Making Intel's Worst Product Better.... And Also Worse
Vložit
- čas přidán 14. 05. 2024
- Receive a $25 credit for Ting Mobile today when you sign up at linus.ting.com/
Create your build at www.buildredux.com/linus
Intel's i9-12900KS is an incredibly fast processor... but what if we made it faster?
Buy a RockItCool Copper IHS Kit:
Buy an Intel i9-12900KS: geni.us/CulTHl
Buy an Asus Maximus Extreme Z690: geni.us/Q0WW6
Buy an Asus Strix RTX 3090: geni.us/sTtsH
Purchases made through some store links may provide some compensation to Linus Media Group.
Discuss on the forum: linustechtips.com/topic/14255...
► GET MERCH: lttstore.com
► AFFILIATES, SPONSORS & REFERRALS: lmg.gg/sponsors
► PODCAST GEAR: lmg.gg/podcastgear
► SUPPORT US ON FLOATPLANE: www.floatplane.com/
FOLLOW US
---------------------------------------------------
Twitter: / linustech
Facebook: / linustech
Instagram: / linustech
TikTok: / linustech
Twitch: / linustech
MUSIC CREDIT
---------------------------------------------------
Intro: Laszlo - Supernova
Video Link: • [Electro] - Laszlo - S...
iTunes Download Link: itunes.apple.com/us/album/sup...
Artist Link: / laszlomusic
Outro: Approaching Nirvana - Sugar High
Video Link: • Sugar High - Approachi...
Listen on Spotify: spoti.fi/UxWkUw
Artist Link: / approachingnirvana
Intro animation by MBarek Abdelwassaa / mbarek_abdel
Monitor And Keyboard by vadimmihalkevich / CC BY 4.0 geni.us/PgGWp
Mechanical RGB Keyboard by BigBrotherECE / CC BY 4.0 geni.us/mj6pHk4
Mouse Gamer free Model By Oscar Creativo / CC BY 4.0 geni.us/Ps3XfE
CHAPTERS
---------------------------------------------------
0:00 - Intro
1:10 - Build Redux
1:22 - Isn't the IHS already copper?
3:30 - IHS Removal
5:40 - TIM Removal
6:52 - Polish
8:02 - Liquid Metal
9:10 - Relid
10:40 - Did it work??
12:00 - Testing
15:27 - Game Testing
16:36 - Ting!
17:15 - Outro - Věda a technologie
This is quite literally the most Linus video in a while, the number of drops, incredible
I usually just chuckle when Linus drops stuff. But during this video I got so frustrated that I actually threw up my hands at one point. HOW DOES HE DO THIS! I remove my RX580 to clean it and treat it like it cost 3 million dollars.
He had to make up for the chip fab video where he was not allowed to touch anything, let alone drop stuff.
I feel like the editors have been editing out the drops for some of these videos, he went like 4 videos without dropping anything of value. I refuse to believe linus can not drop anything for multiple videos consecutively
He probably does it on purpose. Know your audience. Linus media is a powerhouse not a small goofy quirky studio
all the best linus videos have Alex in it
"can you believe they wouldnt send me a second unit after the review?"
"What a bunch of jerks"
"I know right?"
Intel, sponsoring the Intel extreme upgrade: "yes please, insult me more"
You don't mention Linus pulling the IHS off the CPU then dropping it on the floor. Why not send him more ludicrously expensive hardware? ;-)
@@travisbic it was a joke dude
Don't forget the Intel factory visit too 😂😂
Linus did the same a few years back with a much more expensive intel xeon processor and damaged it. Unsurprisingly intel refused to send a replacement then.
What are you doing step-reviewer
I just can't get over Alex's soul leaving his body every time he has to tell Linus
"Don't do that."
a man of wisdom about the sandpaper wank
@@l33tsauce79 clearly speaks from personal experience lmao.
With the massive premium Intel charges for the KS variant cpu, they should include a machined copper IHS kit like shown free.
That and much more. Like guaranteed all core clocks for a specific voltage. Like siliconlottery did it.
The KS series is like an amazing wagyu steak packaged in a terrible paper bag with holes in it. Something you might expect a pretzel to come in.
Honestly it should be solid silver
@@toby1248 Would gladly pay for a pure silver IHS.
@@aluckyshot they used to exist a few years ago. Not sure why they aren't available anymore
I love that Idea, sounds like an amazing way to get your investors to raise their pitchforks
Linus: rips a top of the line Intel CPU apart
Also Linus: Why won't they send us more free processors?
@@Noah-lj2sg but they didnt?
also classic Linus dropping it
@@vsnusv7555 They said Intel would probably send more in the FUTURE
I mean, what is Intel extreme tech upgrade?
I read Linus “rips a line on a top of an intel CPU”
Linus: "I wonder why they won't send us any more CPUs..."
Also Linus: Drops the CPU.... twice
Exactly! That's why they should send him more! So he has spares.
in less than a second
I panicked so hard lol.
"No no! Do it on the floor damn it!"
@@UltimatePerfection really good idea
And it's not the first time. Remember he dropped a $10K Xeon (off camera, sadly), damaging one memory channel. Intel didn't send another...
Seems like a process that should be implemented by CPU producers on their top-tier products.
I actually find it insane that top of the spec products don't even, consider outsourcing cpu lids for a product that just only realistically exists because a competitor is outperforming you.
@@namonaite Not insane at all. That's how halo/flagship products work. Why do people forget that Intel wants high margins.
but u also can said the other way around if they gone that way, that u need to pay more for non crap IHS
For real, I don't buy the explanation that manufacturers couldn't scale this up to mass production when they are producing chip dies that require far lower tolerances than machining smooth IHSs.
@@brice0403 it would also cost them far less because scale.
This may not be the right place to say this but I want Linus and the whole team to know that the love of my life looked up to Linus and loved everything about you guys. We would watch your videos together, and he would explain everything because I'm not techy. He was taken from this world way too soon. Thank you Linus for bringing a smile to his face everyday. Every time you all drop something, too of course.
So sorry to hear, but it’s comforting to know that LTT could be a way to reconnect with those memories.
Finally something in Linus Tech Tips that I might be able to afford: the copper heat spreader.
🤣🤣😅
Why buy a copper heat spreader when you can buy a gold one?
Not if you break your CPU trying to remove the original heat spreader
yeah i might be able to afford it, but i have got zero chance of replicating what they did in this video. Too complicated for my sausage fingers.
@@afriedrich1452 I heard diamond is better! /s
"Your house might burn down."
"But look at these reflections."
That's right; priorities.
"PRIORITIES" Indeed
this comment gave me HEY THIS IS AUSTIN vibes
I can get behind those priorities
Burning the house down might raise the temp a bit more 10 degrees C (at first), but once you're rocking outdoor PC lifestyle, the Winter temperatures should let us get even more performance out of it.
@@FranciscoTChavez i like how you are thinking
Weird to find a compromise in a no-compromise CPU. I wonder how much more Intel would have to charge per CPU if they used a tighter tolerance heat spreader for the massively high end CPUs only. If it doesn't matter for 99% of all chips okay, use the other process, but for that remaining 1%, it seems like billions of dollars of R&D could be bought cheaply simply by refining this one component.
I hope someone with some power and vision at Intel watches this, since AMD is looking to eat their lunch again with Ryzen 7000 and they could use every advantage.
Or someone at AMD could watch this, improve their own high-end and continue embarrassing Intel in the future.
Tbh offering a "top bin" for the higher end CPUs at like 400 bucks more without an IHS or with a lapped IHS may be interesting. I wouldn't buy it but people looking for absolute best performance may want it.
Or, like, design the CPU to use a somewhat reasonable TDP? This thing draws more power than my entire computer
8 months old comment, but anyway, i think the issue may be that the cost may be higher than what the price of buying that lid would suggest and that in practical applications, as shown in the video, the gains are minimal to non-existant:
1)Because of the random factors during manufacturing process (and manufacturers designing around the existance of said random factors), you only get to know if any particular chip can be made into a top tier/mid tier/low-end processor after evaluating (i.e. (mostly?) testing) it. Which means that you have to either
a)disassemble, clean and then reassemble already working processors (which, besides costing money, adds some risk of breaking them),
b)or you have modify your production process to have your testing done before fully assembling the processors, which will probably increase production costs for all of the processors, which, given that only a very small percentage of the processors would actually use that super flat IHS, would effectively make installing that IHS an expencive process (if 100 processors with standard IHSs cost, lets say, 1000$ make and 99 "standard" and 1 "super flat" processors cost $1050, then that super flat IHS effectively costs 50$).
2)You now also have to manage having 2 similar items in your facility, which increases costs of running said facility (i.e. the production cost of all of the processors), which, same as above, means that it is, effectively, really expencive to store those super flat IHSs.
main issue with this is that cupper and gallium will react and liquid metal will "dry out".
Maybe precuring cupper surface with LM will create layer of CuGa2 and it will not dry out anymore, but I did not test that.
For 10celcius, intel could have made a special heatspreader just for the KS mostly for the money. At least it worked flawlessly, nice job.
That would really justify the KS name
why did everyone forget about that concept PC build that had copper wool as a heatsink for processors that was as cool as water cooled systems ? was like a wire mesh of copper wool and it disperses heat way better than fins
@@girlsdrinkfeck which video was this?
I’m excited to see how it holds up long term.
@@girlsdrinkfeck The game is not about heat dispersion for a long time now, CPU cooling is like 95% about heat transfer from the die/IHS to a few milimeters away. That's the bottleneck, especially because CPU/chips are small. Once we can move the majority of heat just a few milimeters from the IHS, it's almost trivial to get rid of it.
I want to see how Silver which is more conductive than copper would perform. Like Linus and a few other youtubers could afford to show testing on Silver heatsinks. I guess it would perform 7% better than copper.
Wouldn't it bend?
Diamond - 2000 - 2200 W/m•K... They should Use Diamond 💎. Better than Silver 5x times 🤣😅
Not too much difference, although it is probably a bit better.
But consider that a copper heatpipe is about 20-200 times better thermally conductive, and you'll realize why hardly any coolers use silver. Ease of manufacturing and corrosion resistance of copper metal play a role too.
Why yes, silver is 7% more conductive than copper(that in this case would at best give you a couple degrees less), it also costs 50 to 150 times more per pound
Do I need to explain why it would be such a stupid thing to do, or you think you can manage that for yourself?
Also, metal liquid is mostly gallium
Silver and gallium make an alloy, Ag2Ga, that can happen spontaneously at room temperature
This alloy increase its electrical conductivity, against its thermal one, enjoy
just looked into that and seems like it would need some sort of plating cause silver alloys with gallium which is in liquid metal, which would ruin it
Man, this takes me back to when I had a phase where I lapped the mating surface of my heatsink as well as the top of the IHS so there was less thermal compound needed to fill the gaps. I certainly didn't get a 10c improvement, but it was pretty cool to have a near-mirror finish on your CPU 😎
That was right in the era of having a DFI LanParty motherboard with UV reactive PCI slots. Wish I had pictures of that setup still!
The stuff they used to get rid of the solder is actually a Gallium, Indium, and Tin alloy called “Galinstan”.
Last week: “Intel let me see their cpu production”
This week: “Let’s exploit and rebuild Intel’s fastest gaming cpu”
next week: "i established a new startup in silicon valley"
Although modding something that you legitimately own is not a sin. It's your CPU after you own it. Not Intel's.
@@bananab3813 Next to next week I am making my startup public, invest now.
When will you buffoons understand that Intel isnt as salty as you people
pissing off Intel month
"This isn't abuse, we're making it faster"
I get the feeling Alex blew up a few engines back in school
Just 5 more psi of boost…
With two strokes, a little lean on fuel makes a little more power
Guess the catch
@@ruikazane5123 knock knock!
@@ruikazane5123 what?
@@mh13mini who's there? mr rodney!
Few hints (from old delidder): 1. Cut the cotton part from ear-stick-cleaner and spread thermal grizzly with plastic sharp part (avoiding any cotton hairs that can stick in), and 2. Don't remove all glue from cpu pcb just gently clean it with isopropyle alchol, (it's okay to have part of it left on pcb) then apply medium-thin silicone glue / whatever you use. Tested many times, 100% best way. And gift tip ... dont wait too much for glue to settle, put in cpu socket and lock it, left it there to settle under pressure of locking mechanism, that way you get correct thermal compound distribution over cpu die + IHS.
10 extra lower degrees of temp for just 40 dolars is actually amazing
“Don’t want to get that in your eyes.”
It’s almost like wearing eye protection/PPE would be a good idea…
Dont be ridiculous thats way too sensible
OSHA staring motherfuckerly O_O
Don't be silly. We are in the "masks don't do nothing" days. Ppe is seen as infringement of rights now.
Linus dropped the eye protection goggles and they cracked or sum
Everytime the CPU boosted higher with the temperature dropping. Now what I want to see is a situation in which the CPU doesn't draw more, to see the true temperature difference. Maybe limiting the CPU clock to whatever the stock CPU was running at. I think it should actually reach the 15°C mentioned online.
Limit the CPU clock in both tests, that way all factors will be the same.
I think this is really what they should have done before thinking about 20W extra, higher clocks, overclocking or whatever
Also.. if they want to have a legitimate comparison, they probably should have used the same cooling system..
As impressive as the temp drop is, the fact that they used different coolers for the 2 stress tests invalidates everything in my eyes.
@@longjohnsilveri3733 I think they used the same setup for both.
nah, nobody cares about the temp of your cpu. The purpose of cooling a cpu is to get more performance out of it, so testing the performance is literally what matters. If a cpu performs the same at 100c and 40c, why would you spend any amount of money to get it cooler? It's not going to help longevity either, you can run a cpu near tjmax it's whole life and you'll need to upgrade your cpu because it's slow as shit due to the time that has passed before anything actually happens to it.
What I really would have liked to see is just how much difference lapping the original IHS would have made before they removed it and installed the replacement IHS. They could also have tried delidding and replacing the solder with liquid metal to see if that made a big difference. As it is we don't really know if the improvement is because of a flatter IHS, using liquid metal as a TIM instead of solder, or if it's the new IHS that makes the most difference.
RockIt Cool: "It's really flat!"
Me, having had to remelt 300kg+ of copper/titanium alloy because parts deviated 1/1000th of a mm over 85mm: "😑.... you should try aerospace MFer!"
Thanks for watching 🤗
Send a direct message right away,I have something for you.🔝🔝🔝🔝
where are you making air locks at
I just finished rendering 5,400 frames of 3D animation, a second less each would have saved me 1.5 hours of downtime so, under the right conditions, it's totally worth it!
Can't you render on gpu though?
@@matejnemec4616 a second less per frame is worth it, either on cpu or gpu --for this video the upgrade was cpu cooling anyway, not the actual cpu
Makes you think about having one of those coppers on your GPU, and putting the whole thing in an actively-cooled liquid bath, to me.
@@davidgoodnow269 precisely! especially considering how sensitive are gpu clocks to temperature these days
Yes, though it's getting to a point where it would just make more sense to have 2 computers render 2700 frames each. Much faster. And after a point, or at a bigger scale it gets cheaper at the same time. Including the power costs.
I remember back in 2013 when AMD brought out the R9 295x2 and Linus said that intel needed a product that pushed the bounds of what's possible and requires watercooling. Looks like he finally got his wish!
Damn even the single R9 290 was a mess to cool. Been running in thermal throttle since day one, had to build my custom WC loop
@@bratisla2040 I had a Tri-X 290 for a while back then. After repasting, it didn't go past 75 degrees EVER.
"Graphine infused liquid coolant"! everyones got to love coolants that instantly short everythink out the momoent they leak.
7:26 Dracula from Hotel Transylvania
AMD: Next generation 3D stacking technology brings top tier gaming performance on high-midrange CPU's
Intel: 300W stock, take it or leave it
AKA - Intel: "Best I can do is 300w stock."
They had the best cooling you can possibly get and this CPU still thermal throttled with stock IHS. What a waste of money this processor is.
@@kamilkilian7861 Not quite the best possible. They have that air conditioner cpu cooler.
Good ol' Intel is just trying to bring liquid nitrogen cooling into the mainstream
@@DarthCiliatus which in it's turn takes 2kw of power and has issues with condensate
Linus: "I can't believe we're already doing this"
Me: "I can't believe Linus is letting himself handle it already"
Back in my day we’d just remove the heat spreader and mount the cooler directly on the die. Who knew that heat spreaders would become so good that you would actually want to mount a custom one to lower temps?
AMD Athlons/Durons are welcoming you ^_^
The nickel plating on the IHS actually does significantly contribute to thermal resistance due to the nickel being heat-soaked. The thin-wall assumption that you spoke of holds primarily for steep temperature gradients, but under nearly isothermal conditions, like a CPU and cooler at steady-state, the contact resistance of the nickel-to-copper interface is actually quite significant relative to the other sources of thermal resistance. I would cite my sources, but it was a textbook I read like 2 years ago and I can't remember what it was called, sorry folks.
His cat standing up for no reason at 13:40 had me dying lmao
For a second, I thought this was going to be another zero compromises build. Still, can’t say no to this.
there are no compromises in this build for me
I just saw a comment just like this
@@Bruh-nb1hj Yeah, they copied me.
@@mineisstupid True, but it’s damn risky at that.
Well it kinda is. just without the case
Pah, heatspreaders! Back in the days we had raw and open CPU cores which we'd chip corners off when putting the heat blocks on! Looking at you, AMD Athlon XP Series!
I feel like changing thermal pads/paste on the gpu may have actually seen results in this setup. My 3090FE had massive cooling gains for memory temps when I did it (about 20 degrees cooler), I think they really have been skimping in the thermal pads.
Having the extra good heat spreader should have been the standard. Intel wants to pump the absurd extreme, they should offer supreme cooling. As if it's not expensive enough already, while there's almost no way to prevent it from throttling.
nhaaa, they should just rework their stuff, all they do is keep upping the power consumption gen after gen, just like ngreedia does. thats great if you live way north but having a space heater with 300w cpu and 400w gpu is the definition of ''playing on a toaster''.
@@jeromegagnon3787 it would be better if they made their general line up better, but it if they are going to make a hot part like this they'd better supply it with ample cooling.
@@Olav_Hansen id rather it come with an ac unit then. the part can stand the heat, i don't .
The stock IHS is domed for a reason: To make sure it will work even with lower quality heatsinks and low quality thermal compound. You can get the exact same result as in the video by just sanding down the IHS flat.
@@jeromegagnon3787 one the many reasons I got a 75w Series S console....The next gen is gonna be 10X my S's 75w power usage. I will keep my old 200w PC for retro/media usage.
15:27 The wallpaper is hilarious 😁😁
I've seen Linus complaining about sticky residue several times. Tech tip: contact cleaner dissolves sticker and glue residue like it is nothing. You can even remove void warranty stickers with that stuff.
Just be sure to use the computer-kind contact cleaner. Automotive contact cleaners sometimes use minute abrasive agents that can ruin the look of some plastics :(
Linus and LTT you should try IHS's made of silver, 24k gold, niobium, and other metals in your new lab. you can get a jeweler to make them using a CNC or laser I would recommend Riva Precision Manufacturing in Brooklyn NY but I know they deal mainly with large orders but I can definitely help you find a jewelry in your area with the right equipment to do it
7:00 It's nice that Linus mentioned us: Polish people also watch LTT.
English is so confusing to foreigners. 🤣
I delided my 8600k, was a lot of fun. I didn't even bother glueing the IHS back on, I just used the mounting pressure of the socket to hold it still. I probably got like a 10c drop just putting the previous IHS back on with liquid metal and lightly planing the IHS + heat sink.
:o
The time consuming part for keeping the tolerances so tight isn't the machine itself nor it's quality (well to a degree on quality of the machine) but the frequency you have to check tolerances and change the tool wear in the settings of your machine, as well as how you mount what you're machining, depending on what you make you may have to choose between something that gives extraordinarily repeatable results and has no play or something thats fast, just depends on what you have to work with because your fixture is definitely important
Hey LMG, I have seen on Jay's Two Cents channel, when he does heat/speed testing on a processor, he will often sand/mill down the top of the cpu to make sure it is as flat as possible. So I can see a 3rd party doing this on purpose, for best surface area contact. Love the channel, keep up the great work!!!
If gold is also an excellent conductor, well I think you gotta test the theory of gold vs copper
Gold is worse in thermal conductivity. By quite a bit.
Actually, silver metal has the best thermal conductivity of any metal at 429 W/mK. Gold is 310, copper is 401, aluminium is 205.
The thermal conductivity of a copper heatpipe is between 10000 and 100000 W/mK.
So a gold heatsink would make no sense at all.
@@ProjectPhysX So... silver coolers when?
A diamond IHS is the way to go
@@Sithhy for a top price CPU silver heatsink actually makes sence. However if you running water cooling, you want to stick to a mono-metal system.
I love Alex's semi-disapproving parent look throughout this video, as Linus drops the CPU multiple times and makes a mess, to the point where Alex just takes over.
@modest mouse colored person he has completely flanderized himself, "this computer legitimately if you have like bad not up to code wiring in your house it might burn down" his schtick has really lost all of its charm.
I thought the thread ripper consuming crazy wattage was nuts. Thanks intel!
Yesterday, I saw the shorts video you made on this, it was so cool and I’m so glad you made a nice video of it today
1:22 y'all pretty much explained Six Sigma manufacturing better than any corporate trainers I been subjected to 🤣 so considering that Intel isn't realistically expecting a perfect finish, I'd take a guess they're wanting an adequate finish and Sigma 3 or 4
6S is even more cringe than 5S. All lean manufacturing is soooo painfully cringe. The proponents and the consultants are the type of people that pump 6S during the work week, and a -pyramid scheme- ... I mean MLM scheme... after hours and on weekends. You'll see them at MLM rallies on Saturday afternoons.
@@littlejackalo5326 what the shitballs are you on about?
I prefer six ligma
Oh boy! Have you heard of LEAN SIX SIGMA!?
@@bigqwertycat I first heard about it in 2009 lol
Petition to have Intel send Linus more free processors to tear apart and to... burn the house down? 😂
Hear Hear!
BUUUURN THE HOUSE DOWN!!!
with flammable lemons
3rd
to…. DROP
Yee nah ;D
Would love to see you guys make a pefect copy of the IHS out of solid silver for that extra couple degrees. The stock metal would prob cost less than what you paid for this
I would love to see the original heat spreader under an optical flat / monochromatic light to observe the interference patterns. Also, how does the part deform when heated?
As electricity usage per square mm goes up, every additional improvement in tolerances will have greater levels of significance (obviously to a point). It’s clear that for the maximum performance units they could do better, but for the overwhelming majority of processors, thankfully, doesn’t matter.
It does raise a question about the value of keeping the prices down on the high end products when quality suffers. Relying on 3rd party improvements has been standard for top-end systems, usually because of some odd deficiency-whether it’s lapping the IHS, package bending (Alder Lake), IHS-die gap (on the interior), de-lidding, die lapping, or use of liquid metal TIM-halo products could benefit from the additional effort, and we could still enjoy warranty coverage.
I personally think paying more for better quality in a complete package off the assembly line is worth it for halo products. I would be willing to pay to not risk ruining something so expensive.
Better quality parts would really justify the KS name and price
Eh. Don't know. The price may not be worth it to them. Computers are not like any other product on the market. They depreciate so quickly. In a year or two the 12900KS will be outperformed by the rest of the high end market, in 5 it'll be outperformed by the majority of the market. Does it make sense for Intel to spin up entire new boutique divisions just for the KS series? I'd say no. A halo chip is for current headlines, it's not like a halo car which could have an impact on the brand for decades. People don't judge chip manufacturers on halo products. They judge them on what they can actually afford to buy. Chips are commodity priced and sold, which we saw over the last two years with the chip shortage. Noone has brand loyalty to Tropicana, if Tropicana OJ is out they'll buy store brand and be fine with it. With how fast chips depreciate it honestly may not be worth it to Intel.
It would be interesting to send the intel IHS to a company that can test its “flatness”.
Its most likely slightly bend now after the delidding.
Why? The real issue is how inconsistent Intels IHS are from one unit to the next. Some might be flat while others arent.
@@randybobandy9828 There is also the issue, on 12th gen specifically, of the Independent Loading Mechanism (ILM) bowing the IHS.
GN tests that.
What you could use to measure it is an optical flat and a low pressure sodium lamp that measures surface flatness by optical interference. It sounds complicated but if you set it up right it's actually pretty simple.
10:34 no new thermal paste?
I challenge you to try combining that Copper IHS with CPu Contact Frame and Diamond infuse Thermal Paste
I actually bought one of these back in 2019 and my temps did go down a good bit, I still have the excel file I made. The core temps went from (90/85/93/86) to (74/73/80/74) using whatever stress test I was using at the time (Probably Prime95).... So they do work pretty dang good.
Does that give you enough headroom to overclock instead of getting a new cpu?
What cpu did you put it on?
And I love that intel sponsor’s LTT no matter what
But Intel didn't sponsor them, they were sponsored by Build Redux and Ting
10:40 nope, you're not tried hard enough to get perfect temperatures... To do this you should sanded the heatspreader and water block up to perfect flatnnes and shine ☝️
flatness yes, shine, no. thermal compound likes a slight texture
@@ljfran2383 No need for thermal compound when the surfaces are as smooth as gauge blocks
damn, linus really turned bri'ish at 1:07
you know when there's tasks requiring dexterity, we're going to get a Linus Drop Tips special, lol
I would love to see this compered to a direct die 12900ks in terms of temperature
This plus lapped OEM IHS. I wanna see how much the flatness is a contributing factor.
if i remembeer correctly applying liquid metal to the die and also lapping the die flat had roughly the same effect but this is a very nice alternative for those that do not want to lap the die
yea never use blend for testing the cpu, that test does a bit of cpu/memory controller/ram. run small and use custom set the max size to the L2 or if you want L3 cache size.
6:48
Linus: Gotta say, the little tissues I dispose of usually don't look like this.
Alex: Sighs like never before.
Linus has some on point jokes sometimes LOL
8:14 Fish Bubbles Sound Effects
You need to invest in a set of gauge blocks. If you can wring the IHS to a gauge block, then you know it’s flat. Same goes for the your coolers touch surface. Having a surface grinder with an aluminum oxide wheel, diamond honed, on had will let you be much more precise than the hand pressure wet wheel method seen before.
6:00
Its a good idea in theory, but in my experience those tape guns are vastly inferior. They have another point of failure AND restrict the way you can use it. It may be fine for a couple of uses at home but try packing two or three hundred packages yourself in a day and you will know the difference. Just get the standard 3M tape gun without that "safety feature" and learn how to not cut yourself with your tools.
Those tape guns may be required by Canadian OH&S.
Intel should definitely still be worrying themselves later this year, now that AMD is fully switching to LGA and potentially brining their 3D cache tech and RDNA2 UHD graphics to do anything into one. After them realizing they’ll be able to break their limits of DDR5 speeds too like XMP, but more insane
Why would Intel be worried about AM5 being LGA?
Man that's gonna push Steam Dieck stonks to the roof.. Hell yeah
I believe Intel themselves will be bringing upgrades with double the small cores alongside higher performance for the big cores. AMD will most likely take the gaming crown while intel workstation crown.
Don't know how lga should make them worried, but 3d cache still needs more work. They won't implement it on R7000
Honestly, lga1700 motherboards are ridicolously expensive when compared to AM4, especally at the lower end. And I expect similar pricing for AM5 boards, because of PCIE gen5 and DDR5 as well as being the hot new shit.
Soooo Ryzen might be a king in performance, but could lose on the value front.
I would be interested in seeing a direct die vapor chamber like in some GPUs and Laptops as an IHS. That would be a lot more thermally conductive than a pure metal IHS and spread the heat even better over the whole area of the IHS.
Actually im not sure it would be. Vapor chambers are used for long distance heat transfere like a laptop or aircooler. Pretty sure when you only need a mm or so that wouldnt even be possible that thin. Also youd transfere the heat from metal to vapor to metal again which should be worse than 1mm of copper i suppose. But im no expert just following my brains logic so pls correct me if theres an expert somewhere here
@@D3nn1s My feeling is that it may still be beneficial, especially as the whole area of the IHS will be more evenly used, instead of the heat being concentrated where the die is.
But this discussion is what would make the experiment interesting! Please Alex/Linus!
@@emmamitchell1582 Dennis is right here, heat pipes work well over several cm plus. Thinner than 3mm and straight copper is objectively better.
@@D3nn1s
Ehhhh heat pipes are used for long distance heat transfers
Vapor chambers are used for large area heat transfer
Or is it?
Vapor chamber is when you want to transfer heat across a huge surface area
9:20 killed that smurf, lmao
I challenge you to try combining that Copper IHS with CPU Contact Frame and Diamond infuse Thermal Paste
At 1:03 is one of the best "Holy crap!"s that Linus has ever exclaimed, hope the editors are clipping it for use in asides in other videos.
init
I put one of these on a 8700K a few years back, dropped temps by about 15C at idle, and allowed me to push it to 5.2GHz, which was pretty fast at the time.
It'd be pretty crazy to see some emulation with this CPU
Yeh
technically any interpreted programming language is emulation
another cpu stress test you can run is star citizen, almost all of it is cpu bound right now. running a stock ryzen 5 3600xt with a 3070ti (yes i know that card is over kill when paired with my cpu... , i do game dev ok ...) but i run arround 40 fps in space and arround 30 when in a city.
"Intel getting close to the sun"
If you had a Pentium 4 PresHot that lives at over 60 degrees (C) idle and over 90 degrees (C) at full send without airconditioning on stock clock, that's nothing new
Would be fun to see an actual car racing endurance radiator on that thing with the baddest fan you could slap on...that gotta be worth a few more degrees, together with a larger water block for enlarged fittings and better flow...don't forget an electric water pump
And a peltier cooler module.
And an air conditioner to blow sub-ambient air through that radiator.
And a blowiematron.
And ihs made of diamond
A Pentium 4 at 90c would've long destroyed itself. I'd know, had one of my chips die on me before the age of thermal throttling.
@@jstar2235 Is that also a Prescott core? The earlier Northwood cores are the better options for OC but suffer from sudden death. We didn't overclock my rig, and runs on the stock cooler. It held all that hotness for over 10 years until it died on a black screen (no POST no boot)
My 2006 VAIO restoration unit with a Core Duo would throttle to 50% at 98 degrees C so that would still be over 90 degrees...
@@ruikazane5123 It was a prescott. That cursed thing made me forever wary of CPU temps, so now I just go with a big fat noctua cooler for every build.
6:01 It's these sorts of things that make you realize Linus is really savvy when it comes to tech, but he can be truly clueless when it comes to "normal" stuff lmao As if it's the first time in his life he has to deal with a tape dispenser...
to be fair, that is a crappy tapegun design to use. esspecially when typical use and abuse happens to break off the top flap and the whole gun is unusable now.
@@Dreddy72 As someone who has to send packages on an almost daily basis, this Staple's tapegun is fine and virtually impossible to break by accident, you'd need to try and break it on purpose. And I'm no gentleman when it comes to taping packages...
Alex:Your house might burn down if you have this computer.
Linus:BUT LOOK AT THESE *REFLECTIONS*
2:52 Hey Linux, CNC machinist here !
The reason why a good surface finish like this costs alot is because we need specialised carbide inserts in the facing tools and alot of machining time to get such a finish. Specialised inserts for insane finishing capabilities are more expensive, as well as the tools to wield them. So its not really a question of a better machine or not, but more like better tools and more work time, therefore the value is higher.
That intel IHS i would say is MORE likely pressed into shape in most part instead of CNC machined. As its sooo much cheaper.
More like cast "with recycled materials"
These are usually stamped, with the die-facing side machined for higher degree of flatness.
You are right, it is stamped, that is the reason why every machined IHS is better than the original one.
@@TheCampi86 Kinda. There are ways to flatten stamped metals, intel just chooses not to due to the cost of doing so. At scale it wouldn't even be particularly expensive. Honestly I don't know why AMD or Intel doesn't spend a bit more money on the highest end (5900x, 5950x, 12700k, 12900k) replacing the normal stamped IHS with proper machined (or stamped then machined is most likely)/lapped ones. Would cost them probably less than $15 per IHS, and would improve performance by a marginal, but measurable amount.
10:45 DAMN! It sounded like a dog barkimg
I used copper shims to repaste my Alienware M15 R5. Ryzen 9 5900hx/3070 130w
I used 20mm x .3mm for the GPU, and 15mm x .5mm for CPU.
I have not seen the CPU over 83 since, and GPU hasn't been over 79. Those highest temps were during a back to back Time Spy/ Afterburner overclock session.
11,091 time spy score with it. This is my first experience with copper shimming. I used to have a computer that instantly throttled during anything more than web browsing.
i wish you guys would have a solidworks single core stress test on 250meg file loads for some real world performance indications.. not render so much, but single thread pressure on the cpu, which is primarily large model loading. at work, this can typically be a 15-20 minute load...
Here's an idea for cooling experiment.
Lap that IHS and a cooler for a perfect flatness on both surfaces to see if you don't need thermal paste. You could also try to attach the cooler to the ihs in vacuum to make a cold weld contact.
^^ *THIS* ^^
_Vacuum_
i think a more 'accurate and scienfitic' method would be like der8auer did - measuring the average thickness of the thing with a micrometer and evaluate a standard deviation. Just measure like 20-50 spots on the heat spreader and compare the standard deviation to the copper one.
I did this to my 12900k a month ago. rockitcool didn't ship it with the liquid-silver to remove the old TIM so I had to use a razor blade. I ran tests before and after and got 10c drop in temperature (difference over ambient temp), also all the cores were %16 closer in temperature to each other then before. Der8auer is rumored to come out with a direct-die frame soon, so can't wait to try that out.
Thanks for watching 🤗
Send a direct message right away,I have something for you.🔝🔝🔝
The fact that even with a high-end cooling setup and exotic liquid to boot is STILL not enough to realistically keep that chip under wraps only further proves to me that Intel still has a long way to go if they want to properly catch up to AMD. Yeah, that thing is faster than my 5950X. But, I can still get reasonably close to it in performance all the while drawing significantly less power from the socket and putting out far less heat. I can run my chip on an NH-D15 comfortably all day long w/PBO enabled. I can't run that 12900KS let alone the 12900K on one out of the box without severe thermal throttling. That's just ridiculous.
NH-D15 is what I was/am planning to get for my 5950 but my temporary NH-U12 is keeping it surprisingly cool.
The fact that I can run PBO and do a full 4K cpu video exporting while not seeing 90*C made the U12 alot less temporary than I was expecting
@@ValentineC137 Same on a be quiet Dark Rock 4 with a 200w PBO 5950X, it's simply incredible how efficient modern AMD CPUs are. Draws the same as a significantly slower R7 2700X, and I've measured in some workloads 3x better performance
Windows 11 is actually the ideal version of Windows for the CPU since its kernel actually knows how to properly handle the efficiency cores.
Nice try, microsoft
@@manuelkimothy6998 I'm just saying the truth about the difference between Windows 10's and Windows 11's versions of the NT kernel.
Nothing you say will make me switch from running XP!
Windows 11 is shit and will remain shit throughout its life cycle. Windows 12 will be built on windows 11 with all bugs cleared out and new futures polished. It happens every cycle.
Maybe, but so does Linux. And honestly, I don't really care anyway. Fuck Windows 11. I'm not running it. And fuck 10 too for that matter, although that's not relevant at the moment.
The 12900ks and the 3090ti should be renamed the global warming combo
Trade a degree or so for the earth over the next hundred years for top teir gaming for life? Where do I sign up? Lol
08:43 at this point I would polish this surface to get most clean, flat and shiny surface. Just because... You could produce content and have two tests - one with delivered part and one with some tuning. 🤷♂️
@3:19, They interpret they are interpreting the spread of the thermal compound on the heat sink incorrectly. It actually was making better contact on the center leaving more thermal paste on the CPU and was less effective on the sides. JMO.
I'd like to see a mid - high performance build with a focus on low electricity usage as electricity its getting expensive in some areas of the world
AMD is your friend on that regard
First time I actually got really scared when Linus dropped something. That double-take was something else, man
Seeing Linus play with the FLIR, reminds me of our department crawfish boil, using our $20000 Search and rescue flir to check oil temps on the fish fryer.
15:58, well good thing i redid my wireing to 2.5 mm2 and an 25 amp breaker XD, 230v btw
what would be the temperature from just delidding the CPU and putting liquid metal under the stock IHS? Would be nice to know for comparison
Not much. Der Bauer (der 8auer) already tested it on his channel some time ago ^^
@@StitchExperiment626 did He? I only found the Video where He killed his 12900k
I'm also curious about how this aftermarket IHS performs compared to a lapped stock IHS
Since it's soldered I think it would be about the same, it's not like the older CPU's that used some cheap paste.
I think part of the difference might be the thickness of the copper lid. I have one of those (no name brand) for X series and the lid is about 2mm thick, vs maybe 0.5mm stock. More thickness would allow the heat to spread out more better from the CPU die to the rest of the IHS.
8:00 I want to die
The more copper it is (impurities and oxygen) and the microstructure could be ways to improve the heat spreader. Could make it just thicker and bigger too?
by the time i desolder the IHS, why not do direct die cooling?
paying 80$ more for a special waterblock sounds way easier at that point