Back in my day, lapping meant sanding BOTH the CPU and heatsink so get a perfectly smooth mirror finish mating surface that eliminates the use of thermal paste. It was supposed to be a form of cold welding, or the effect you get when two mirror finishes of metals come in contact to each other, they tend to stick together very well.
+Kirill Ostapenko That's what lapping is. Luke didn't get that HS anywhere near mirror finish and I'm assuming the CPU went untouched. The whole video was a waste of time, but I guess they got the views, so they get the ad revenue, and that's all that matters.
+zefrum3 wow ,i thought you said masochist instead of machinist , i still dont think i was wrong xD sanding cpu and heatsink for hours is something M would love to do
+Kirill Ostapenko Even a mirror finish is still not even close to smooth enough to eliminate the use of thermal paste. For that you would need a surface that's smooth at molecular level.
RisseK1 I didn't say it's not interesting. My comment was based on my perception of Lukes feelings. The null result is perfectly valid and interesting, but it isn't fun to put in hours of work just to find out none of it mattered, unless you're the sort of person who does enjoy that. And not having fun can lead to burn out. Which would suck, because these videos are interesting. Also, next time, leave your condescending attitude in the trash, where it belongs. Or else expect to be ignored or deleted.
+Ryan P Have you seen the state of the CPU they use? Due to their poor taste in thermal paste it is scratched up REALLY bad. Lapping a heatsink then using that CPU won't make a difference when their CPU heat spreader is in a much worse condition then the heatsink before lapping.
+Ryan P Maybe a comparison of coolers based on design. Test a swath of direct contact, stock, indirect contact, 4 pipe, 6 pipe, ect. We all know stock suck but the difference in performance would be cool to know when purchasing.
As someone who works with a lot of heat transfer (Nuclear Navy and Civilian Reformers) it kind of surprises me the amount of effort you guys are putting into this. The temperature you deal with in PC's is so low that as long as the surface is relatively smooth and flat you aren't going to see a difference due to polish. You are tightening down the cooler and forcing the two metals to make very nice contact, so that's all you need for nice heat transfer. As long as the material of the heat sink transfers energy rapidly. If you were trying to transfer something more like 300 F and up then maybe you would see the polish affect it but it would still be negligible. Differential temperature is still a better factor, so if your air cooled system is under performing you either need better fans or a better fin set, if that still doesn't work you need a better medium like water. The direct metal connection should be the last thing you put your time and energy into.
Another big issue is the thin pipes of copper that have to transfer energy a long way before the sink. The larger the differential the better the heat transfer and a long pipe makes for a slow gradual temperature change. It would be more ideal if the sink started earlier.
+Ian Tusa After taking heat transfer courses It surprises me how much effort people put into things that don't really matter. Metal on metal contact is going to have such a high heat transfer rate that nothing like this was ever going to matter much.
+Ian Tusa Not to mention that the manufacturers would polish it themselves if it made a difference. I mean that would be a super cheap way to improve performance of your product, but most don't do it.
+Michael Kinkema While you guys might have qualifications in heat transfer you obviously don't know much about computers, imperfections on heat sinks allows nucleation of the blue smoke and results in a slow leak that degrades clock speeds over time. I've seen a drop in base clock of 50MHz in just one year from poor machining. I'm sure in another couple of years people will have been putting LEDs in computers for so long that they'll start to think there's some performance benefit to it.
Us old school overclockers invented lapping to fix the really shitty heat sink bases that were around back in 99 when all the extreme overlocking and watercooling started. Lapped Celeron chips down to the copper, and this was before AC5 was around, which was another factor back then, shitty white thermal paste that couldn't conduct good enough. Now days bases are made much truer, and with AC5 or other silver thermal compounds, it really fills in any of those minor machine scratches that are sometimes left. Extreme overclocking might see a bit of a help but that is only under extreme cooling, such as trying to get the chip as cold as possible with refrigeration.
Old school RUles!! right! i remember in Argentina only found crappy heatsinks made in china , only bad formed aluminium with anodyzed colours, 1cm height or 2 cm max, and cracppy fan coolers.. i started some mods on heatsinks around 1996 with Pentiums or AMD , I sell computers and I did not like to sell those crap, I started to make copper heatsinks and use Sunon fans, the best I could get in my city, those some computers were sold to Radio Transmitters where the computer worked 24 hours a day without stopping , at the most it was restarted every x days .. but never failed any even the audio playback computer had a cyrix 233 or 300 mhz by 1996 and that they warmed up a lot, with those modifications and holes in the cabinet all Fresh and worked wonders. Not only had to use ingenuity and Lapping
actually current IHS have still very poor flatness and lapping improves cooling, this is important when overclocking. Here are very good measurments: www.overclockers.com/heatsink-mounting-pressure-vs-performance/
@@nopochoclos So basically you were buying crappy coolers, improving them and re-selling? Nice business! Did you have problems with original manufacturers?
Your sanded surface didn't look that much better from the non-sanded one, in case people don't know a higher grit removes less material so you need to do more work with them, and you keep going until no scratches from the previous grit are visible. With 2000 grit wet sanding should leave a near mirror finish. Not that this would make a big difference but if you get into this might as well do it proper. And I imagine convex heatsinks only do well on lidded CPUs as they push the cover closer to the dye, should show a pretty substantial improvement if you are one of the unlucky people that came with badly placed covers.
Hey Luke, a definitive conclusion that comes from the scientific method is always useful, even if the answer is "Whatever" now we have proof that cable management is only for aesthetics and if anyone makes a big deal about it saying that it will effect performance, we now know that it wont thanks to you putting in the time to test it, that it will not effect performance. knowing that is important and i hope you make more workshop videos because Science!
Hey, I've got about 3 CPU degrees down doing a good cable management when replacing my PSU... (I have a single case fan for HDDs only) Of course it's in idle. Don't you think it will help under big load?
I can assume that all things together do somehing. Properly cable managed case + lapping of cooler and IHS + using good paste + replacing Intel's paste with LM + proper airflow in the case. Every single part might only give a little bit, but it should add up.
+Tea drinking crumpet munching British dude Wanna hear something really fucked up? I was looking up Tupac's mom and interviews with her and the next day she died.
+News Cartridge It is worse, but so long as you're not rubbing your feet on the ground like an idiot and making sure to touch your case before touching your parts its gonna be fine.
I've built about 15 pcs + upgrades/management to them on carpet and never had an issue. BUT I went and bought a anti shock wristband just in case as electro static build up IS real and CAN/HAS broken pc parts for people. it's very unlikely to happen but it can.
I like it when these things end up not mattering Luke, it means I don't have to go through the tedium of doing them. I felt particularly vindicated by the whole "cable management doesn't matter" video, that was fucking amazing.
Hi Luke. Lapping a heatsink will never make a difference when thermal paste is used. As the role of thermal paste is to nullify any imperfections. If you redo the tests without the paste you will notice different results. Cheers!
lapping the CPU is more effective. most coolers are flat enough or a little convex and work fine. But, often the CPU is concave. My current 4930K had a very poor heat spreader. When I put a straight edge over it and shined a flashlight at the joint you could see the rather large gap at the center. After lapping temps dropped almost 10C. Been lapped a bit over a year. I have had it for about three years now. maybe four. I also had to lap my Q6600 back in the day. Same issue. If you don't have a big gap under the straight edge test, it's not worth doing.
+Chad Bremer well said but i think 10c is a stretch. I got 4C drop on my old C2D and 3C on dated but current i7 860. My first gen TRUE was pretty flat so i never bothered to lap that but both CPUs were sunken in. Used a straightedge and feeler gauge at the time to measure how much.
Lawrence Sinanan like I said, it was a rather large gap in the center. Way bigger than normal. So, couple extra degrees most likely from that. I said almost 10C. Don't recall the exact difference. Around 8c maybe. It was a year ago. Lol
+Speedyjens Yes, that's exactly what it does. The idea with lapping is to minimize the need for the paste, as solid copper conducts heat better than any thermal paste.
+Speedyjens Can't be as good as metal on metal. However you would have to literally weld it on to do that, and it's not that easy. Also there's thermal paste UNDER the heat spreader that you can replace with better aftermarket stuff.
+wannabesq But didn't they disprove the need for precise thermal paste in the last video that did not have a deceive answer besides don't use to little?
+Jasoman that video was talking about thermal paste application techniques, not the quality of the thermal paste itself (which can be subpar when it came with the stock cooler).
it is kind of nice to see someone test something for me, and to show that it doesn't matter IMO is still important, it means that it doesn't matter that much with thermal paste application, I don't need to spend all that extra time fiddling with cable management. So when a freind brings up one of these things you've tested, I can say that I've seen the experiment, I've seen the results and I know where it all stands. TY for doing these Luke, I really enjoy watching the results even if it's not that ground breaking
WARNING!! DO NOT TWIST THE HEATSINK WHILE STILL ON THE PAPER WHEN TURNING 90°!! MAKE SURE TO GENTLY LIFT IT FIRST , THEN TURN 90° (while in air), PLACE BACK DOWN & CONTINUE!! If you don't then each time you do this it will leave unwanted round scratches. You also have to repeat this process many more times than 5 each way with each grit of sandpaper so be prepared to be there for a while doing this process of lapping. [I have lapped heatsinks before so I have first hand experience with this.] Temp Results: As for how much of a difference it makes in the temp. My testing resulted in no more than a 1-2c change in temp that fluctuated and could of been accounted for as a margin of error. I honestly would just buy some decent thermal compound like Prolimatech PK-3 and not waste your time with Lapping. Tips & Experience: The process is not fun and a huge pain in the arse like wet sanding is not a pain as it is, you have to deal with the heatsink trying to skip across the surface once your water is not fresh clean. If you can build a setup to have constant flowing water across the sandpaper it would make this much easier. You should also seal the sandpaper with good quality duct tape ALL the way around, not just at each side. Sealing it helps prevent water from getting underneath the sandpaper and weakening its stability by waterlogging it.
+sleepup7931 No, he wasted his time doing it on a heatsink that was already flat. If he had done it on a cpu that had a concave/convex or a heatsink that had that, it would have shown improvement.
+Francefire That baffled me. The Hyper Evo 212 is already a great piece of hardware that I have seen in more than one PC, and people are happy with it. A concave heatsink would have been far more useful in this video. There are unfortunately just a handful of them, and I know of one that's been showing promises but kept being disappointing due to the often-reported concave contact surface, the 130W TDP-rated vapor-chamber ID-Cooling IS-VC45 cooler.
The entire point of using heatsink compound is to put a thermally conductive material between the CPU and heatsink surface that will fill any surface imperfections or gaps. If the CPU case and the bottom of the heatsink were true and perfect flat surfaces, I don't think you'd need or care about the compound, if the surfaces made perfect contact at all points, the heat would conduct without the compound. So in my eyes at least the use of the heatsink compound makes lapping an exercise in futility.
The whole idea behind a mirror finish is to ensure that there is a solid contact between cpu/paste/cooler, the smoother it is the more likely it is to make more contact, since even microscopic bumps can cause crevices that don't actually make contact. A rougher surface could allow for air bubbles or gaps.....but there is a thick enough layer of paste that it shouldn't be an issue if you make sure you apply proper pressure when mounting to spread the paste, especially if it's done from a single drop in the center since that will push any air potential pockets out. Still, two mirror surfaces will make more surface contact. But in the end, heat dissipation comes down to mass. The base collects the heat, the heat pipes transfer the heat to the fins, and the fins radiate the heat away.
+Matlockization Dude, you're 5 months late to the party. The issue was that certain coolers were exceeding the recommended specification mounting pressure. Which hasn't changed in Skylake from previous generations. Skylake silicon being on a thinner substrate (NOT WAFER) brought the fact that these were exceeding specification into light. This was dealt with and sorted within about 2 weeks by the aftermarket cooler manufacturers that this was related to..
I can give some excellent advice on this because I used to run a small business that did exactly this type of work. DON'T BOTHER. Here's the details - My father had a small manufacturing business. I had access big money CNC type grinders, polishers, drill presses, lathes, etc. I started a small PC customizing biz using his equipment during off hours. I could polish CPU coolers to insane mirror level finishes that you could have used to pop your pimples! No joke. This was industrial level gear. I did hundreds of CPU coolers for just over a year. I tested batches of my finished products regularly to know and track the results and see if I could improve my techniques and processes. There was such a small gain (IF ANY AT ALL), that just wasn't worth the cost. Customers realized this and the business died. At first, I was doing 10-20 per day and over the course of a year it dropped off. Because once you got it back and saw your own personal before and after temps, or compared your polished CM HYPER temps to your buddies out-of-the-box temps - you wouldn't be impressed or happy. I voluntarily stopped offering that service because I could not improve anything. I could only take a sucker's money and I won't roll like that. Here's why - IT"S THE THERMAL PASTE THAT MATTERS! Read below to understand. #1 - A quality CPU cooler (well reviewed and tested by the good sites out there) with a copper base will give you almost perfect results. #2 - A quality thermal paste (well reviewed and tested by the good sites out there) will give you even better performance and improve upon #1. #3 - A quality thermal paste is between the CPU and the COOLER. It's the thermal paste that's the center of the thermal exchange. As long as the cooler device is smooth and flat and well made, it doesn't matter if it is factory smooth or a polished to a mirror. The thermal paste will do the work! #4 - Even after all of the above, MAYBE, POSSIBLY letting me polish it might net you 2 more measly degrees. #5- Aside from the above, the only noticeable differences would come from either improving a poorly made cooler or fixing a defective unit that some how slipped through quality control. The conclusion - A cheapo brand or poorly made CPU cooler can be improved upon and you will make noticeable gains, but would you really want to risk using a cheapo product that needs to be improved OOTB on that expensive OCed rig of yours? NO! Is it worth all that extra time and aggro or money for a business like mine charged for 1-2 degrees of improvement? NO! The best and simplest ways to gain lower temps are - Buy a quality brand, well reviewed and tested cooler. Use a quality brand, well reviewed and tested thermal paste and (this is really important but almost never done) remove it and re-apply fresh paste every 6 months. Also, do not use paste older than 1 year. This is debatable but most of the info I've read over the years says to always use new paste, not year old stuff from the toolbox. Why not right? Get a better case that will allow for better cooling/air flow. Get better fans. Remove obstructions and do clean cable management. Lower the ambient room temperature. Use better or different quality components. Example - I had an expensive brand name power supply but no matter how you positioned it, it gave off tons of freaking heat and turned my rig into a tiny oven. I did my research and bought another quality brand. Guess what? My temps dropped dramatically and my cooling problems were instantly and easily resolved.
They make them really well these days. Years ago, early 2000s, it was different. Now, the studies and lab results are there and they know what to do. But it's like any thing these days. You have to keep an eye on testing and reviews. There are some excellent websites out there doing really great work. They'll guide you on what to buy. If polishing to mirror levels would knock out the competition, they'd be doing it and bragging from the mountain tops. Poorly made products can be improved. Quality brands are great OOTB. No need to go crazy and polish unless your OCD or doing it for some other personal reasons. Enthusiasts might do it just for bragging rights. Sorry. I don't have time to waste on that bullcrap. I'd rather go for a motorcycle ride, see a movie, go to the beach...you know...live real life. lol
Dude, I closed the business. I don't have numbers if that's what you mean. It doesn't matter. What I said above applies. The improvements were 1-2 degrees at best. I did various brands and people could send in their own for modifications. Everything you need to know is in my OP. It's just that simple.
Doing this isn't going to do much because your thermal paste is going to fill any surface imperfections - that's why it's used. It might make a significant difference if you can't use thermal paste for whatever reason, but in that case, you would also have to finish the surface of the CPU heatspreader as well, not just the cooler. The thermal paste should conduct heat better than the cooler material, so you won't be able to benefit from having 2 perfectly polished surfaces, but no paste. Also, contact pressure has nothing to do with thermal transfer. It is all about surface area and the materials. The reason convex surfaces get a benefit is because there is more surface area, with thermal paste making sure there is contact all the way to the edge. Try comparing a convex surface to a flat one with no paste, the convex surface will run a lot hotter. Concave surfaces don't get the same advantage because air gets trapped in the cavity, preventing the thermal paste from making full contact.
+Henry Chan lol, i really thought about it, when i started watching the movie. 'does it even matter'? 'I bet he'll say: i doesnt even matter that much'. "-_- Make a comparison between a normal CPU cooler, or a after market/third party cooler. Does that even matter?
I'm like really late to this conversation but the fact that most of these workshop videos just come up with no difference is actually really interesting to me. Because it shows that the hardware engineers and the industry at large has already tackled these questions at some point when their design decisions were made and ensured that the peak efficiency range of the product is so above and beyond any variations that may be encountered in day to day operations. Parallel, perpendicular, push, pull, etc whatever. And I think that's kinda cool. It doesn't leave too much room for enthusiasts to tinker because the ceiling is so high right now but it also means most novice or average builders/users will never have to worry about these variables.
sorry, it matters not at all. the thermal paste, when applied correctly, fills in all gaps, crevasses and curves and allows for proper heat transfer. furthermore, scratching away the surface of the metals will actually increase the possibility of heat induced oxidation which will hurt the ability of the heat sinc to transfer the heat. DO NOT DO THIS
***** no, increased surface area of the fins dissipates the heat, the surface area of the sinc that makes contact with the cpu, as long as it at least covers the entire cpu surface, does not increase the heat transfer from the cpu to the sinc. because there is only so much surface to transfer from. the size of the cpu is the bottleneck
+Tigglet Fills, as in removes air which is a insulator. That's a whole different thing then conducting as well as a solid piece of copper or aluminum. And if you remove Air...you don't have oxidation.
Despite the results, it was a great video to demonstrate and document one observation of the experiment. Thanks for taking one for the team Luke. You're a credit to the community.
Dear luke , you could try what linus did wrong I mean delliding your CPU but this time put liquid metal between the die and the IHS and thermal paste between the IHS and the heatsink, if you clean the glue the right way (not like linus)you should see from 10 to 20 degrees improvement, I tried this myself it works...
+spaceminions I'm not sure what you mean... but the IHS that goes glue to the cpu die may contain aluminum of some sort, I think the liquid metal does corrode a little the IHS but is just a plate , if years later you wanna change it back you can remove the IHS and install the heatsink directly on the die, all of this is because intel doesn't wanna solder the cpu and puts thermal paste instead 'cause it's cheaper
Nahuel Cutrera I've seen pure gallium weaken aluminum so much that you can tear it easily; a gallium alloy might not be a good idea around an aluminum alloy.
+spaceminions this is the one I used www.amazon.com/Coollaboratory-Liquid-Thermal-Interface-Material/dp/B0039RY3MM not sure if it contains gallium, and here there's a bunch of guys doing comparisons only they used the liquid metal on the IHS too ... www.overclock.net/t/1351984/coollaboratory-liquid-ultra-vs-liquid-pro-vs-phobya-liquid-metal
You have to lap the IHS in addition to the heatsink to have any results. It doesn't really matter if one is flat if the other isn't. I have lapped my CPU several times before, and it definitely helps, but the softer copper is prone to warping and may cancel out or be worse over time with pressure from the mounted heatsink.
I sharpen knives and a tip you learn quickly to see if you have leveled a surface is to cover the surface in sharpie and do a couple passes against the stone or abrasive. The spots making contact against the abrasive will have the sharpie removed, the spots not making contact will not have it removed. Also, since copper and aluminium are much, much softer than hardened steel, I would suggest starting at 400 grit, then moving to 1000 grit and finishing with 2000 grit. Making an evenly convexed heat sink with an abrasive by hand would be near impossible.
I LOVE Luke's "WHATEVER" videos. He is the myth-buster of computer builds -- not that he's likely to change anyone's mind. It would be nice if he'd say something about fan speeds, does he keep them at constant speed or not?
+DeutscheJunge Why didn't he just make it a video about how CPU lapping doesn't matter? Instead he made a video about "I'm going to prove why CPU lapping works, oh wait I'm stupid". He could have done all the research, reached his conclusion, then made a video about it... Is it just me?
Nice, you did some good research. That sandpaper progression is nearly exactly what I use when sharpening a knife to a mirror finish. Except I go on to leather strops after 1,600 grit.
NASA created a ball bearing so smooth that if it were the size of the earth, the highest imperfection would only be a few metres high. It is the smoothest object ever created. Ofc, they did it with LazerPewPew. You should test a heat sink made by NASA.
Take a look on Veritasium's channel, he posted a video about the roundest object in the world. It's his most popular video IIRC. Basically, it's a metal sphere about 6" across, and if it were scaled up to the same size of the earth the highest mountain and lowest valley would only be 14 metres apart (7m above and 7m below sea level). Impressive considering the Marianas Trench is 10,809m below sea level and Mount Everest is 8,848m above sea level. NASA did have the record, but this one has it beat.
Luke: I find the "Whatever" videos to be extremely useful. Like the cable management one. I always fretted over cable management but have never had a glass side panel so it was always about air flow never about looks. Imagine my joy after watching your video and finding out that cable management equals squat! Now I get hours of my life back for every build. Thanks so much. See - - even the whatever videos have a lot of value to someone.
Curious as to why you didn't lap both the CPU and heatsink? Part 2? I feel like intel just doesn't give 2 craps about their IHS (as well as thermal interface between the die and IHS, but thats another story). Cpu cooler manufacturers make their coolers nice and flat and shiny to sell more of them. Lapping the cpu i feel would net better results. But of course lapping the more expensive part, and voiding the warranty on that is an undesirable thing. I believe my cpu's IHS had some issues with "flatness" as i had some very high variation in core temp. I'm custom water cooled with an EK supremacy evo and 4790k. I didn't lap my cpu, but when i removed my IHS and got the "naked ivy" mounting kit for the supremacy evo, my temps went down drastically, as well as my variation in core temps. (i was delidded with CLU prior to installing the naked ivy kit and was not happy with my temps/core temp variations). I do still have some core temp variation (core 2 runs ~5C hotter, which is better than its previous ~10C) and overall temps dropped as well.
Luke Get a granite surface block, and some prussian blue, put the prussion blue on the granite, wipe leaving a thin coat, then rub the cooler plate over the granite plate, and then look at the cooler plate to see if there is a uniform coat of blue (ie, it's flat) or blue on the outer periphery (ie, it is concave) or it is just blue over the center (ie, its convex). Having sorted the cooler by the plate geometry, then try the lapping again on the concave and convex ones and test to see what happens Grizly sells an affordable 12x12 granite surface plate if you need a source, or just check with a local machine shop, they will have them and probably let you do the tests on their plate John
Hey Luke, I just want to let you know I still appreciate and enjoy you doing these workshop videos. Still super interesting to see it in action since I don't have the resources to do it myself.
You know I respect people who didnt like this video and idea but I am actually very pleased that these guys are working really hard and undeniable enthusiasts in this community. Everybody can give average ass info about any subject but these guys are putting their soul into what they are doing and we all as a community should appreciate that. They are trying to find a balance between usefull (applicable) and "huh? interesting, thats cool to know!" type of content. All and all this is a business and they're gonna get that sweet youtubemonay but we all should admit they very well deserve it.
+Friederrich Eggeling if you have a 2x2cm cpu and a 2x2cm heatsink you have 4cm sq of surface area maximum once any gaps are filled in with thermal paste. if the heatsink had a sawtooth arrangment that perfectly interlocked with a sawtooth design on the cpu with no air in there what so ever (unlikely) then you could get the area over 4cm sq however the IHS on the cpu and base of the heatsink still have the same cross dimensional area so will still have the same internal heat conduction so it still wouldnt be any better. FWIW I heard in college that some BMW cylinder heads were machined so flat that they didn't need a head gasket between the head and block, never seen it first hand though.
+Djhg2000 The problem is air. If the surface is too rough, or especially concave, you can trap tiny pockets of air in between the paste and the surface of the cooler, making it less efficient.
+Toms Tech Engineers have designed heat transfer solutions for cooling for years now. There's a reason there's no jigsaw. Adding in jigsaws would actually be less beneficial to thermal conductivity, because you'd increase the volume of copper the heat has to go through before reaching the liquid in the heatpipes. Heatpipes rely on a phase-change design, with liquid that becomes gas, and travels to (one of) the other end of the heatpipe way faster than heat can travel its way through copper over the same length. Less volume of copper is better for heat to reach the liquid faster, so flat surface is better.
Hey, you guys should pioneer research on these topics... be the first company to provide definitive data on these workshop videos. Use the forum to gather lots of data and publish it... Keep these workshop videos coming! They are wonderful, and I appreciate them! :D
Im a machinist , I do this from time to time at work with hydraulic pump parts during rebuilds. When lapping if possible it should be stressed in the exact same way it would be in use during the lapping process. Once mounted to the cpu the base of the cooler may be stressed and bent slightly from said stress. Both aluminum and copper are soft very ductile materials and may be effected greatly depending on the stresses involved in mounting . I dont know how u would solve this on a cup cooler but just wanted to note something I knew about lapping that may or may not be critical in this situation.
A quick guide to ultimate contact. 1: Put the heatsink on a drill press. 2: Place the CPU under the heatsink. 3: Start the drill press. 4: At a relatively low drill speed setting (don't allow the drill to pull your heatsink apart), gently lower the spinning heatsink onto the CPU. 5: As the two components are grinding on each other in a circular motion, any imperfections are being flattened, and any curvature is being aligned. 6: Apply enough pressure to heat up both components, making sure not to overheat the CPU (give it a break every minute or so.) 7: After about 30 minutes (allowing for breaks), your two components should now fit onto each other as snugly as were they 1 piece of metal. Only those two components will ever fit each other though. So make sure you have a perfect heatsink for that CPU that you aren't going to want to change for a while.
Just speaking as a machinist. (I grind parts with a tolerance of .00005 inches) When lapping or checking for high spots by *hand* you should always try to create a figure eight motion on a flat surface. This limits the creation of a convex shape on the material. Turning the heat sink 90 degrees just makes it convex in two directions. The removed material will always be in between your part and the lapping medium you are using. Which lifts up your parts and causes it not to sit flat. IMHO use some blueing compound on the mirror and lightly press the the heat sink on the glass. The ink will only transfer to the high spots. Sand those down (or use a stone, which probably would be better) and repeat until you have 85-90% transfer. This is usually good down to .0002 flatness.
U forgot the CPU luke. Otherwise thanks for proving a point. I know others will say this but let me repeat it. The WHOLE POINT OF THERMAL PASTE is to fill in those imperfections and the curves. If their is some major curves than lapping might do a little bit, mabye. If it was something easy and quick it woukd not really matter, everybody would do it. But it is not, it takes lots of time, takes some work, and can possible void a warrenty. Unless u are some crazy overclocker i dont see any reason to do this. I will admit i have never done this so i dont know from experiance, so if it works for u ... school me. But i dobt see any reaso ti do it
If you lap your heatsink base, you also need to lap the CPU IHS. i was already doing it back a few years ago on a 3.9Ghz E6750 under a CM 212 and it was a big improvement Im currently using a lapped/dellided 3770K @ 4.9Ghz under a HN-D14, and this is the perfect setup for using liquidmetal !
I don't get the high amount of dislikes. Luke obviously put TONS of effort to produce this video. Maybe change the name of the series to Myth Busters or something like that.
Luke is just too horrible! I've been subscribed (not anymore) for only one month, and he just bores the hell out of me, so annoying. Can't stand him......
The heatsink isn't always what you want to lap. The CPU lid is. If you delid your processor and lap it... you'd be happy to see it DOES make a huge difference at high temps . :)
Very poorly done lapping, You should hold by the base, not high by the fins, best to press the base itself gently in the very centre of it, and the result should be mirror like which this wasn't. After 3h this was very shitty result
+3majsie maciek Not really, he could have used up to only 600-800grit paper. If he wanted mirror he could have gone 1000+. Through all the cpu's I have lapped- going for the mirror finish has any negligible difference. However, he should have lapped the IHS on the cpu if he wanted to see any significant difference.
I think you guys are going into a lot of these without doing research, basically making another useless "conclusion" which is just adding to the already mind boggling confusion that exists around the topic. If you want to do something meaningful, do a full investigation like that of SPCR and gather a large amount of information and sift through it to get the useful stuff instead of trying to stick to some upload schedule by taking shortcuts and making useless content.
I have an ARCTIC Freezer XTREME Rev. 2 that originally came with the worst contact surface I've ever seen. The surface was full of scratches and not flat at all. In fact when I placed the heatsink on a table, it could almost make it swing from left to right because of the curved surface. So I decided to lap it and, even if I didn't finish the process because after 3 and an half hours I got really tired, I saw a considerable improvement on the performance. Before, my FX-8350 was thermal throttling after 15-20 seconds in full load but now it doesn't thermal throttle at all and the max temperature reaches about 60 degrees celcius. And there's a lot more work I can do on it.
Lapping the heatsink is one side of the matter. One would also need to lap the CPU Integrated Heat Spreader. This would then ensure that both are relatively flat surfaces, thus allowing for maximum contact and transfer of heat (with minimal use of thermal paste). As to whether it would have any actually noticeable effect, can't say as I haven't bothered/needed that level of performance. I do recall a few forum posts around the net, having mirror polished both the heatsink and CPU IHS; to the point where the CPU stuck to the heatsink due to suction.
I'm sorry this is way too much effort to get a few degrees difference in temps. much rather just spend the extra and get a good aio to get a more significant change.
+Lorentz the annoying Llama Have you seen the state of the CPU they use? Due to their poor taste in thermal paste it is scratched up REALLY bad. Lapping a heatsink then using that CPU won't make a difference when their CPU heat spreader is in a much worse condition then the heatsink before lapping.
+Lorentz the annoying Llama They do tend to liberally apply TIM before putting a cooler on, just a rice grain size drop in the center is all that's needed not a stripe clear across the CPU like they do. The TIM is probably preventing any decrease in temperatures.
+SweBass Yes you're exactly right, that is the thermal paste's job, so this is pretty much stupid to do anything for those scratches, groves and other things. but a flat surface is important, which sanding will not achieve without HOURS of work, You will want to instead take it to a machining shop and have them flatten it.
Sean Anderson You can get flat results from lapping, hell you can get mirror finish results without the slightest distortion with lapping like in the video. It's just doing a good job takes time and practice which were obviously both missing from the host of the video. When he was lapping on camera you could see some shudder once when he changed directions, what's shown in the final video is the best footage they have, there was that one screw up in the best footage they had so I will assume there were far more and far worse. Taking it to a machine shop isn't needed, just time and patience as well as practice blocks of copper.
I for one appreciate the "Whatever" results. People argue endlessly and venomously over these things. These videos remind me that people will go to great lengths to argue something that doesn't matter and helps me ignore it all.
i call BS on lapping making any diff unless your heatsink had some kind of manufacturing defect to begin with. All those minor imperfections will be filled with thermal compount once you seat them together filling in any imperfections / small voids.
What about different liquid cooling liquids. It makes a difference in cars due to the different thermal capacity of water vs ethylene glychol vs polypropylene glychol etc. Maybe another option exists, does methanol +water (windshield wiperfluid) or something similar work better? Just a thought. I switched my race car to polypropylene glychol (sometimes found at walmart as RV antifreeze) and my engine runs cooler yet my coolant temp is higher because its pulling away more heat, but it doesnt flow as well so you need a bigger radiator. So might be a fun test and should show results for sure.
+Merp Water cools better than just about anything, including automotive coolant/antifreeze and rv coolant. The additives are to prevent corrosion and freezing and in general the higher the additive content, the worse the cooling. The only thing that can really make water cool a little better is an additive that reduces surface tension.
garfunkle187 Generally true, water can move more heat, however water also microvaporizes around very hot things causing cavitation preventing good contact with the hot object reducing colling effectiveness. Also water has to be pressurized to take higher heats where other higher boiling points do not have to be. Not sure if it matters for CPU cooling but would be interesting, but you also cannot running straight water in coolers or you will get corrosion or mold. Also many overclockers claim Methanol is better than water, and what they sell in water coolers is glychol based and many claim polypropylene is better. So I think it would be a valid test.
I agree the test would be valid simply because people do use those different coolants, i just wanted to make the point about water being a good coolant inherently. To clarify further: Yes some liquids boil at higher temps than water, but ANY liquid will boil at a higher temp at higher pressure and vise versa and most coolants do NOT have a higher boiling point at a given pressure than water. The cavitation you speak of is what is helped by the surface tension reducers (eg water wetter) i specifically mentioned. Regardless, ANY liquid will be subject to these same effects. The main difference between ethylene glycol and proplylene glycol is how they react to different metals, specifically nickel, aluminum and steel.
garfunkle187 True, although proplylene glycol does have a 370 F boiling point. And ethylene glycol is 387 F So quite a bit more than water, so you do not need to worry about pressurizing to increase boiling points :)
What you did in the video would be considered simply wet sanding. My understanding of the lapping process would require lapping compound (a paste with abrasive grit in it used commonly in automotive valve lapping). It further would require lapping both contact surfaces which would include the CPU lid. I don't know if it would be a good idea, since any remaining compound might foul the CPU, but it may be worth a shot! I'd love to see a quick revisit with proper lapping technique.
+Hayden Ross No kidding, Linus Media Group has really been going downhill with these strange ideas and just plain boring ones. I don't think I would ever care about the shape of my heatsink. I'm not designing a brand new computer in my garage, I'm simply going to be spending some money every now and then upgrading or building an enthusiast grade machine.
+Hayden Jacobson now I've lapped my g3258 the actual cpu itself and it did shed a few degrees not much but a few. I've never heard of someone doing it to the actual heatsink. Only seen people do it to the cpu itself.
lapped my big typhoon like 8 years ago when I had an Opteron 165. Dropped load temps about 5 or 6 degrees. The heatsink was slightly concave from the factory. Lapping made a huge difference for me. I also removed the IHS. That did another couple degrees as well.
+Aoru Isn't thermal paste a lower-thermal conductor than nickel or copper? If so then yes this would matter (although probably a very small amount, almost unnoticeable amount). However if the thermal conductivity is the same then, from my understanding of the laws of thermal dynamics, the edging should not matter as the transfer of heat energy would be the same.
+Aoru Thermal paste fills into microscopic cracks, using it to fill macroscopic imperfections (because of either surface being curved) would result in poorer performance because it is actually a worse conductor than copper or aluminium.
***** Not always. If thermal paste truly is useless you wouldn't waste money on it and just put the slab of heatsink directly on the chip. The paste is there to transfer heat better to the copper and aluminum that dissipate heat. As an example, GPu vs CPU. CPU might survive without paste, but not a GPU. So thermal paste can be a macroscopic factor in that sense.
It is significant, but my gaming pc uses a crappy HP heatsink and it runs fine. I don't see how 5 degrees is worth hours of my time rubbing it on a piece of sandpaper.
I've made today a very good experience with lapping my cpu & heatsink. I had problems with overclocking my Q9650 past 3.4 Ghz. When stressing the system with Prime95 after overclocking it crashed due to too high temperatures. It took me the whole afternoon but i would say it was worth it. Temperatures dropped about 5 degrees Celsius at minimum and the cpu now is stable in prime with 3.6 Ghz (air cooled). Probably results may vary for other people.
Lapping is something I do on a regular basis so I'm going to give this a shot on my video card heat sink. It rests directly on the silicon wafer so black polishing should provide maximum contact. Thanks.
Heat transfer through conduction is pretty intuitive, as in contact area helps. We could have assumed that when using a good paste (which we always do) it's not going to matter significantly if surface is rough or glassy. If for some weird reason you weren't using paste, it would likely matter more since glassy and flat should slightly increase contact when 'dry'. There are ways to measure pressure in different areas under mounted tension and of course it follows that the best shape would be one that 'deflects to even pressure' instead of having higher pressure around the edges. But of course that would all depend on the size, thickness, shape and material of the face, and the cpu heatspreader. It's easy to imagine that often there would be so little deflection that flat would still be best, but only measurement could tell. I think it's hard to beat a decent all copper air design for bang for the buck/trouble/etc. I'm looking forward to Pascal release tomorrow.
I figured the surface wouldn't matter since the thermal paste is there. Thermal paste is basically meant to fill all of the gaps between the two surfaces with a very conducting substance, instead of it just being air. It may have made a bigger difference on older models where manufacturing quality was lower, and maybe the surfaces were really bad. At that time thermal paste might have been worse. It would also make a difference if you didn't have thermal paste, for some reason. You would then need both surfaces to be as smooth as possible to make the best contact and allow as little air between as possible. For those that don't know, air is a thermal insulator, so trapped air means bad conduction. That's by double pane glass works well.
Sounds exactly like something I'd like to never have to do.
Heeyyy snazzy!
+Snazzy Labs Shots fired.
Wait who let you here?
+Snazzy Labs Afterall, it is a one-time procedure.
+Bloodstainer like all those virginities I took
Back in my day, lapping meant sanding BOTH the CPU and heatsink so get a perfectly smooth mirror finish mating surface that eliminates the use of thermal paste. It was supposed to be a form of cold welding, or the effect you get when two mirror finishes of metals come in contact to each other, they tend to stick together very well.
+Kirill Ostapenko amazing
+Kirill Ostapenko yup, what this guy said; any machinist can confirm this
+Kirill Ostapenko That's what lapping is. Luke didn't get that HS anywhere near mirror finish and I'm assuming the CPU went untouched. The whole video was a waste of time, but I guess they got the views, so they get the ad revenue, and that's all that matters.
+zefrum3 wow ,i thought you said masochist instead of machinist , i still dont think i was wrong xD
sanding cpu and heatsink for hours is something M would love to do
+Kirill Ostapenko Even a mirror finish is still not even close to smooth enough to eliminate the use of thermal paste. For that you would need a surface that's smooth at molecular level.
Aw, poor Luke. We really need to find you something to test that actually matters.
RisseK1 I didn't say it's not interesting. My comment was based on my perception of Lukes feelings. The null result is perfectly valid and interesting, but it isn't fun to put in hours of work just to find out none of it mattered, unless you're the sort of person who does enjoy that. And not having fun can lead to burn out. Which would suck, because these videos are interesting.
Also, next time, leave your condescending attitude in the trash, where it belongs. Or else expect to be ignored or deleted.
Ryan P Big boi and his big words :3 adorable.
+My eyebrows are GONE!? Perception, interesting, condescending...
I feel like those were the only words with over 10 letters ._.
+Ryan P Have you seen the state of the CPU they use? Due to their poor taste in thermal paste it is scratched up REALLY bad. Lapping a heatsink then using that CPU won't make a difference when their CPU heat spreader is in a much worse condition then the heatsink before lapping.
+Ryan P Maybe a comparison of coolers based on design. Test a swath of direct contact, stock, indirect contact, 4 pipe, 6 pipe, ect. We all know stock suck but the difference in performance would be cool to know when purchasing.
keep pressing 5 when the video begins, something seems to be hurting Luke...
+LoLUnderstated Well spotted.
+McJaews Why thank you good sir (tips hat)
+LoLUnderstated You should have more likes
+mostafa “smile gamer” al-shawesh Lol
+LoLUnderstated LMFAO
As someone who works with a lot of heat transfer (Nuclear Navy and Civilian Reformers) it kind of surprises me the amount of effort you guys are putting into this. The temperature you deal with in PC's is so low that as long as the surface is relatively smooth and flat you aren't going to see a difference due to polish. You are tightening down the cooler and forcing the two metals to make very nice contact, so that's all you need for nice heat transfer. As long as the material of the heat sink transfers energy rapidly. If you were trying to transfer something more like 300 F and up then maybe you would see the polish affect it but it would still be negligible. Differential temperature is still a better factor, so if your air cooled system is under performing you either need better fans or a better fin set, if that still doesn't work you need a better medium like water. The direct metal connection should be the last thing you put your time and energy into.
Another big issue is the thin pipes of copper that have to transfer energy a long way before the sink. The larger the differential the better the heat transfer and a long pipe makes for a slow gradual temperature change. It would be more ideal if the sink started earlier.
+Ian Tusa After taking heat transfer courses It surprises me how much effort people put into things that don't really matter. Metal on metal contact is going to have such a high heat transfer rate that nothing like this was ever going to matter much.
Steve Exactly
+Ian Tusa Not to mention that the manufacturers would polish it themselves if it made a difference. I mean that would be a super cheap way to improve performance of your product, but most don't do it.
+Michael Kinkema While you guys might have qualifications in heat transfer you obviously don't know much about computers, imperfections on heat sinks allows nucleation of the blue smoke and results in a slow leak that degrades clock speeds over time. I've seen a drop in base clock of 50MHz in just one year from poor machining.
I'm sure in another couple of years people will have been putting LEDs in computers for so long that they'll start to think there's some performance benefit to it.
Us old school overclockers invented lapping to fix the really shitty heat sink bases that were around back in 99 when all the extreme overlocking and watercooling started. Lapped Celeron chips down to the copper, and this was before AC5 was around, which was another factor back then, shitty white thermal paste that couldn't conduct good enough. Now days bases are made much truer, and with AC5 or other silver thermal compounds, it really fills in any of those minor machine scratches that are sometimes left. Extreme overclocking might see a bit of a help but that is only under extreme cooling, such as trying to get the chip as cold as possible with refrigeration.
Exactly right.
It may help a crappy heat sink, but with modern parts, there's a chance that the "lapping" process actually harms performance.
Old school RUles!! right! i remember in Argentina only found crappy heatsinks made in china , only bad formed aluminium with anodyzed colours, 1cm height or 2 cm max, and cracppy fan coolers.. i started some mods on heatsinks around 1996 with Pentiums or AMD , I sell computers and I did not like to sell those crap, I started to make copper heatsinks and use Sunon fans, the best I could get in my city, those some computers were sold to Radio Transmitters where the computer worked 24 hours a day without stopping , at the most it was restarted every x days .. but never failed any even the audio playback computer had a cyrix 233 or 300 mhz by 1996 and that they warmed up a lot, with those modifications and holes in the cabinet all Fresh and worked wonders.
Not only had to use ingenuity and Lapping
actually current IHS have still very poor flatness and lapping improves cooling, this is important when overclocking. Here are very good measurments: www.overclockers.com/heatsink-mounting-pressure-vs-performance/
Wow, so water cooling started in 99?? I'm surprised.
@@nopochoclos So basically you were buying crappy coolers, improving them and re-selling? Nice business! Did you have problems with original manufacturers?
I recommend going to 0:00 and keep hitting the "J" key :p
Seepy, see, see, see, see pee you...
He's good at blinking.
+Prashank Abhishek try pressing 5 :D
+Jürgen Ruut Thats a rap song!!!
+Prashank Abhishek "ow"
I had a Duron 1.3Ghz about 13 years ago... after I sanded the cpu heatsink till it became a mirrorred bottom, the CPU temps were 10ºC lower...
Your sanded surface didn't look that much better from the non-sanded one, in case people don't know a higher grit removes less material so you need to do more work with them, and you keep going until no scratches from the previous grit are visible. With 2000 grit wet sanding should leave a near mirror finish.
Not that this would make a big difference but if you get into this might as well do it proper.
And I imagine convex heatsinks only do well on lidded CPUs as they push the cover closer to the dye, should show a pretty substantial improvement if you are one of the unlucky people that came with badly placed covers.
When working with glass, we used 1800 for near perfect transparency and 3200 for perfect transparency. Literally perfect, at least to the naked eye.
The hardest part is manual sanding is hard to determine whether if its totally flat or slanted
Hey Luke, a definitive conclusion that comes from the scientific method is always useful, even if the answer is "Whatever"
now we have proof that cable management is only for aesthetics and if anyone makes a big deal about it saying that it will effect performance, we now know that it wont thanks to you putting in the time to test it, that it will not effect performance. knowing that is important and i hope you make more workshop videos
because Science!
Hey, I've got about 3 CPU degrees down doing a good cable management when replacing my PSU... (I have a single case fan for HDDs only)
Of course it's in idle. Don't you think it will help under big load?
I can assume that all things together do somehing.
Properly cable managed case + lapping of cooler and IHS + using good paste + replacing Intel's paste with LM + proper airflow in the case. Every single part might only give a little bit, but it should add up.
Linus should do what Gordon Ramsey did and go on a "how to build a pc class" undercover
+Tea drinking crumpet munching British dude Wanna hear something really fucked up? I was looking up Tupac's mom and interviews with her and the next day she died.
+Bulletproof_Gaming CS:GO savage
wolfblood21432 :>
+GamesN' GENERAL "YOUR CPU IS SO UNDERCLOCKED, IT GOT OUTBENCHED BY A PENTIUM 4!" - Linus Ramsay, 2016
+GamesN' GENERAL
CPU
IN
SOCKET
CLOSE
DONE
luke, please do a "building on carpet vs not on carpet" to show people it doesn't matter. XD thanks.
lol
+News Cartridge YEAH
+News Cartridge It is worse, but so long as you're not rubbing your feet on the ground like an idiot and making sure to touch your case before touching your parts its gonna be fine.
I've built about 15 pcs + upgrades/management to them on carpet and never had an issue. BUT I went and bought a anti shock wristband just in case as electro static build up IS real and CAN/HAS broken pc parts for people. it's very unlikely to happen but it can.
+News Cartridge My 290x doesnt agree with you. Only card that ever died on me, was when i changed the paste in my couch (with a rug mat)
I like it when these things end up not mattering Luke, it means I don't have to go through the tedium of doing them.
I felt particularly vindicated by the whole "cable management doesn't matter" video, that was fucking amazing.
Hi Luke.
Lapping a heatsink will never make a difference when thermal paste is used. As the role of thermal paste is to nullify any imperfections. If you redo the tests without the paste you will notice different results.
Cheers!
+Overclocked Gamer mine craft = autism
+Long Qua Missionary
false. Its one of the greatest gaming innovations ever. You clearly are missing the point
+PewPewDie You are a literal garbage human being.
+Long Qua Missionary I am autistic
+Kak Pis why would you announce that
Can we expect to see a "Taskbar - Locked or Unlocked?" video next week?
+dpersonalizedALT good joke
locked
+dpersonalizedALT I'm assuming the reason for this video is because everyone has been kind busy/dead because they've been at PAX/SDC
You should test different kinds of dust filters, how much they impact airflow and how effective they are at stopping dust.
honestly dont think that these are pointless, if anything these are great for debunking myths. I still like your content Luke, you do a good job.
lapping the CPU is more effective. most coolers are flat enough or a little convex and work fine. But, often the CPU is concave. My current 4930K had a very poor heat spreader. When I put a straight edge over it and shined a flashlight at the joint you could see the rather large gap at the center. After lapping temps dropped almost 10C. Been lapped a bit over a year. I have had it for about three years now. maybe four. I also had to lap my Q6600 back in the day. Same issue. If you don't have a big gap under the straight edge test, it's not worth doing.
+Chad Bremer well said but i think 10c is a stretch. I got 4C drop on my old C2D and 3C on dated but current i7 860. My first gen TRUE was pretty flat so i never bothered to lap that but both CPUs were sunken in. Used a straightedge and feeler gauge at the time to measure how much.
Lawrence Sinanan like I said, it was a rather large gap in the center. Way bigger than normal. So, couple extra degrees most likely from that. I said almost 10C. Don't recall the exact difference. Around 8c maybe. It was a year ago. Lol
Isnt that what thermal paste does? It fills all the imperfections and conducts heat.
+Speedyjens Somewhat
+Speedyjens Yes, that's exactly what it does. The idea with lapping is to minimize the need for the paste, as solid copper conducts heat better than any thermal paste.
+Speedyjens Can't be as good as metal on metal. However you would have to literally weld it on to do that, and it's not that easy. Also there's thermal paste UNDER the heat spreader that you can replace with better aftermarket stuff.
+wannabesq But didn't they disprove the need for precise thermal paste in the last video that did not have a deceive answer besides don't use to little?
+Jasoman that video was talking about thermal paste application techniques, not the quality of the thermal paste itself (which can be subpar when it came with the stock cooler).
it is kind of nice to see someone test something for me, and to show that it doesn't matter IMO is still important, it means that it doesn't matter that much with thermal paste application, I don't need to spend all that extra time fiddling with cable management. So when a freind brings up one of these things you've tested, I can say that I've seen the experiment, I've seen the results and I know where it all stands.
TY for doing these Luke, I really enjoy watching the results even if it's not that ground breaking
Curves are never imperfections ;)
WARNING!! DO NOT TWIST THE HEATSINK WHILE STILL ON THE PAPER WHEN TURNING 90°!! MAKE SURE TO GENTLY LIFT IT FIRST , THEN TURN 90° (while in air), PLACE BACK DOWN & CONTINUE!!
If you don't then each time you do this it will leave unwanted round scratches. You also have to repeat this process many more times than 5 each way with each grit of sandpaper so be prepared to be there for a while doing this process of lapping.
[I have lapped heatsinks before so I have first hand experience with this.]
Temp Results:
As for how much of a difference it makes in the temp. My testing resulted in no more than a 1-2c change in temp that fluctuated and could of been accounted for as a margin of error. I honestly would just buy some decent thermal compound like Prolimatech PK-3 and not waste your time with Lapping.
Tips & Experience:
The process is not fun and a huge pain in the arse like wet sanding is not a pain as it is, you have to deal with the heatsink trying to skip across the surface once your water is not fresh clean. If you can build a setup to have constant flowing water across the sandpaper it would make this much easier. You should also seal the sandpaper with good quality duct tape ALL the way around, not just at each side. Sealing it helps prevent water from getting underneath the sandpaper and weakening its stability by waterlogging it.
so the result is " WHATEVER" ?
+sleepup7931 yes
No, it's because he tested on a 212 evo
+sleepup7931 No, he wasted his time doing it on a heatsink that was already flat. If he had done it on a cpu that had a concave/convex or a heatsink that had that, it would have shown improvement.
DaMirrorLink you're right !
+Francefire That baffled me. The Hyper Evo 212 is already a great piece of hardware that I have seen in more than one PC, and people are happy with it. A concave heatsink would have been far more useful in this video. There are unfortunately just a handful of them, and I know of one that's been showing promises but kept being disappointing due to the often-reported concave contact surface, the 130W TDP-rated vapor-chamber ID-Cooling IS-VC45 cooler.
The entire point of using heatsink compound is to put a thermally conductive material between the CPU and heatsink surface that will fill any surface imperfections or gaps. If the CPU case and the bottom of the heatsink were true and perfect flat surfaces, I don't think you'd need or care about the compound, if the surfaces made perfect contact at all points, the heat would conduct without the compound. So in my eyes at least the use of the heatsink compound makes lapping an exercise in futility.
+compactc9 So lapping still doesn't matter, if you use a thermal compound?
Next video, "Does CPU Cooling Matter At All?"
Conclusion...somehow, no.
+MikeM8891 if cpu lapping mattered then copper shims would not be a thing. It is as useless as the thermal paste spread...it does not really matter.
MikeM8891 Conclusion "Whatever"
The whole idea behind a mirror finish is to ensure that there is a solid contact between cpu/paste/cooler, the smoother it is the more likely it is to make more contact, since even microscopic bumps can cause crevices that don't actually make contact.
A rougher surface could allow for air bubbles or gaps.....but there is a thick enough layer of paste that it shouldn't be an issue if you make sure you apply proper pressure when mounting to spread the paste, especially if it's done from a single drop in the center since that will push any air potential pockets out. Still, two mirror surfaces will make more surface contact.
But in the end, heat dissipation comes down to mass. The base collects the heat, the heat pipes transfer the heat to the fins, and the fins radiate the heat away.
It's probably the convex heatsinks breaking those Skylake CPUs.
+Nicholas Li Any heatsink will break a Skylake CPU.
+SirCodeye nor mine:P
+Nicholas Li if you broke a skylake cpu with a heat sink then you are doing it wrong
SirCodeye 'Mine doesn't' doesn't explain why the wafer is so much thinner than Broadwell. It also doesn't explain how its happening to so many people.
+Matlockization Dude, you're 5 months late to the party.
The issue was that certain coolers were exceeding the recommended specification mounting pressure. Which hasn't changed in Skylake from previous generations. Skylake silicon being on a thinner substrate (NOT WAFER) brought the fact that these were exceeding specification into light.
This was dealt with and sorted within about 2 weeks by the aftermarket cooler manufacturers that this was related to..
I can give some excellent advice on this because I used to run a small business that did exactly this type of work. DON'T BOTHER.
Here's the details -
My father had a small manufacturing business. I had access big money CNC type grinders, polishers, drill presses, lathes, etc. I started a small PC customizing biz using his equipment during off hours. I could polish CPU coolers to insane mirror level finishes that you could have used to pop your pimples! No joke. This was industrial level gear. I did hundreds of CPU coolers for just over a year. I tested batches of my finished products regularly to know and track the results and see if I could improve my techniques and processes. There was such a small gain (IF ANY AT ALL), that just wasn't worth the cost. Customers realized this and the business died. At first, I was doing 10-20 per day and over the course of a year it dropped off. Because once you got it back and saw your own personal before and after temps, or compared your polished CM HYPER temps to your buddies out-of-the-box temps - you wouldn't be impressed or happy. I voluntarily stopped offering that service because I could not improve anything. I could only take a sucker's money and I won't roll like that.
Here's why - IT"S THE THERMAL PASTE THAT MATTERS! Read below to understand.
#1 - A quality CPU cooler (well reviewed and tested by the good sites out there) with a copper base will give you almost perfect results.
#2 - A quality thermal paste (well reviewed and tested by the good sites out there) will give you even better performance and improve upon #1.
#3 - A quality thermal paste is between the CPU and the COOLER. It's the thermal paste that's the center of the thermal exchange. As long as the cooler device is smooth and flat and well made, it doesn't matter if it is factory smooth or a polished to a mirror. The thermal paste will do the work!
#4 - Even after all of the above, MAYBE, POSSIBLY letting me polish it might net you 2 more measly degrees.
#5- Aside from the above, the only noticeable differences would come from either improving a poorly made cooler or fixing a defective unit that some how slipped through quality control.
The conclusion -
A cheapo brand or poorly made CPU cooler can be improved upon and you will make noticeable gains, but would you really want to risk using a cheapo product that needs to be improved OOTB on that expensive OCed rig of yours? NO!
Is it worth all that extra time and aggro or money for a business like mine charged for 1-2 degrees of improvement? NO!
The best and simplest ways to gain lower temps are -
Buy a quality brand, well reviewed and tested cooler.
Use a quality brand, well reviewed and tested thermal paste and (this is really important but almost never done) remove it and re-apply fresh paste every 6 months. Also, do not use paste older than 1 year. This is debatable but most of the info I've read over the years says to always use new paste, not year old stuff from the toolbox. Why not right?
Get a better case that will allow for better cooling/air flow.
Get better fans.
Remove obstructions and do clean cable management.
Lower the ambient room temperature.
Use better or different quality components. Example - I had an expensive brand name power supply but no matter how you positioned it, it gave off tons of freaking heat and turned my rig into a tiny oven. I did my research and bought another quality brand. Guess what? My temps dropped dramatically and my cooling problems were instantly and easily resolved.
+Barda Wulf excellent post! I bought a corsair h80i a year ago and it's been great OOTB, 2500k 4.2 idles at 25, runs 100% at 45
They make them really well these days. Years ago, early 2000s, it was different. Now, the studies and lab results are there and they know what to do. But it's like any thing these days. You have to keep an eye on testing and reviews. There are some excellent websites out there doing really great work. They'll guide you on what to buy. If polishing to mirror levels would knock out the competition, they'd be doing it and bragging from the mountain tops. Poorly made products can be improved. Quality brands are great OOTB. No need to go crazy and polish unless your OCD or doing it for some other personal reasons. Enthusiasts might do it just for bragging rights. Sorry. I don't have time to waste on that bullcrap. I'd rather go for a motorcycle ride, see a movie, go to the beach...you know...live real life. lol
+Barda Wulf Wheres the details?
Dude, I closed the business. I don't have numbers if that's what you mean. It doesn't matter. What I said above applies. The improvements were 1-2 degrees at best. I did various brands and people could send in their own for modifications. Everything you need to know is in my OP. It's just that simple.
So no actual lapping of the CPU?
"Whatever" results on such "experiments" is a good thing. It liberates the consumer from marketing propaganda. Trust physics, not companies.
Doing this isn't going to do much because your thermal paste is going to fill any surface imperfections - that's why it's used. It might make a significant difference if you can't use thermal paste for whatever reason, but in that case, you would also have to finish the surface of the CPU heatspreader as well, not just the cooler. The thermal paste should conduct heat better than the cooler material, so you won't be able to benefit from having 2 perfectly polished surfaces, but no paste.
Also, contact pressure has nothing to do with thermal transfer. It is all about surface area and the materials. The reason convex surfaces get a benefit is because there is more surface area, with thermal paste making sure there is contact all the way to the edge. Try comparing a convex surface to a flat one with no paste, the convex surface will run a lot hotter. Concave surfaces don't get the same advantage because air gets trapped in the cavity, preventing the thermal paste from making full contact.
Great video! I don't think you should be discouraged by negative results. They are still very meaningful.
Don't even waste your time watching this. The answer is, again, it doesn't matter at all whatsoever.
Yeah don't watch it plus he did it wrong
+Henry Chan lol, i really thought about it, when i started watching the movie.
'does it even matter'? 'I bet he'll say: i doesnt even matter that much'. "-_-
Make a comparison between a normal CPU cooler, or a after market/third party cooler.
Does that even matter?
I'm like really late to this conversation but the fact that most of these workshop videos just come up with no difference is actually really interesting to me. Because it shows that the hardware engineers and the industry at large has already tackled these questions at some point when their design decisions were made and ensured that the peak efficiency range of the product is so above and beyond any variations that may be encountered in day to day operations. Parallel, perpendicular, push, pull, etc whatever. And I think that's kinda cool. It doesn't leave too much room for enthusiasts to tinker because the ceiling is so high right now but it also means most novice or average builders/users will never have to worry about these variables.
sorry, it matters not at all. the thermal paste, when applied correctly, fills in all gaps, crevasses and curves and allows for proper heat transfer.
furthermore, scratching away the surface of the metals will actually increase the possibility of heat induced oxidation which will hurt the ability of the heat sinc to transfer the heat.
DO NOT DO THIS
this is correct, to actually see the affect of lapping you would want to test before and after without thermal paste.
*****
no, increased surface area of the fins dissipates the heat, the surface area of the sinc that makes contact with the cpu, as long as it at least covers the entire cpu surface, does not increase the heat transfer from the cpu to the sinc. because there is only so much surface to transfer from.
the size of the cpu is the bottleneck
+Tigglet Fills, as in removes air which is a insulator.
That's a whole different thing then conducting as well as a solid piece of copper or aluminum.
And if you remove Air...you don't have oxidation.
+Tigglet Please use capital letters, it's annoying to read a text without them.
Hannes Dendoncker
no
Despite the results, it was a great video to demonstrate and document one observation of the experiment. Thanks for taking one for the team Luke. You're a credit to the community.
Dear luke , you could try what linus did wrong I mean delliding your CPU but this time put liquid metal between the die and the IHS and thermal paste between the IHS and the heatsink, if you clean the glue the right way (not like linus)you should see from 10 to 20 degrees improvement, I tried this myself it works...
+Nahuel Cutrera Does a cpu contain aluminum?
+spaceminions I'm not sure what you mean... but the IHS that goes glue to the cpu die may contain aluminum of some sort, I think the liquid metal does corrode a little the IHS but is just a plate , if years later you wanna change it back you can remove the IHS and install the heatsink directly on the die, all of this is because intel doesn't wanna solder the cpu and puts thermal paste instead 'cause it's cheaper
Nahuel Cutrera I've seen pure gallium weaken aluminum so much that you can tear it easily; a gallium alloy might not be a good idea around an aluminum alloy.
+spaceminions this is the one I used
www.amazon.com/Coollaboratory-Liquid-Thermal-Interface-Material/dp/B0039RY3MM
not sure if it contains gallium,
and here there's a bunch of guys doing comparisons only they used the liquid metal on the IHS too ...
www.overclock.net/t/1351984/coollaboratory-liquid-ultra-vs-liquid-pro-vs-phobya-liquid-metal
Nahuel Cutrera I guess there must not be any on the cpu/ihs itself then, if all those people used it without trouble. Cool, literally!
Although the results didn't seem very conclusive, I appreciate you trying with this series and hope it continues.
Not sure why I watched this...
+Stephen Cortez yeah me too i dont even have a heatsink
+Matthew Smith what is a heat sink
Simple Engineering how did you watch this video without knowing what a heatsink is
+Matthew Smith lol
+Simple Engineering im 13
You have to lap the IHS in addition to the heatsink to have any results. It doesn't really matter if one is flat if the other isn't. I have lapped my CPU several times before, and it definitely helps, but the softer copper is prone to warping and may cancel out or be worse over time with pressure from the mounted heatsink.
I ask this question every day when I wake up.
Ohh my good i do too
That Guy Good is god. Good created the universe, and it would be god if Good would send me heaven.
That Guy Oh good, this is terrible.
That Guy Dont you know good? -_-
I sharpen knives and a tip you learn quickly to see if you have leveled a surface is to cover the surface in sharpie and do a couple passes against the stone or abrasive. The spots making contact against the abrasive will have the sharpie removed, the spots not making contact will not have it removed.
Also, since copper and aluminium are much, much softer than hardened steel, I would suggest starting at 400 grit, then moving to 1000 grit and finishing with 2000 grit.
Making an evenly convexed heat sink with an abrasive by hand would be near impossible.
I heard that polishing your PCBs will make the electrons go faster.
I LOVE Luke's "WHATEVER" videos. He is the myth-buster of computer builds -- not that he's likely to change anyone's mind.
It would be nice if he'd say something about fan speeds, does he keep them at constant speed or not?
sooooo in other words.... whatever?
+DeutscheJunge Why didn't he just make it a video about how CPU lapping doesn't matter? Instead he made a video about "I'm going to prove why CPU lapping works, oh wait I'm stupid". He could have done all the research, reached his conclusion, then made a video about it... Is it just me?
+DeutscheJunge yep like most of the small things you can do to your computer to make it cool "better"
Nice, you did some good research. That sandpaper progression is nearly exactly what I use when sharpening a knife to a mirror finish. Except I go on to leather strops after 1,600 grit.
Linus Tech Tips Whatevershop
This is a metalography process and its done just like this in university! Congrats Luke! It matter a lot!
NASA created a ball bearing so smooth that if it were the size of the earth, the highest imperfection would only be a few metres high. It is the smoothest object ever created. Ofc, they did it with LazerPewPew. You should test a heat sink made by NASA.
Take a look on Veritasium's channel, he posted a video about the roundest object in the world. It's his most popular video IIRC. Basically, it's a metal sphere about 6" across, and if it were scaled up to the same size of the earth the highest mountain and lowest valley would only be 14 metres apart (7m above and 7m below sea level). Impressive considering the Marianas Trench is 10,809m below sea level and Mount Everest is 8,848m above sea level. NASA did have the record, but this one has it beat.
Luke: I find the "Whatever" videos to be extremely useful. Like the cable management one. I always fretted over cable management but have never had a glass side panel so it was always about air flow never about looks. Imagine my joy after watching your video and finding out that cable management equals squat! Now I get hours of my life back for every build. Thanks so much.
See - - even the whatever videos have a lot of value to someone.
Curious as to why you didn't lap both the CPU and heatsink? Part 2?
I feel like intel just doesn't give 2 craps about their IHS (as well as thermal interface between the die and IHS, but thats another story). Cpu cooler manufacturers make their coolers nice and flat and shiny to sell more of them. Lapping the cpu i feel would net better results. But of course lapping the more expensive part, and voiding the warranty on that is an undesirable thing.
I believe my cpu's IHS had some issues with "flatness" as i had some very high variation in core temp. I'm custom water cooled with an EK supremacy evo and 4790k. I didn't lap my cpu, but when i removed my IHS and got the "naked ivy" mounting kit for the supremacy evo, my temps went down drastically, as well as my variation in core temps. (i was delidded with CLU prior to installing the naked ivy kit and was not happy with my temps/core temp variations).
I do still have some core temp variation (core 2 runs ~5C hotter, which is better than its previous ~10C) and overall temps dropped as well.
Luke
Get a granite surface block, and some prussian blue, put the prussion blue on the granite, wipe leaving a thin coat, then rub the cooler plate over the granite plate, and then look at the cooler plate to see if there is a uniform coat of blue (ie, it's flat) or blue on the outer periphery (ie, it is concave) or it is just blue over the center (ie, its convex). Having sorted the cooler by the plate geometry, then try the lapping again on the concave and convex ones and test to see what happens Grizly sells an affordable 12x12 granite surface plate if you need a source, or just check with a local machine shop, they will have them and probably let you do the tests on their plate
John
Science is negative results too. Thanks for taking the time, Luke.
I love your workshop videos by the way! Please keep doing them, the "how many fans should you put in our system" one is my favorite.
Just fill that in with thermal paste. That's what it is for!
Well, TIM is consumable and not cheap, but polishing is quite forever.
Hey Luke, I just want to let you know I still appreciate and enjoy you doing these workshop videos. Still super interesting to see it in action since I don't have the resources to do it myself.
NYESSSSSSSSSSSSSSS
K
+ClaudeAlin you are a maniac
bitc!!!!
You know I respect people who didnt like this video and idea but I am actually very pleased that these guys are working really hard and undeniable enthusiasts in this community. Everybody can give average ass info about any subject but these guys are putting their soul into what they are doing and we all as a community should appreciate that. They are trying to find a balance between usefull (applicable) and "huh? interesting, thats cool to know!" type of content. All and all this is a business and they're gonna get that sweet youtubemonay but we all should admit they very well deserve it.
Shouldn't the thermal paste compensate for imperfections? Or is this about bad CPU lids?
think making the Surface rougher would actually increase surface area and together with the thermal paste could improve performance....
+Djhg2000 It does, but compared to solid copper, TMI looks pretty much like a thermal insulator.
+Friederrich Eggeling if you have a 2x2cm cpu and a 2x2cm heatsink you have 4cm sq of surface area maximum once any gaps are filled in with thermal paste.
if the heatsink had a sawtooth arrangment that perfectly interlocked with a sawtooth design on the cpu with no air in there what so ever (unlikely) then you could get the area over 4cm sq however the IHS on the cpu and base of the heatsink still have the same cross dimensional area so will still have the same internal heat conduction so it still wouldnt be any better.
FWIW I heard in college that some BMW cylinder heads were machined so flat that they didn't need a head gasket between the head and block, never seen it first hand though.
+Djhg2000 The problem is air. If the surface is too rough, or especially concave, you can trap tiny pockets of air in between the paste and the surface of the cooler, making it less efficient.
+Toms Tech
Engineers have designed heat transfer solutions for cooling for years now. There's a reason there's no jigsaw.
Adding in jigsaws would actually be less beneficial to thermal conductivity, because you'd increase the volume of copper the heat has to go through before reaching the liquid in the heatpipes. Heatpipes rely on a phase-change design, with liquid that becomes gas, and travels to (one of) the other end of the heatpipe way faster than heat can travel its way through copper over the same length. Less volume of copper is better for heat to reach the liquid faster, so flat surface is better.
Hey, you guys should pioneer research on these topics... be the first company to provide definitive data on these workshop videos.
Use the forum to gather lots of data and publish it...
Keep these workshop videos coming! They are wonderful, and I appreciate them! :D
Whatever
Im a machinist , I do this from time to time at work with hydraulic pump parts during rebuilds. When lapping if possible it should be stressed in the exact same way it would be in use during the lapping process. Once mounted to the cpu the base of the cooler may be stressed and bent slightly from said stress. Both aluminum and copper are soft very ductile materials and may be effected greatly depending on the stresses involved in mounting . I dont know how u would solve this on a cup cooler but just wanted to note something I knew about lapping that may or may not be critical in this situation.
Cool.. another video that boils down to just an ifixit ad and content that doesn't get us anywhere by the end.
+Daldaren I just itched by scrotum for the first time today and it felt really great.
john papple ikr?
A quick guide to ultimate contact.
1: Put the heatsink on a drill press.
2: Place the CPU under the heatsink.
3: Start the drill press.
4: At a relatively low drill speed setting (don't allow the drill to pull your heatsink apart), gently lower the spinning heatsink onto the CPU.
5: As the two components are grinding on each other in a circular motion, any imperfections are being flattened, and any curvature is being aligned.
6: Apply enough pressure to heat up both components, making sure not to overheat the CPU (give it a break every minute or so.)
7: After about 30 minutes (allowing for breaks), your two components should now fit onto each other as snugly as were they 1 piece of metal.
Only those two components will ever fit each other though. So make sure you have a perfect heatsink for that CPU that you aren't going to want to change for a while.
that's what the paste is for wow !!!!
Just speaking as a machinist. (I grind parts with a tolerance of .00005 inches)
When lapping or checking for high spots by *hand* you should always try to create a figure eight motion on a flat surface. This limits the creation of a convex shape on the material. Turning the heat sink 90 degrees just makes it convex in two directions. The removed material will always be in between your part and the lapping medium you are using. Which lifts up your parts and causes it not to sit flat.
IMHO use some blueing compound on the mirror and lightly press the the heat sink on the glass. The ink will only transfer to the high spots. Sand those down (or use a stone, which probably would be better) and repeat until you have 85-90% transfer. This is usually good down to .0002 flatness.
U forgot the CPU luke. Otherwise thanks for proving a point.
I know others will say this but let me repeat it.
The WHOLE POINT OF THERMAL PASTE is to fill in those imperfections and the curves.
If their is some major curves than lapping might do a little bit, mabye.
If it was something easy and quick it woukd not really matter, everybody would do it.
But it is not, it takes lots of time, takes some work, and can possible void a warrenty.
Unless u are some crazy overclocker i dont see any reason to do this.
I will admit i have never done this so i dont know from experiance, so if it works for u ... school me. But i dobt see any reaso ti do it
THANK YOU! for overvolting when testing anything to do with thermals!
LTT is becoming Buzfeed. They have done lots of good videos in the past but there are no more topics to cover.
+Bjeah_ well I just think that there is less activity at this time of year
just wait till about Q3 of this year. they'll have a metric ton to talk about
+EDDSWORLDMEGAFAN yep thanks
+Manav Anand I'm not very good at explaining :}
+TheSupergary69 Please don't compare this to the parasitic and toxic filth that is Buzzfeed and their vile videos.
If you lap your heatsink base, you also need to lap the CPU IHS.
i was already doing it back a few years ago on a 3.9Ghz E6750 under a CM 212 and it was a big improvement
Im currently using a lapped/dellided 3770K @ 4.9Ghz under a HN-D14, and this is the perfect setup for using liquidmetal !
This, they seemed to forget that you need to lap the CPU heatspreader.
NCRranger24 - Gaming and More! by experience i can say that CPU's IHS are never flat :D
I don't get the high amount of dislikes. Luke obviously put TONS of effort to produce this video. Maybe change the name of the series to Myth Busters or something like that.
+Ricardo Martins The new trend is for people to dislike any video they do lately, regardless of quality. These things come and go fortunately.
Luke is just too horrible! I've been subscribed (not anymore) for only one month, and he just bores the hell out of me, so annoying. Can't stand him......
The heatsink isn't always what you want to lap. The CPU lid is. If you delid your processor and lap it... you'd be happy to see it DOES make a huge difference at high temps . :)
Very poorly done lapping, You should hold by the base, not high by the fins, best to press the base itself gently in the very centre of it, and the result should be mirror like which this wasn't. After 3h this was very shitty result
+3majsie maciek Not really, he could have used up to only 600-800grit paper. If he wanted mirror he could have gone 1000+. Through all the cpu's I have lapped- going for the mirror finish has any negligible difference.
However, he should have lapped the IHS on the cpu if he wanted to see any significant difference.
I thoroughly appreciate this video as i was going to pick up sandpaper tomorrow, for my hyper 212 evo lol
I think you guys are going into a lot of these without doing research, basically making another useless "conclusion" which is just adding to the already mind boggling confusion that exists around the topic. If you want to do something meaningful, do a full investigation like that of SPCR and gather a large amount of information and sift through it to get the useful stuff instead of trying to stick to some upload schedule by taking shortcuts and making useless content.
I have an ARCTIC Freezer XTREME Rev. 2 that originally came with the worst contact surface I've ever seen. The surface was full of scratches and not flat at all. In fact when I placed the heatsink on a table, it could almost make it swing from left to right because of the curved surface. So I decided to lap it and, even if I didn't finish the process because after 3 and an half hours I got really tired, I saw a considerable improvement on the performance. Before, my FX-8350 was thermal throttling after 15-20 seconds in full load but now it doesn't thermal throttle at all and the max temperature reaches about 60 degrees celcius. And there's a lot more work I can do on it.
What is the point of this discussion? Not a productive use of time!
Money for views. That is all.
+Peter M The point is to INFORM you that it isn't a productive use of time, at least according to this individual test.
Lapping the heatsink is one side of the matter. One would also need to lap the CPU Integrated Heat Spreader.
This would then ensure that both are relatively flat surfaces, thus allowing for maximum contact and transfer of heat (with minimal use of thermal paste).
As to whether it would have any actually noticeable effect, can't say as I haven't bothered/needed that level of performance.
I do recall a few forum posts around the net, having mirror polished both the heatsink and CPU IHS; to the point where the CPU stuck to the heatsink due to suction.
I'm sorry this is way too much effort to get a few degrees difference in temps. much rather just spend the extra and get a good aio to get a more significant change.
Honestly, Luke is why I'm subscribed to linustechtips
a thermal paste's reason of being isn't actually to cover these microscopic groves to have an even heat transfer?
+Lorentz the annoying Llama Have you seen the state of the CPU they use? Due to their poor taste in thermal paste it is scratched up REALLY bad. Lapping a heatsink then using that CPU won't make a difference when their CPU heat spreader is in a much worse condition then the heatsink before lapping.
+Lorentz the annoying Llama They do tend to liberally apply TIM before putting a cooler on, just a rice grain size drop in the center is all that's needed not a stripe clear across the CPU like they do. The TIM is probably preventing any decrease in temperatures.
+SweBass Yes you're exactly right, that is the thermal paste's job, so this is pretty much stupid to do anything for those scratches, groves and other things. but a flat surface is important, which sanding will not achieve without HOURS of work, You will want to instead take it to a machining shop and have them flatten it.
Sean Anderson You can get flat results from lapping, hell you can get mirror finish results without the slightest distortion with lapping like in the video. It's just doing a good job takes time and practice which were obviously both missing from the host of the video. When he was lapping on camera you could see some shudder once when he changed directions, what's shown in the final video is the best footage they have, there was that one screw up in the best footage they had so I will assume there were far more and far worse. Taking it to a machine shop isn't needed, just time and patience as well as practice blocks of copper.
I for one appreciate the "Whatever" results. People argue endlessly and venomously over these things. These videos remind me that people will go to great lengths to argue something that doesn't matter and helps me ignore it all.
i call BS on lapping making any diff unless your heatsink had some kind of manufacturing defect to begin with. All those minor imperfections will be filled with thermal compount once you seat them together filling in any imperfections / small voids.
I feel like i have to say this. I SO FREAKING MUCH LOVE THIS KIND OF VIDEOS, no matter the result!!! :D Keep it up Luke :D
What about different liquid cooling liquids. It makes a difference in cars due to the different thermal capacity of water vs ethylene glychol vs polypropylene glychol etc. Maybe another option exists, does methanol +water (windshield wiperfluid) or something similar work better? Just a thought. I switched my race car to polypropylene glychol (sometimes found at walmart as RV antifreeze) and my engine runs cooler yet my coolant temp is higher because its pulling away more heat, but it doesnt flow as well so you need a bigger radiator. So might be a fun test and should show results for sure.
+Merp Water cools better than just about anything, including automotive coolant/antifreeze and rv coolant. The additives are to prevent corrosion and freezing and in general the higher the additive content, the worse the cooling. The only thing that can really make water cool a little better is an additive that reduces surface tension.
garfunkle187 Generally true, water can move more heat, however water also microvaporizes around very hot things causing cavitation preventing good contact with the hot object reducing colling effectiveness. Also water has to be pressurized to take higher heats where other higher boiling points do not have to be. Not sure if it matters for CPU cooling but would be interesting, but you also cannot running straight water in coolers or you will get corrosion or mold. Also many overclockers claim Methanol is better than water, and what they sell in water coolers is glychol based and many claim polypropylene is better. So I think it would be a valid test.
I agree the test would be valid simply because people do use those different coolants, i just wanted to make the point about water being a good coolant inherently.
To clarify further: Yes some liquids boil at higher temps than water, but ANY liquid will boil at a higher temp at higher pressure and vise versa and most coolants do NOT have a higher boiling point at a given pressure than water. The cavitation you speak of is what is helped by the surface tension reducers (eg water wetter) i specifically mentioned. Regardless, ANY liquid will be subject to these same effects.
The main difference between ethylene glycol and proplylene glycol is how they react to different metals, specifically nickel, aluminum and steel.
*and copper
garfunkle187 True, although proplylene glycol does have a 370 F boiling point. And ethylene glycol is 387 F So quite a bit more than water, so you do not need to worry about pressurizing to increase boiling points :)
What you did in the video would be considered simply wet sanding. My understanding of the lapping process would require lapping compound (a paste with abrasive grit in it used commonly in automotive valve lapping). It further would require lapping both contact surfaces which would include the CPU lid. I don't know if it would be a good idea, since any remaining compound might foul the CPU, but it may be worth a shot! I'd love to see a quick revisit with proper lapping technique.
Really scrapping the bottom of the barrel for ideas huh?
+Hayden Ross More like scraping the bottom of their heatsink.
I see what you did there
+Sam Watson NICCCE
+Hayden Ross No kidding, Linus Media Group has really been going downhill with these strange ideas and just plain boring ones. I don't think I would ever care about the shape of my heatsink. I'm not designing a brand new computer in my garage, I'm simply going to be spending some money every now and then upgrading or building an enthusiast grade machine.
+Hayden Jacobson now I've lapped my g3258 the actual cpu itself and it did shed a few degrees not much but a few. I've never heard of someone doing it to the actual heatsink. Only seen people do it to the cpu itself.
lapped my big typhoon like 8 years ago when I had an Opteron 165. Dropped load temps about 5 or 6 degrees. The heatsink was slightly concave from the factory. Lapping made a huge difference for me. I also removed the IHS. That did another couple degrees as well.
Kinda pointless doing this..Thats what thermal paste is used for, filling in the gaps, to spread heat.......
+Aoru Isn't thermal paste a lower-thermal conductor than nickel or copper? If so then yes this would matter (although probably a very small amount, almost unnoticeable amount). However if the thermal conductivity is the same then, from my understanding of the laws of thermal dynamics, the edging should not matter as the transfer of heat energy would be the same.
+Aoru Thermal paste fills into microscopic cracks, using it to fill macroscopic imperfections (because of either surface being curved) would result in poorer performance because it is actually a worse conductor than copper or aluminium.
***** Not always. If thermal paste truly is useless you wouldn't waste money on it and just put the slab of heatsink directly on the chip. The paste is there to transfer heat better to the copper and aluminum that dissipate heat. As an example, GPu vs CPU. CPU might survive without paste, but not a GPU. So thermal paste can be a macroscopic factor in that sense.
asymsolutions Depends on the kind of thermal paste. Some thermal paste are conductive, while others are non-conductive.
Aoru All thermal paste is a thermal conductor. That's the whole point
Wow a whole 5 degrees it must make so much difference. Totally worth about 5 hours of my time grinding down my heatsink.
+GDbluerebel 5 degrees celsius is actually quite significant in the electronics world, especially when overclocking.
It is significant, but my gaming pc uses a crappy HP heatsink and it runs fine. I don't see how 5 degrees is worth hours of my time rubbing it on a piece of sandpaper.
fell asleep after the first 2mins... Jesus what is going on with their videos lately...
I've made today a very good experience with lapping my cpu & heatsink. I had problems with overclocking my Q9650 past 3.4 Ghz. When stressing the system with Prime95 after overclocking it crashed due to too high temperatures. It took me the whole afternoon but i would say it was worth it. Temperatures dropped about 5 degrees Celsius at minimum and the cpu now is stable in prime with 3.6 Ghz (air cooled). Probably results may vary for other people.
Luke's time wasted again.
Lapping is something I do on a regular basis so I'm going to give this a shot on my video card heat sink. It rests directly on the silicon wafer so black polishing should provide maximum contact. Thanks.
is it me or the channel is getting boring?
It's linusTECHtips not linusENTERTAINMENTtips it's meant to be informative!
Heat transfer through conduction is pretty intuitive, as in contact area helps. We could have assumed that when using a good paste (which we always do) it's not going to matter significantly if surface is rough or glassy. If for some weird reason you weren't using paste, it would likely matter more since glassy and flat should slightly increase contact when 'dry'. There are ways to measure pressure in different areas under mounted tension and of course it follows that the best shape would be one that 'deflects to even pressure' instead of having higher pressure around the edges. But of course that would all depend on the size, thickness, shape and material of the face, and the cpu heatspreader. It's easy to imagine that often there would be so little deflection that flat would still be best, but only measurement could tell. I think it's hard to beat a decent all copper air design for bang for the buck/trouble/etc. I'm looking forward to Pascal release tomorrow.
Come on... stop with all this irrelevant DIY projects.
just checked my cooler(noctua push pull) it had a bent in it Cpu -> ])
WHATEVER
Kidding, glad to hear you gave lapping a shot! I thought the it was the CPU that is more known for being lapped down to the copper though.
I figured the surface wouldn't matter since the thermal paste is there. Thermal paste is basically meant to fill all of the gaps between the two surfaces with a very conducting substance, instead of it just being air. It may have made a bigger difference on older models where manufacturing quality was lower, and maybe the surfaces were really bad. At that time thermal paste might have been worse. It would also make a difference if you didn't have thermal paste, for some reason. You would then need both surfaces to be as smooth as possible to make the best contact and allow as little air between as possible.
For those that don't know, air is a thermal insulator, so trapped air means bad conduction. That's by double pane glass works well.