl888]

@Gamers Nexus conCAVE, as in it's caved in so it means the inside part is lower than the outside, it helped me remember :)

[@radicalxedward8047]

Wouldn’t it be more efficient (lapping generally, not for LN2) to totally polish the IHS? The point of thermal paste is filling in micro texture/scratches, so if it was a mirror finish wouldn’t that mean smaller texture to fill?

[@pino_de_vogel]

Should get someone to 3d print a socket with a grip for this.