Lapping a CPU, the proper way [AMD 3950x]

แชร์
ฝัง
  • เผยแพร่เมื่อ 20 พ.ย. 2024

ความคิดเห็น • 305

  • @rapthis
    @rapthis 4 ปีที่แล้ว +116

    Came here after seeing the article on PC Gamer. Very excited to see what comes out of this.

    • @litapd311
      @litapd311 4 ปีที่แล้ว +1

      samesies

    • @nathanphillips575
      @nathanphillips575 4 ปีที่แล้ว

      Were you disappointed?

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +37

      I'd like to thank PC Gamer and its readership for all the positive attention. I understand there's also a bit of frustration that my video ends on a cliffhanger, I assure you that I share your desire for BAR GRAPHS but while I could go out and spend two days holed up in my workshop and publish those juicy numbers for you I did after all sell my CPU and its matching water block to a hardware news and review publication with the implicit understanding that the story was part of the sale. I'm not going to go out and spoil the scoop for a friend.

    • @GeoffRichards_GP
      @GeoffRichards_GP 4 ปีที่แล้ว +5

      was a bit confused by the intro (or lack thereof). It felt like someone had just ripped der8auer and reposted. But then the "real video" starts. I look forward to learning more!

    • @Dunkopf
      @Dunkopf 4 ปีที่แล้ว +1

      Same

  • @OrganicBASSd
    @OrganicBASSd 4 ปีที่แล้ว +79

    Can't wait to see temps without thermal paste. It's going to be a good day.

  • @privatepadilla
    @privatepadilla 4 ปีที่แล้ว +30

    For the love of god...
    ...please start making more videos.
    Your attention to detail and narration voice would make you an instant sensation on TH-cam IMO...
    PC related, machining related, doesn't matter.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +8

      I still have my regular job to keep up with and spending time with my family often takes priority when I'm home so my writing/recording/editing pipeline is always full.

  • @satnav9699
    @satnav9699 4 ปีที่แล้ว +85

    dude you have a great voice, very nice to listen to.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +22

      @X Oh well, can't please everyone.

    • @blunden2
      @blunden2 4 ปีที่แล้ว +13

      Paul over at PC Gamer seems to agree:
      "He partially chronicles the process in a 10-minute video, detailing various steps in an extremely soothing voice. Seriously, I could listen to him read a phone book (remember those?)."
      Source: www.pcgamer.com/a-machinist-lapped-an-amd-ryzen-16-core-cpu-within-a-nanometer-of-its-life/

    • @williambennett9764
      @williambennett9764 4 ปีที่แล้ว +4

      Great voice. Here, have my subscription.

    • @darealmaul
      @darealmaul 4 ปีที่แล้ว +4

      I mean...damn...I mean dudes voice got me feeling like a baby ready to sleep out of comfort. Yeah he can ready me a couple bedtime stories! 👍🏿

    • @2railnation
      @2railnation 4 ปีที่แล้ว +3

      @@Penrowe More suited for a murder story than tricks for overclocking a PC.

  • @bluegizmo1983
    @bluegizmo1983 4 ปีที่แล้ว +37

    That's a freaking thing a beauty right there! Bout time someone did a true lapping job on a CPU! Can't wait to see the results. Even if the temp difference isn't massive, I'd still be happy knowing that incredible lapping job has been done and that the CPU and cooling block can be rung together without paste!

    • @alanfidurski6860
      @alanfidurski6860 4 ปีที่แล้ว +8

      Right?
      Liquid metal might even be overthrown...
      For all we know...
      If this works totally gunna pay him to do a proper lapping job for me.

    • @VndNvwYvvSvv
      @VndNvwYvvSvv 10 หลายเดือนก่อน

      ​@@alanfidurski6860Honeywell PTC already beats in most practical applications. This is neat to see, but you don't actually want it perfectly flat. You want a slight curve, concave IIRC, due to thermal expansion, so it's flatter at load.

  • @manoftherainshorts9075
    @manoftherainshorts9075 8 หลายเดือนก่อน +3

    So, there are no even absolute results for this experiment after all of this? Man, you can't have us hanging like this :'(

  • @NameyNames
    @NameyNames 4 ปีที่แล้ว +101

    Was that "Beve Sturke" I heard, Steve Burke's alter ego? Gamers Nexus!

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +59

      I am not at liberty to disclose such information.

    • @XeoNIvan
      @XeoNIvan 4 ปีที่แล้ว +7

      Yes, thank you, I knew I was right!!! That's definitely Gamers Nexus!

    • @MakubexGB
      @MakubexGB 4 ปีที่แล้ว +16

      @@XeoNIvan When?! When do we get to see part 2?

    • @weignerg
      @weignerg 4 ปีที่แล้ว +3

      9:59 I thought I heard that too.

    • @Outland9000
      @Outland9000 4 ปีที่แล้ว +6

      As soon as he mentioned bar graphs....

  • @FirewaII33
    @FirewaII33 3 ปีที่แล้ว +10

    I've worked in Mc shops for almost a decade and worked as a QC tech. It always killed me when people would "lap" with sandpaper. It's like milling with a lathe. Sure you can get a flat surface, but it's not the same thing! I've never seen an IHS rung with a coldplate and I have to admit I'm pretty excited to see what Beve can show with that. It's so counter intuitive that the lap plate needs to be softer than your work piece, but this has inspired me to learn more about this and makes sense when you think of the charging process, but it's still a little "wait, wut" if you are used to traditional machining where the cutting tool is harder and the cause of material removal. But if you think of the lap plate as a tool holder, and the tool removing material is the abrasive, it becomes a little more intuitive. Great video, thank you for making this!

  • @VladislavGoranov
    @VladislavGoranov 4 ปีที่แล้ว +5

    Wow! So smooth that the Casimir effect is even observed :O

  • @SaifBinAdhed
    @SaifBinAdhed 2 ปีที่แล้ว +2

    I'm still waiting.. the hanger broke, but I fell on the floor safely, but still waiting.

  • @rdyer8764
    @rdyer8764 4 ปีที่แล้ว +16

    Biggest lie you told: "At the end of the day I'm JUST a machinist..." To the contrary, this is great work and your presentation very understated. I'm so glad you're investigating this phenomenon. I've asked other PC-Performance TH-camrs about the effect of wringing a CPU IHS and a cooling block for heat transfer performance, but all I've ever gotten was 'crickets'.
    I just subscribed. Can't wait to check out your other videos.

  • @bosross6992
    @bosross6992 3 ปีที่แล้ว +4

    It's been 8 months. I'm still waiting for the part 2 of this video

    • @N0N0111
      @N0N0111 3 ปีที่แล้ว +3

      Sorry dude something happened, and we won't hear why.

  • @GreybushGamer
    @GreybushGamer 4 ปีที่แล้ว +5

    This was in my recommended. Thank you, youtube.

  • @leifleifnephew5154
    @leifleifnephew5154 4 ปีที่แล้ว +42

    That was weirdly fascinating.
    Please tell me you're making a follow-up to this.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +40

      I can't promise when but there will absolutely be a follow up to this video.

    • @SimianIndustries
      @SimianIndustries 3 ปีที่แล้ว +4

      @@Penrowe 7 months later...very curious to see the results considering I had great success years ago but seems like all of TH-cam says it doesn't work

  • @danielwarme5374
    @danielwarme5374 4 ปีที่แล้ว +8

    Very interesting! I really enjoyed the backround, never thought about precision flatness this way. Looking forward to the next part.

  • @AXHMR
    @AXHMR 4 ปีที่แล้ว +12

    I did this to my P4 and Heatsink back in the day.
    I learnt the hard way that they would become inseparable, ripping the CPU from the socket. When I went to remove the heatsink.

    • @Lan_ny
      @Lan_ny 2 ปีที่แล้ว

      I ripped my A8-3800 out of the socket just from old paste being in the system (6 years not being turned on or changed). Good thing I only was taking it out to try lapping

  • @JMUDoc
    @JMUDoc 4 ปีที่แล้ว +5

    I was silly - I used sandpaper on glass when I should have used my milling machine and cast iron lapping plates.
    Oh well... next time, maybe :)

  • @shi_mo_neta
    @shi_mo_neta 4 ปีที่แล้ว +23

    I can't be the only one who got a braingasm from the Three Plate Method

    • @joshuakuehn
      @joshuakuehn 4 ปีที่แล้ว +3

      The basis of all precision.

    • @ChucksSEADnDEAD
      @ChucksSEADnDEAD 4 ปีที่แล้ว +5

      You wouldn't believe the rabbit hole I had to go into to find that out before youtube machinists/engineering channels started talking about it.

    • @henryong4100
      @henryong4100 4 ปีที่แล้ว +3

      best comment ever...I had the exact same experience

    • @taiiat0
      @taiiat0 4 ปีที่แล้ว +1

      definitely not! one of those 20/20 hindsight situations.

  • @cadibiagio66
    @cadibiagio66 4 ปีที่แล้ว +4

    Tuned in for a brief but interesting cpu modification, got info on the history and science of flatness. Loved it, awesome video

  • @connortremblay1259
    @connortremblay1259 4 ปีที่แล้ว +3

    I saw this video like a week after it was posted and have been anxiously awaiting the results the entire time. The wait is killing me

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +1

      I too am feeling a little restless, been sitting on results from my own testing for a while that I'm looking forward to publish.

    • @SaifBinAdhed
      @SaifBinAdhed 4 ปีที่แล้ว +1

      @@Penrowe so when should we expect anything? I am eating more than normal just because of waiting in anxiety from the result if that makes any sense. lool

  • @MengskTV2
    @MengskTV2 4 ปีที่แล้ว +7

    "Hello, this is Beve Sturke." Sounds like we're about to get some killer content on this!

    • @MengskTV2
      @MengskTV2 4 ปีที่แล้ว +2

      @Nolo Kobo It's Steve Burke's alter ego he used to use to make orders without being recognized by the PC manufacturers. If you like this kind of content, you should check out the GamersNexus TH-cam Channel!

    • @henryong4100
      @henryong4100 4 ปีที่แล้ว +2

      yeeeeeeeeessssss

    • @MengskTV2
      @MengskTV2 4 ปีที่แล้ว

      @Sugarz maybe next time answer the guy's question in the first reply instead of being snarky about someone who doesn't know what you know.

  • @maninarush2112
    @maninarush2112 4 ปีที่แล้ว +2

    I knew about gauge blocks, but damn, that 3 plate method blew my mind.

  • @TechLevelUpOfficial
    @TechLevelUpOfficial 2 ปีที่แล้ว +2

    still waiting for the update, come on man this is super interesting.

  • @Veektohr
    @Veektohr 3 ปีที่แล้ว +5

    Finally, a true lapping vid. It has always bothered me watching people sand dies with obvious edge rounding. I'm looking forward to seeing a follow up. I've always wanted to see somebody toss theirs on a surface grinder too.

  • @TheChodax
    @TheChodax 4 ปีที่แล้ว +11

    Right, where is part 2! You've left me hanging bro! 😁😂

  • @2020Tech4U
    @2020Tech4U 2 ปีที่แล้ว +1

    ok this was great. I said lapping and sanding are different but u know big channels and answering viewers is not a big thing unless it's a mass of ppl. Thank you for this, now I'm off to sand the IHS on my 5900x...Good luck everybody!

  • @MinorKing245
    @MinorKing245 4 ปีที่แล้ว +1

    Oh man! Can't wait for the follow-up! EPIC video! GOOD JOB!

  • @karankshah
    @karankshah 4 ปีที่แล้ว +4

    Absolutely fascinating. Eager to see what the results come back as.

  • @angus3963
    @angus3963 4 ปีที่แล้ว +1

    Eagerly awaiting those graphs, very fascinating work here! 👍🏻

  • @NolePTR
    @NolePTR 4 ปีที่แล้ว +1

    I had to watch a second time because your voice is too soothing it almost put me asleep.

  • @JustifyTheseHeathens
    @JustifyTheseHeathens 4 ปีที่แล้ว +1

    I didn't know what CPU lapping was 15 minutes ago and now I'm all for it.

  • @turboshoturassmerica7269
    @turboshoturassmerica7269 3 ปีที่แล้ว +4

    Amazing video!
    Gosh, this makes me really want to go the extra mile and try no TIM, so I’m a machinist and have done lapping for 15+ yrs, also XOC overclocker, with my water blocks and cpu dies, Direct die 9900k/10900k, I ground the die flat to 2 light bands then lapped it lightly from there, along with water block cold plate, but with water block assembled to avoid the mistake and make everything square to itself, die to PCB, ect. On 9900k chips, took my best 6 I also tested different material removal to get closer to the transistors which did make a significant improvement, and let’s just say the results with both surfaces at 1 light band flat they will wring together, but getting them to wring while assembling cpu/waterblock in motherboard would be an extremely challenging process. Non the less on my golden 9900k I have insane core to core deltas of +/- 1c controlled coolant temp (chiller) and on my golden 10900k +/- 2c using liquid metal. And my temps compared to stock saw significant decrease of around 10c to 25c depending on chips and silicon quality. My 10900k was just going to have IHS lapped but I had core to core delta issues even after lapping IHS, so a delid was needed.
    Only thing stopping me is time, look forward to seeing another video.

  • @nexxusty
    @nexxusty 3 ปีที่แล้ว +2

    This is uncanny.
    I was talking with someone about launching this exact business the other day.
    I too believe in this process, I believe it to be the best and most effective way.

  • @ren7a8ero
    @ren7a8ero 4 ปีที่แล้ว +1

    This is the real deal. 2020, and too few people actually uses the resources available to improve stuff.
    Great!

  • @Laeknishendr
    @Laeknishendr 4 ปีที่แล้ว +1

    Getting serious Ahoy vibes from this video. Great video. Subbed

  • @Teatime4Tom
    @Teatime4Tom 4 ปีที่แล้ว +2

    PC Gamer sent me here. But I subscribed because I must know more about this mystery man, Beve Sturke. I wonder if he means old Steve Burke? I don't know anyone named Beve, but old Steve lives out beyond the Dune Sea.

  • @SLEEPYJK
    @SLEEPYJK 4 ปีที่แล้ว +1

    Man, you've got one hell of a good voice-over voice 👍🏻

  • @LawrenceTimme
    @LawrenceTimme 4 ปีที่แล้ว +1

    Finally a proper lapping video. It's very annoying when people just sand their cpu and say they have lapped it.
    I have used a set a grade 0 blocks. They wring together very well and I have always wondered what would happen if you wring the cpu and cooler together.
    Personally I think once the cpu heats up the thermal expansion will completely ruin all the efforts, they will separate as the ihs and cooler expand at different rates.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +2

      I'm not sure the deformation you speak of would be that bad. A temperature difference of 50­°c will cause a 1mm thick copper sheet to deform by 1 micron. For one that's still within the scale of wringability and actual real world temperature deltas between the edges of the IHS/cold plate are unlikely to be anywhere near that bad. Either way it'll become obvious if it's a problem.

    • @LawrenceTimme
      @LawrenceTimme 4 ปีที่แล้ว

      @@Penrowe it would be across in x and y that might be the problem. The amd ihs is 37mm square so in x and y that's 37 microns. The base of the cooler is a different thickness to the ihs as well as a different width. The water temp will be 40c max, and the cpu will be running at 80c
      Will be interesting to see though. I look forward to the results.

  • @SavouryLobster
    @SavouryLobster 4 ปีที่แล้ว +2

    This is absolutely incredible and very interesting. Great video!

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +2

      Thanks, glad you liked it

  • @rdasher7747
    @rdasher7747 4 ปีที่แล้ว +1

    I'm excited for the next part

  • @chapstickbomber
    @chapstickbomber 4 ปีที่แล้ว +10

    it would be interesting to see the effect of mounting pressure on thermal performance under wringing

    • @Cynyr
      @Cynyr 4 ปีที่แล้ว +1

      probably minimal given how hard it is to pull apart gauge blocks that have been wrung together. You have to slide them off of each other.

  • @kxder5688
    @kxder5688 4 ปีที่แล้ว +3

    This is some true expert and engineering stuff bruhh keep it up 🔥👍

  • @MrQwerty2524
    @MrQwerty2524 4 ปีที่แล้ว +1

    Never heard of this before in my life, tanks for the informative video!

  • @andrewhendrix2297
    @andrewhendrix2297 4 ปีที่แล้ว +1

    Mmmm, dat voice dough. Lap me up, sweet siren. Professor Rick would approve of your attempt at true flatness.

  • @N0N0111
    @N0N0111 3 ปีที่แล้ว +3

    @Penrowe
    Hi i am back, any progress?
    Tech Ingredients and Linus Tech Tips did a paste video review.
    So i am here again to ask if you had the new tvideo ready for us?

  • @dstevens6243
    @dstevens6243 4 ปีที่แล้ว +1

    Can't wait for the next video... hopefully?

  • @joreberking
    @joreberking 4 ปีที่แล้ว +2

    Great stuff! Very interested in the follow-up video :)

  • @noleandersson748
    @noleandersson748 4 ปีที่แล้ว +1

    Fun project, looking forward to see their results.

  • @blubery2
    @blubery2 9 หลายเดือนก่อน

    Klockren! vilken underskattad kanal. Vi behöver en revanchvideo

  • @pizzashtndragon
    @pizzashtndragon 4 ปีที่แล้ว +1

    Hey man, I hope I haven't missed anything here and I don't sound rude or offend you. Just your channel is really small, but this is a damn good video. I didn't know any of the stuff you talked about in the video but you made it so interesting and I've definitely learned a few things. If you made more content like this I think your channel could really grow!

  • @NightVisionOfficial
    @NightVisionOfficial 4 ปีที่แล้ว +1

    Dude, love video, love quality, love your editing, love your voice, and love how you, like others mechanics or electronic engineer can explain stuff that in school it would probably take month to learn xD

  • @docbrazen
    @docbrazen 4 ปีที่แล้ว +5

    You know you play to much Divinity: Original Sin 2 when you can recognize its music right away in a video.

    • @betaslayers
      @betaslayers 4 ปีที่แล้ว +2

      Also knew I heard this immediately

  • @johnfrian
    @johnfrian 3 ปีที่แล้ว +1

    Cool concept, awesome video, then.. nothing.. Are there any updates to this? The result of these efforts deserves to be shared!

  • @ryanmanner2503
    @ryanmanner2503 4 ปีที่แล้ว +1

    I AM SOOOO FUCKING EXCITED! You blew my mind when you discussed the history of lapping and how it was invented. I couldn't believe how simple and elegant the solution to creating a perfectly flat surface was. Now I await the second video and I believe Gamers Nexus will do this test justice.

  • @Laundry_Hamper
    @Laundry_Hamper ปีที่แล้ว +2

    Gotta use an optical pitch plate instead of that aluminium foil!

  • @nate6386
    @nate6386 4 ปีที่แล้ว +1

    Wow! I learned a whole bunch of stuff like the counterintuitive bits about how hard the lapping plates should be.

  • @SergenTemucin
    @SergenTemucin 4 ปีที่แล้ว +2

    What a fucking good video. I'm locked on it.

  • @piggfeathers
    @piggfeathers ปีที่แล้ว +1

    thank you

  • @Beverain
    @Beverain 4 ปีที่แล้ว +4

    This video makes me wanna become a machinist
    also, dat voice tho

  • @djr8585
    @djr8585 4 ปีที่แล้ว +1

    Can't wait to see the follow up from Gamers Nexus! I better get a detailed notification! Lol. Great video!

  • @theskinkworks63
    @theskinkworks63 4 ปีที่แล้ว +11

    I really hope Steve from GN will do the tests... how exciting!!!

    • @HMarcBower
      @HMarcBower 4 ปีที่แล้ว +1

      He does have a levelness tester (I know, my highly technical language is astonishing) now. :)

  • @kommandokodiak6025
    @kommandokodiak6025 3 ปีที่แล้ว +4

    Time for an update Penrowe

    • @Penrowe
      @Penrowe  3 ปีที่แล้ว +5

      Thinking it might be time to get my hands on a 5950x.

    • @N0N0111
      @N0N0111 3 ปีที่แล้ว +2

      @@Penrowe Yeah, the package you send to the person seems to be forgotten.
      Review from your side on 5950X lapping would be very HYPE :)

    • @Penrowe
      @Penrowe  3 ปีที่แล้ว +3

      @@N0N0111 Just had a look and not a single retailer anywhere near me has it in stock.

    • @kommandokodiak6025
      @kommandokodiak6025 2 ปีที่แล้ว +2

      @@Penrowe did 'they' scrap the testing after the rittenhouse video?

  • @KamiCrit
    @KamiCrit 2 ปีที่แล้ว +4

    So? What are the results?

  • @8lec_R
    @8lec_R 4 ปีที่แล้ว +1

    Great video. Next time please just lower the music by 5db or so, you have a great voice. Let us listen to it even more 😁

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +2

      Fair point

  • @h3ldermax
    @h3ldermax 4 ปีที่แล้ว +4

    Instant subscribe! Just have to know what happens next...

  • @XacTactX
    @XacTactX 3 ปีที่แล้ว +1

    Penrowe, this is incredible work and I'm so excited to see you make follow up videos about lapping processors and heatsinks. About 10 years ago I had a CPU cooler and when I lapped it with sandpaper and polishing compound I lowered the temperature of my CPU by 3-5* C (my testing methodology was extremely primitive so take this with a pinch of salt), since then I've always had a weird fascination with it. I've read quite a bit about lapping over the past 10 years but I think this is the first time that someone with your background and experience is tackling this subject. Please, please, please continue making videos and thank you for your time and effort
    Can you give any advice for people who are trying to lap their CPU, heatsink, or waterblock? Most people will not go to the lengths that you have gone to in order to get a perfectly flat surface, but are there any tips you can give to get closer to a flat surface with a relatively small cost? Most of the advice I have read in the past focuses on using a flat piece of glass, using different grades of wet/dry sandpaper, and sometimes it suggests using polishing compound to get a mirror finish. Is this correct, or how should it be changed?

  • @denny8360
    @denny8360 4 ปีที่แล้ว +1

    great content tbh, such a interesting and widely misunderstood topic

  • @ricsonchua4264
    @ricsonchua4264 2 ปีที่แล้ว +2

    Was there ever a follow-up to this? Would definitely still be good content even if the results were disappointing

  • @dotes12
    @dotes12 4 ปีที่แล้ว +1

    Mounting the heat sink directly on the silicon is the most thermally efficient way to cool it. Direct die cooling. The inside of the IHS doesn't even make contact with the top of the silicon! Manufacturers use either solder or thermal paste to transfer the heat from the silicon to the IHS, and then you use thermal paste again to transfer the heat from the IHS to the heat sink. The trick is making sure you don't crack the silicon!

    • @gubx42
      @gubx42 4 ปีที่แล้ว +1

      How flat is the die? The etching process is so ridiculously precise that I expect the die to be super flat too. And BTW, many CPUs in the early 2000s didn't have an IHS. You put the heat sink directly onto the die. And yes, lots of chipped dies, shops has signs explicitly telling they don't accept returns for chipped dies. Probably the reason why modern CPUs have IHS.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +1

      I do have some older CPU's I could attempt to delid to measure their flatness and check for wringability but even if they're not flat already it's something I should be able to solve. The only issue is time.

    • @obscureredtheoryaudiobookc5456
      @obscureredtheoryaudiobookc5456 4 ปีที่แล้ว

      3D stacked chiplets from AMD will eventually kill direct die cooling as the height of the individual chips will be of varying height.

  • @Kavster92
    @Kavster92 2 ปีที่แล้ว +3

    DUDE. Can you please tell us where to find the follow up. I can't find it anywhere and the hardware review publications only mention this video but no testing or follow up. Can you please direct us to where to find the follow up?

    • @Penrowe
      @Penrowe  2 ปีที่แล้ว +4

      I can't speak for GN but I've been busy working on projects that put food on the table. Sorry I haven't been able to finish this thing but it just doesn't pay a regular salary. I hope to close this out once I have the time and money to spare on my less serious projects.

    • @Kavster92
      @Kavster92 2 ปีที่แล้ว +3

      @@Penrowe Thanks man. I understand your situation and respect that your life is not lapping CPU's! Thanks for the highly informative video, I found it super interesting. Your style is really good.

  • @mochageico
    @mochageico 4 ปีที่แล้ว +1

    Content of this quality but 400 subs? Very very nice, can’t wait for the follow up. Though I’m sure we will be lol, takes a lot of time and effort

  • @selvaniklesh9332
    @selvaniklesh9332 4 ปีที่แล้ว +1

    This video is so underrated!

  • @ericdarriaga8326
    @ericdarriaga8326 3 ปีที่แล้ว +1

    this is quality content.

  • @v3xx3r
    @v3xx3r 4 ปีที่แล้ว +3

    I've been wondering for over a decade what would happen if you could wring a cpu to a cooler plate. I'd you really did send this to Steve I will be so happy.
    Also pcgamer wrote an article about you. That's how I find your channel.

  • @JosephLorelli
    @JosephLorelli 4 ปีที่แล้ว +1

    Omg the cliffhanger!

  • @jeremymcginnis7258
    @jeremymcginnis7258 3 ปีที่แล้ว +2

    Any updates on that last Beve Sturke comment?

  • @kubetail12
    @kubetail12 3 ปีที่แล้ว

    I have done my fair share of metallographic cross-sections for electron microscopy. Final polishing of cross sections is done with on soft polishing clothes with a colloidal polishing compound. This cloth is on a rotating table. A small amount of compound is used charge the cloth. A small stream of water goes on the table as you polish. It works very well. It's not really for flattening however. The surface with flat be to itself, but not to a particular plane.
    But I do not think a mirror finish is desirable. It will never prefect and there will be slight air gap at the interface. If the air is removed, heat transfer across such voids will be EM radiation (radiant heat transfer), which is not every efficient at low temperatures. You need something to fill this gap.
    However having the CPU and heatsink flat as possible with maximize the contact area.

  • @N0N0111
    @N0N0111 4 ปีที่แล้ว +1

    I hope he made 2 super flat CPU's and cooling blocks and send one set to Steve and one to Roman.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +3

      Wrote to him a while back but have heard no reply so have no idea if he'd be interested or not.

    • @N0N0111
      @N0N0111 4 ปีที่แล้ว +2

      @@Penrowe Well at least you can get him jealous now :)

  • @blindsniper35
    @blindsniper35 4 ปีที่แล้ว +3

    Finally somebody has done it
    It's been bugging me that I couldn't find anyone who actually lapped a CPU
    I'd like to see what you could do lapping an Intel die. Thanks for the interesting video.
    Edit
    Do you think making aluminum lapping plates would work / be worth it.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +1

      "Do you think making aluminum lapping plates would work / be worth it."
      I made a set of aluminum plates too out of 5754. Complete pain the arse and not worth the time spent on them. Tin is a more promising candidate but to make a set I'd need 10kg give or take, not all that easy to find since the only people buying/selling the stuff are tin soldier hobbyists and niche metalsmiths.

    • @benjaminalexander1998
      @benjaminalexander1998 4 ปีที่แล้ว +1

      @@Penrowe I'm not much for metals but I love puzzles. Would zinc be an option?
      I was working in my mind. Lead had the associated health issues. Magnesium has the burning issues. Zinc could be too hard.
      A few Google searches turned up a Bismuth/Tin alloy 281. Might be worth a look.

  • @sp00n
    @sp00n 3 ปีที่แล้ว +2

    Sooo, 6 months later, where is that follow up? 🤔

  • @JMUDoc
    @JMUDoc 4 ปีที่แล้ว +2

    Interesting point about the plates being too hard to do copper; the only softer plates I can find are made of tin!

  • @burchds84
    @burchds84 4 ปีที่แล้ว +2

    what's the update on this? Is there an ETA on any future experiments?

  • @GamingKing545
    @GamingKing545 ปีที่แล้ว +1

    time to do this with a 13900ks

  • @ИванГришаков-з5п
    @ИванГришаков-з5п 4 ปีที่แล้ว +1

    good precision for school project)

  • @SebLukaTech
    @SebLukaTech ปีที่แล้ว +2

    WHERES the BEEF?

  • @SpieleETC
    @SpieleETC 4 ปีที่แล้ว +4

    But what if there are no V O I D S ?
    Man this got deep.
    Or did it?

  • @madtrade
    @madtrade 4 ปีที่แล้ว +1

    this remember me when i was doing my own telescope mirror !
    good job man !@

  • @stephenhauer9055
    @stephenhauer9055 3 หลายเดือนก่อน +1

    Was there ever a followup made? I saw this video when you first posted it and always waited on someone doing a followup video on it but never did see one.

  • @highonpotnus7890
    @highonpotnus7890 4 ปีที่แล้ว +1

    I NEED MORE.

  • @djludowici536
    @djludowici536 4 ปีที่แล้ว

    This is amazing!! Would you please consider making a video showing how you made the lapping plates using the 3 plate method and cutting the channels into them etc? Also if you lap the CPU and thermal conductor together how does that work with the channels and such?

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +4

      Check the links in the description, oxtoolco's videos on hand lapping are several hours long in total and give you all the information you'd want. The only thing I did different was milling the groove pattern.

  • @MrBl00
    @MrBl00 4 ปีที่แล้ว +1

    Härlig video :D Ser fram emot testerna från Steve

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +1

      *Beve

  • @sTOPzTARINGAlright
    @sTOPzTARINGAlright 4 ปีที่แล้ว +1

    Damn part 2

  • @Caseycallender
    @Caseycallender 4 ปีที่แล้ว +1

    The suspense. Will you be releasing the video of the results or will I need to subscribe to another Chanel to see them?

    • @DizzTheDoc
      @DizzTheDoc 4 ปีที่แล้ว +1

      Gamers nexus channel is were the results will be when testing had concluded.

    • @Caseycallender
      @Caseycallender 4 ปีที่แล้ว +1

      @@DizzTheDoc thanks, already subscribed.

  • @usp211816
    @usp211816 4 ปีที่แล้ว +2

    The question of running a wrung processor/heatsink pair without paste has been kicking around in my mind for a bit.
    Perhaps this would be easier with an air cooler, even if it's only a single sample with a sample before and after test.
    Interested on seeing how this works out with the flattened IHS and paste though.

    • @dariuscastor2331
      @dariuscastor2331 4 ปีที่แล้ว +1

      I suspect it would be difficult, as the two may unwring when the processor heats up.

  • @emileturcotte4626
    @emileturcotte4626 2 ปีที่แล้ว +5

    @Penrowe has there been an update on this? What you have done is so cool!

  • @creepercharge1277
    @creepercharge1277 3 ปีที่แล้ว +2

    I only can find silicon carbide powder can you link where I can get the silicon carbide you used

    • @Penrowe
      @Penrowe  3 ปีที่แล้ว +2

      I use autosol but I'm sure other brands work fine too (mother's for example). You could also look for valve lapping paste from any auto parts store since those almost always include silicon carbide as the abrasive.

  • @lordsqueak
    @lordsqueak 4 ปีที่แล้ว +1

    Awesome stuff.
    I am however curious in regard to increasing the surface area, by using a rough paper on the flat surface. (much like the rifling on the bottom of cooking pans helps.) Any ideas of how that could be accomplished?

    • @taiiat0
      @taiiat0 4 ปีที่แล้ว +1

      if you want to rough up your surfaces, get a fine abrasive surface of your choice, and create your minor grooves in a specific direction. should take just a few seconds to make the grooves you're thinking of. make the grooves in the same orientation on the surface it will be mated to.
      then fill in the imperfections you just added with Paste and wonder if it was worth bothering.

  • @tryAGAIN87
    @tryAGAIN87 4 ปีที่แล้ว +2

    I came, I saw, and I came again... Crude but in reality, your video and the audio was like what I imagine people feel through ASMR... I saw this linked on /r/Amd. I am here in a my skilled capacity as Physicist & Nanotechnologist to impart a few points of wisdom. Your video strikes me as you are not just a machinist. Not just a man curious about flat metal. You strike me as someone who wants to understand exactly what he is doing. That is why I stopped here, subscribed and to drop you this line, and to impart some of the physics in what I am guessing is a project you with to take to fruition?
    First things first. If you are half the Abom79 you seem to be, you have researched the gauge block story. It's the first stop one makes before learning about micron level measurement.
    But gauge blocks are a step past this. You don't use clumsy thumb screw pins to measure the lands and grooves in a flattened surface any more. You use the wavelength of light that refract off the surface through stressed glass. A laser that spans the width of a few mW will do. Red laser, green, blue, makes no difference at the micron scale because the wavelength is measured on a scale that a one micron wide scratch, not deep but wide, is enough that even a red 650 nm laser pointer on the order of 5mW will create enough decoherence of the beam in that pit, the light will begin to scatter very spectacularly.
    Now on the scale of gauge blocks, the flatness is on the order of between 2 and 3.3 magnitudes higher. It depends on the quality of the gauge block. But the wringing effect is not what people assume it to be. It is not actually the wringing out of the atmosphere between the blocks that causes the 'sticking' of the blocks. It is a mixture of physical effects and sometimes chemical effects that cause the wringing to happen. on the scale of nanotechnology, we have to measure defects of 0.5 nm pits using much more pompously named devices. (I hate that physicists of old were so up themselves.) But suffice to say, the simplest way of measuring the flatness of two mating surfaces is by seeing if it can be permeated by the kinetic diameter of a given gas molecule. Sounds difficult right?
    Well the actual process is very simple. Attach a hose to the side of the mated surfaces and see if you can separate the two just with the gas force alone. If it can force them apart by gas pressure alone, you know you have defects that span the surface that are at least as large as the gases kinetic diameter. For Oxygen gas, the molecule being O2, the kinetic diameter is approx. one third of a nanometre. And this is where the scale of the physical forces comes in. On that order of scale, if you are unable to force two gold plates apart, you never will unless you cut them apart, as the pits between the gold atoms are roughly that same size. And gold loves to form molecular bonds with more gold to make more crystal lattice. At this scale, the electromagnetic force, the gravitational force, and the energies spread between potentials and kinetics reaches an equilibrium where the two surfaces are no longer separate surfaces. At this point, you wouldn't need thermal paste. Until you applied power to the processor. At which point, due to the different expansion rates of the steel integrated heat spreader, and the heat sink material, the heat sink would catastrophically 'pop' off the CPU. Or, if that's not bad enough, the bonding force of the two surfaces may be enough that the heat could bend the IHS in some nasty way that may crack the silicon crystal it is attached to (can't remember if Zen is soldered, but I know the bulldozer chip was...). This bimetallic strips. Your indicator has one. One side copper, the other side an allow of tin, copper and zinc. The differences in the crystal lattice cause one strip to expand ever so slightly further as the current passes through them creating heat. The more heat, the more acute the bend. It keeps bending until it makes that famous click-clack-click-clack sound. That is literally metal bending itself so far out of shape it overcomes a magnetic force on it and arcs as it separates. Then it cools so rapidly it slams back into the contact point again.
    That would be the perfectly lapped CPU. Lapped so well to perfection it throws the heatsink out of the computer, shatters the crystal under that perfectly lapped metal, right before it goes into meltdown. It would be glorious. But alas, no room to stretch no place for the heat to travel. The thermal paste, thermal grease, or whatever tack compound you use, it doesn't just transfer the heat, it allows the cpu and heatsink to 'breathe' at varying rates while still keeping the surfaces in contact.
    Now I know the catastrophic end is a little out of proportion, but you would be surprised at the amount of potential energy held in that zone between two different flat, wrung metal surfaces. As the heat climbs to thermal criticality, so does the potential for pieces of metal to fly off at speeds on the order of multiples of the speed of sound...
    Anyway, I had a great time watching your video, and I have to say, not only is watching you teach yourself how to do this stuff satisfying, but so is your voice. It's on par with Morgan Freeman

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +1

      I've heard concerns from a lot of people that the unequal thermal expansion of the metals involved could lead to a catastrophic breakdown of the system - or at least degrade the thermal conductivity to the point where either performance throttling or damage were the only possible outcomes.
      I'm not prepared to publish results yet but in the system I've been running under full load now for several weeks it's not what I've observed. I do have a large temperature delta between the two core dies that would be a cause for concern if I were using some thermal interface material but here it's just a reminder that oh right, flat is an ideal and 300nm is still a lot of deviation.
      I'm not sure how much energy you'd have to put into the system to provoke the kind of catastrophy you describe but I have a suspicion something else would have to give long before that point were reached.
      I was not aware of macro-scale cold welding being accomplished with gold, if there's a publicly available research article on the topic I'd enjoy reading it.
      The next project will be a set of scratch built stainless steel angle plates using the three plate method.

    • @tryAGAIN87
      @tryAGAIN87 4 ปีที่แล้ว +1

      @@Penrowe I should clarify that I was being hyperbolic about the scale of the catastrophic failure. The type of failure would be runaway thermal spike.
      The mode of failure, and it's level of risk, would be in the worst thermal scenario. Cold standby, and booting immediately into running multi-threaded recursive mathematics such as a mersenne prime search. It would need to create a shelf between the two metals, not a gradient. Having a gradient means there is still thermal transfer between the two separate parts. Now, from memory, steel expands by about 10 micron per meter Kelvin. and after doing a little searching, I found, and was extremely surprised to find this out, AMD and Intel both use nickel plate copper IHS's. Copper's thermal expansion is significantly higher, and much closer to that of aluminium. now, because that expansion rate is so incredibly small, you would need a large and fast increase in temperature to cause there to become a shelf between the two materials. Looking at the thermal conductivity of the two, aluminium has just above half the thermal conductivity of copper. so that would double the amount of thermal mass needed to radiate the same heat as the copper area transferring the heat to it. But, because the two surface meet at equal area's that cannot happen. Instead, the copper becomes hotter and thus more efficient at radiating and conducting its own heat to the aluminium.
      That would roughly equalise the two surfaces expansion rate. I actually predict using ANY heatsink, and ANY processor of the combinations Al/Cu or Cu/Cu would actually be able to transfer the heat properly. If fact, the standard heat transfer of thermal paste (the standard arctic silver paste or equiv.) has a thermal conductivity just over 1% that of copper. And heatsinks are still able to transfer heat just fine through that.
      So with careful consideration of the materials you are using, and considering the ductility of copper, I honestly think you have nothing at all to be concerned about by trying it, other than thermal throttling in case the mating of the surfaces is not quite good enough. my prediction is actually that you TDP will increase from the wraith prisms average of 125W to a much higher value, of the order of 200W(aluminum heatsink)/350W(copper heatsink) just by getting the surfaces to mate well. Realistically you would be looking at about half that increase because even micron pis are enought that the surfaces are instead of using conduction, will instead rely on radiation and absorption to transfer the heat. And unfortunately there will always be random amounts of backwards transfers as a result where the heatsink radiates some back at the IHS. Hence, hot spots.
      And you are correct. 300nm is massive. But it is still small enough that unpolarised light can still pass through a half wavelength and never get the chance to refract off the side of it and thus would create destructive interference to any wavelength multiplets. But if the fringe is all you are looking for, and your sensor is good enough, using a blue-violet laser around the 400-450 nm mark, you can get light to create large constrictive fringes on the scale of 1 nm pits all you need is the light to not finish its wavelength before hits the bottom of a pit and it will create some very beautiful scattering fringes. The other thing with high sensitivity light refraction flatness testing is that you must tune the optical piece to be some exact multiple of the center of the wavelength span you are using. That is where the difficulty comes in with optics. the more intense the light you are attempting to measure with, the smaller tolerances you have to hit. Comes down to Heisenberg's uncertainty principle. You can know the depth of your pit to a much greater accuracy the larger it gets. But the smaller it gets? At some point, that ceases. It usually occurs after the nano scale. Anything below 1-2nm will have to take into account the atomic structure of the two materials, and if you still want to, then you can always use smaller and smaller wavelengths of light. But using anything more powerful than violet light, it becomes ionising radiation, and at that point, you are no longer measuring the surface, rather you are evaporating it :D
      I've always wanted to lap a processor and heatsink together as a project myself, and I am keen to see what the outcome of this will be. I am also a very curious learner such as yourself. So I have nothing but the utmost respect for what you are doing and I thoroughly appreciate anyone doing work like this.
      As for the macroscale cold welding, there are not a lot of publicly available documents on it, otherwise patented welding technology could be free to all. I think that's bullshit, but alas the world has succumbed to capitalism. So instead I'll link a few youtubers who have done research on this exact topic:
      1. AvE seeing if he can cold weld two 5 ounce gold bars
      2. Cody's Lab doing a collab after AvE sent the bars to his to see if he could do some more testing on them.
      3. Veritasium explaining how everything that is prepared for exposure to space must have an oxide layer, or made of plastics/ceramics and other non crystalline materials that must react to be able to fuse so cold welding wont occur.
      I know these three are only anecdotal, but the theory matches it, and two of them produced the same results (alas with the very same pieces of metal, but thats life.). To throw a spanner in the works, the last video is a surprising, yet again anecdotal, contradiction to the theory. But it is absolutely fascinating.
      th-cam.com/video/mHAhfX7iRjs/w-d-xo.html
      th-cam.com/video/GtcuURSYgvo/w-d-xo.html
      th-cam.com/video/Y2nQ8isf55s/w-d-xo.html
      th-cam.com/video/07CShwlSnms/w-d-xo.html
      If you have access through a Uni or other research institution with partnership to Nature, here is a great article that explores cold welding gold nanowires to create nano machines without the need of heat jointing and how it brings the possibility of smaller and smaller joints into the picture. If you don't have access through an institution, then the abstract, images they show and their references are interesting to look at either way:
      www.nature.com/articles/nnano.2010.4

    • @tryAGAIN87
      @tryAGAIN87 4 ปีที่แล้ว +1

      Oh and please please please keep me up to date with this, I don't just want to know the gist of it (although with your voice I would still be interested lol) , I want to understand it. And I'm guessing you'd be the right person to talk with that we may be able to help each other understand more about this :D Good luck, and I have confidence that the project will work out surprisingly well, regardless of how un-accessible it will likely be to the home user.
      P.S. magazine paper has a thin layer of coating on it that creates the reflective shine. I know this as I come from a long line of print finishers the powder has to be be micron sized to be able to stay suspended in the curing agent, so it will be the perfect polishing implement. layer that on a true flat, and you have yourself the best possible polisher. Truly flat, truly ductile enough to capture the swaff, with hardened particle sizes so small you you'd never find it even in a machinist shop, and the best of all, 2 bucks a book of 100 odd sheets lol.

    • @Penrowe
      @Penrowe  4 ปีที่แล้ว +1

      I work ten hour days beyond which I've got a family and a house to take care of so most of my projects get really drawn out - and that's if I don't go back and redraft a script ten times over.
      If you want to check on the status of things here's a tool I use to track what I have in my pipeline: trello.com/b/gBZeei76/pipeline
      What kind of magazine paper should I be looking for? Gloss or matte? Approximate paper weight? Would it make much difference if it's from a particular manufacturer?

  • @akashk911
    @akashk911 11 หลายเดือนก่อน +1

    This is exactly what i was looking for. Duck gamer nexus and kingpin, when th😊ey dont value true Engineering principles to get flat surfaces. RINGING it's the keyword!!

  • @Merkky
    @Merkky 4 ปีที่แล้ว

    Great video! Looking forward to the follow up, take my sub! Såg nog PostNord, antar att du är från Norden? 👍