Thanks for the English video on this. I really appreciate your work and love being able to watch without subtitles. Your vids are are worth a watch even if I have to read subs, but still appreciate it so much.
@@brandonlittle6444 yes but it doesn't mean that as an English speaker we are entitled for everything to be in our language and he is correct in that it is more work to record for 2 languages
Ugh...I’m a chef and I was just thinking the same thing. In fact, I’ve cut my fingers so many times over the course of my career I could feel the pain of the razor blade cutting my fingers after slipping off the die just from watching the video.
Oh man tell me about it. I had a very deep cut in my left finger about a year ago when trying to flushcut a cable tie... yeah i know, why not buy a knipex flushcutter... well did since then :P
That was unexpected. Indium's thermal conductivity is 81.8 W/(m·K) while Conductonaut's is 73 W/m·K. Intel used too much indium for soldering or that isn't pure indium, it's a alloy.
There is no chance that this much temperature difference can be explained by difference in thermal conductivity, there must be something else at play here.
Ohh boy, such a rarity underwent a torture. When I saw delidding process my skin crawled. But little that CPU knows, true Torture didn't even start... :) As of this writing, true torture is happening at this moment...
@@monotrope Judging by the bottle, it's Laura nail polish from Netto. Pretty much the cheapest you can get in Germany. It works well though, even for higher voltages.
Agreed, Indium looks a lot thicker than prior videos on delidding, but 12 degrees from that?!? practically unbelievable! The only other thing I can think of is that part of the indium wasn't making good contact, or the indium his cpu had had a ton of impurities.
@@Appri At this point the major hardware channels should employ a proper engineer/scientist who would be able to solve the 1d heat transfer equation. Because intel doesn't care and needs to be blamed for this.
@@pavelgorlachuk1460 I believe this is an issue of manufacturing tolerances being ignored, like they have the technology to set the die tolerances to the nanometer but ignore tolerances on the iHS.
@@Appri tbh would you be more interested in a "7nm cpu!" or a "Smaller manufacturing tolerances in the contact distance between the IHS and CPU dye" when reading marketing material? Lol
Micrometer sir. A caliper is a totally different instrument. It measures exterior, interior and depth. Where the micrometer only measures interior, but it's more precise
I wonder if he measured the inner depth of the IHS. That would tell you the solder layer thickness. But that should be the same as the Conductonaut minus the black RTV they use to stick it down
He has a lot of money and those cpus aren't even good. It's not like it's his personal gear or rare stuff. When you spend month wage doing this, that is real courage.
Just curious, why not heat the CPU first, before applying the delidding tool? It seems like the Indium releases quite easily once heated, and then it wouldn't take so much force with the tool. Is it because of the risk of the die separating from the substrate? Also same for removing the Indium from the die. Why not heat it and remove it when still hot?
You first need to loosen the silicone from the IHS/PCB, which is what he has done when he put it in his delid-die-mate. I don't think he would have much luck with a razor blade due to the transistors and the fact that soldered cpus won't flex and make way for the blade, as the solder still hold it tightly in place.
@@pavel2058 Adhesive always releases better with heat. Oven first would make it a lot easier since the glue will expand, and when it contracts again it loses some of it's integrity. However, when heated, all the other components on the cpu are in danger of falling off, which is probably why he didn't heat it up first.
Back when I had a 7700k Delidding took me from 87-89°C highs to 65-67°C after going from stock TIM to Conductonaut. I even have the results logged it's insane.
Intel should have spent all this 11th Gen porting/marketing money on cutting the prices of the 10th Gen and then bring Alder Lake as quickly as humanly possible.
Rocket lake was more an R&D expense for Intel, the knowledge from the engineering side is probably worth it and selling the product would recuperate a little.
The fact that delidding has such a huge impact tells you all you need to know about intels current state. I do hope they get their shit together next Gen.
Intel got complacent, and AMD took advantage. Intel is still being complacent, though, and it seems that they havent REALLY learned from being usurped. Not that I am complaining.
Yeah, it seems like this gen is just being sold as a "well, we have it so might as well". Probably next gen as well, but I'm hopeful it Intel will be competitive after 2022. (It takes much longer than a year or two for their investments and new hires to play out in production.)
I couldn't watch that part. I closed my eyes when you cleaned the solder with the razor, man one slip of the blade and your thumb and index finger would have battle scars.😄
from the enginiers of the intel termal paste, now on the 11th cpu generation the indium alloy. Feel the emotion of change it for liquid metal and get 12c° cooler again
I have never seen anything like this before, and didn't know that it could even be done! Really interesting, and I enjoyed the running technical comentary on the process. I wish my German was as good as this guy's English!
And this is the reason why I love to watch his videos. Even tought something looks awesome, always question your results. Even better by asking to see more results. Top one Roman!
@@mdd1963 Well, americans also pronounce meter as "three feet" and kilometer as ".7 miles". They're just always wrong with their pronunciations. The word micrometer is clearly, logically and intended to be a combination of "micro" and "meter".
I want to try this for the 11900k. This would be my first time doing something like this. If i follow directions to a tea should i run into any problems?
Sad part is as much as these flagship CPUs cost, this should already be done at the factory. You shouldn't be able to tell a bit of difference by doing this. Thank you for the video as always you're smarter than the manufacturer. 👍
for my 9900k the vccsa and vccio where 1.39v and i clocked it back to 1.17 and its stable just fine. and im running 4266 cl16-16-16-32-400. ym motherboard voltage was 1.49v and my vcore was 1.395v which i both lowerer to 1.25v I dont know whats up with boards screwing up voltages these days. temps also went down 30c... Haave you tried to see how low voltages can go? 1.38v still feels stupid high. and what did you cant up putting the vccsa and vccio on.
Board voltages are getting crazy. I had an Asrock board trying to push 1.46v through a 3600x and temps were way too high. Set my own voltage to 1.275v and it was much cooler and sustained max clocks without issue.
@@dracer35 Yeah its nuts i get they try to set a voltage that works for every chip but frankly 0.07 volts over the minimum should work for any chip or i would freaking rma that thing anyway. if my intel would need ore then 1.29 volts i would not have kept it at the time seeing most people get even lower voltages then me i think thats not to much to ask. no need to make it 0.17 higher.
Max 70°C @ CPU Package, 67°C avarage (the same test as in this video @17:17) 11900k, 5000MHz, 1.29V, 12k non AVX (10 min warmup), 212W CPU power; Asus Maximus 12 Extreme, Cooler Master ML240L RGB; CPU die is milled down to about 0.3 mm thickness.
Wouldn't it be better to pre tention the lid and put the whole delidder and cpu into the oven? It would have to be a 2. step process. First time to offsett the glue and the 2.nd time to get rid of the lid. The benefit would be a lot less pressure would be needed. Perhaps it could be tested with a dead cpu?
What load is it under? 6:38 It doesn't look like cinebench is currently running & 80-90+ degrees under watercooling doesn't seem that great, of course the only thing I have to compare to is my 49x 7700k overclock at 1.34v
Knowing Gigabyte, that's probably a chipset or VRM cooling fan, they like to put those little screamers on their high-end motherboards for that kind of purpose lol Happy midweek Roman
Roman it might be worth sending the indium to a testing lab to see its thermal conductivity the alloy might be able to handle less heat the than Liquid metal
it's kinda surprising to see just how much biggeer the 11900k die is vs the 10900k. so much more room but for 2 more cores. but that also somewhat explains the temps. bigger die surface area plus the bigger ihs surface area means better cooling. thats probably the one thing intel did rite with rocket lake.
well, taking in account the difficulty of the delidding process itself. If you have good soldering equipment, smd removal alloy and good soldering skills. It would be very easy to first remove the smd components, usually just some capacitors, one by one, recoding location of each component, then delid without a worry and place the components back into the original location afterwards.
Could you maybe unsolder those capacitors on the outside of it, remove the IHS, and then solder them back on? Also, the term for that measuring tool in English is "micrometer," pronounced "My-KROM-eter."
That's a frightening delid process, have you tried a fluxed solder ribbon on an iron to clean up the In instead of risking a chunk of your thumb to a razor? Does the LM dissolve the indium residue? I'm really curious what the story is with the larger die despite dropping two cores. Does correlate with your results though, better thermal transfer through a larger surface area and less cores adding heat to begin with. Maybe Intel just wanted to send off that legacy silicon with a robust physical design.
Using a razor across the die? Any surface scratches is bad. Couldn’t you just spot heat it up and use a solder debraid to remove excess material? Thanks for sharing you are a braver man.
my 10900k with rockitcool ihs, lapped, with LM on both sides does 5.3all core at 1.385v and doesnt exceed 90c with arctic 420aio. around 325-330w. also, instead of using a gasket maker to glue it down i used kingpin thermal grease to grab heat from the pcb as well.
just heat up the ihs with soldering bolt BEFORE putting it in to the delid tool... those pre soldered die's could easly get destroyed if you place them cold in the delid tool...
I'm half a year late but finally retiring my 5820k today for one of these with the MSI Z590-A Pro, hoping for a huge upgrade, but can't decide on doing a delid
I don't even need to read anything, and the very first words tell me, that you are a native German speaker XD And das is sehr good, weil I mag the Klang of deinem accent XD P. S. Super spannender Prozess, vielen Dank!
Probably a stupid question, but why do you remote the indium with a blade rather than heating it up and wiping it off ? Is it because of the risk of it spilling around ?
@@taiiat0 yea but he is heating it up enough to melt it anyway in the delidding process. As for working with molten metal, it should be no different than removing solder with soldering wick unless the indium compound doesn't wet as well to copper solder wick for some reason.
I just built a 10900kf a couple weeks ago, and I am fully stable for multiple runs of cinebench with 5ghz on all cores using a 240mm radiator and voltage exactly where yours is at. I was hoping for a bit faster but the stability doesn't seem to be there, maybe if I delided like you did I could actually push it faster, but im too scared to do it x.x
My Ryzen 7 5800X makes 5998 pts (multicore) without delidding ;) My temp was 88*C, and power draw was 151W during test (highest values). Anyway, very good job and improvement of temperatures by delidding an i9!
Well done. You really need to take that uhd750 iGpu for a spin now if your overclock (or somesuch mod) that 11900k cpu and its uhd750. Can you make it defeat an nvidia1030 at 720p and 1080p in Red-Dead-Redemption dx12 at 24fps or higher? Likewise Vulkan in MSWindows and Linux, especially games like Doom-Eternal. You could even try a machine learning software in a battle vs the cuda-cores of the 1030. Not buying a nvidia1030 for that Mboard saves £75 to £100 and 30watts and space and a fan cooling (e.g. redirecting airflow in a case the 1030 has obstructed)... Whereas the 11900k can be an upgrade cpu which still uses a watercooler the pc owner already had with no need for a new fan.
Thank you Roman! Always appreciate your English videos. The temperature reduction is impressive, but the chip’s performance is almost laughable. My 5900hs low voltage laptop chip is getting close to 90% of cinebench score shown here.
Could it be you kind of attached the cooler wrong the first time? I mean there was a temp difference of 10 degrees between cores, and that setup is the only thing you didn't double check, right?
Is the delid plus liquid metal stable long term to leave in a PC? Why wouldn't Intel just use this instead of indium solder? Is it less reliable to manufacture?
80W/mK at a die size of 2.76e-4 m2 would cause 12K at 300W if the indium solder was t=2.76e-4*80*12/300=0.8mm. So if the indium solder was that thick, it could explain the temperature drop. But it seems like it was originally much thinner, so this is hard to explain.
I wish Intel would release a special CPU SKU that either has liquid metal pre-applied, or leave the IHS unglued and unsoldered, so it is easy for overclockers and enthusiasts to liquid metal the CPU. You'd think it'd be a core focus for Intel, given their advantage over AMD being that monolithic die being so good for liquid metal and extreme OC...
Thanks for the English video on this. I really appreciate your work and love being able to watch without subtitles. Your vids are are worth a watch even if I have to read subs, but still appreciate it so much.
+1 on this. It must suck to have to make videos twice - but it is greatly appreciated.
@@TheStigma Smart people can accept that English is a more universal language than German.
For me it's a deal breaker due to an accessibility issue.
@@brandonlittle6444 yes but it doesn't mean that as an English speaker we are entitled for everything to be in our language and he is correct in that it is more work to record for 2 languages
@@brandonlittle6444 "Smart people can accept that English is a more universal language than German." *laughs in Chinese*
"This oven probably has seen more hardware than pizza." Lol
@sw4gr1d jokes on you, I usually eat pizza between 1 and 2 time a week, as many other friends do, here in Italy
What's for dinner today?
11900K
Is that with pineapple?
@sw4gr1d I can't speak for other people but I only get pizza once or twice a month usually
@sw4gr1d Me as an Italian that lives in Italy (so outside of the USA): "Wuot is dis man traing to sai? I cant anderstand!"
Still unconvinced to buy intel 4 - 40 yrs of lying.
My stomach turns every time I see him pushing that razor blade with so much force, right at his fingers....
Ugh...I’m a chef and I was just thinking the same thing. In fact, I’ve cut my fingers so many times over the course of my career I could feel the pain of the razor blade cutting my fingers after slipping off the die just from watching the video.
not everyone is clumsy
How not to hold tools. Jesus get a holder and push it away with your thumb.
fingers will grow up, cpu won't xD
Oh man tell me about it.
I had a very deep cut in my left finger about a year ago when trying to flushcut a cable tie... yeah i know, why not buy a knipex flushcutter... well did since then :P
That was unexpected. Indium's thermal conductivity is 81.8 W/(m·K) while Conductonaut's is 73 W/m·K. Intel used too much indium for soldering or that isn't pure indium, it's a alloy.
You can see there's a lot of indium there
Oh I highly doubt that it’s pure Indium. It’s almost certainly an alloy. it probably has some tin and other things in it.
I reckon the glue being removed and reducing the height made the difference.
@@anthonypugliese7340 That probably made half of the difference. The indium really do be thicc.
There is no chance that this much temperature difference can be explained by difference in thermal conductivity, there must be something else at play here.
This delidding process is more intense than pushing the Evergreen Ship out of the Suez Canal.
lol
+1 for current event reference
I'm confused. Is it called Evergreen or Ever given? I've read both names in news articles several times now.
@@maxrei8786 I think Evergreen is the company and all ships are named Ever+ ***.
Ohh boy, such a rarity underwent a torture. When I saw delidding process my skin crawled. But little that CPU knows, true Torture didn't even start... :) As of this writing, true torture is happening at this moment...
14:57 Roman resells his girlfriend's nail polish as *_Thermal Grizzly TG Shield Conformal Coating_*
I read roman as rusian and I think that's more fitting 😂
Lol ya and it's pretty damn pricey too. They aren't shy to advertise it as a nail polish as well XD. Any idea what nail polish it is?.
@@monotrope Judging by the bottle, it's Laura nail polish from Netto. Pretty much the cheapest you can get in Germany. It works well though, even for higher voltages.
@@heyarno Netto 😂
lol, also useful for painting usb cable ends if you have a ton and don't want to guess at a rats nest.
Looks to me that indium solder is too thick, so that it introduces additional thermal resistance. Otherwise it's a miracle.
Agreed, Indium looks a lot thicker than prior videos on delidding, but 12 degrees from that?!? practically unbelievable! The only other thing I can think of is that part of the indium wasn't making good contact, or the indium his cpu had had a ton of impurities.
@@Appri At this point the major hardware channels should employ a proper engineer/scientist who would be able to solve the 1d heat transfer equation. Because intel doesn't care and needs to be blamed for this.
@@pavelgorlachuk1460 I believe this is an issue of manufacturing tolerances being ignored, like they have the technology to set the die tolerances to the nanometer but ignore tolerances on the iHS.
@@Appri tbh would you be more interested in a "7nm cpu!" or a "Smaller manufacturing tolerances in the contact distance between the IHS and CPU dye" when reading marketing material? Lol
I'd say it's a miracle!
The English word you were unsure about for the measurement device is either caliper or micrometer 😁
Micrometer
Vernier caliper?
Micrometer sir. A caliper is a totally different instrument. It measures exterior, interior and depth. Where the micrometer only measures interior, but it's more precise
@@Matpermad micrometers primarily measures the exterior but depth micrometers are also available
@@horrormoviemaker1 you are right, but either way, they are not Calipers for sure. Those are 3 in 1, less precise and completely different LOL
That Indium solder looked as thick as a 0.020" / 0.508mm copper sheet, so 10-12°C seems right. Good work Roman !
I wonder if he measured the inner depth of the IHS. That would tell you the solder layer thickness. But that should be the same as the Conductonaut minus the black RTV they use to stick it down
I had to check the cat for breathing haha. Staying so still between shots. What a pro and a cutie
Just the courage at 8:54 deserve my thumbs lolol omg...
He has a lot of money and those cpus aren't even good. It's not like it's his personal gear or rare stuff. When you spend month wage doing this, that is real courage.
Well the tool is designed to do just that isn't it?
@@Lisa_Minci96 yeah, the tool was, but the processor...
@@Acenis Yeah it's like for a normal person risking 5 bucks CPU and getting few times more the money from content.
Just curious, why not heat the CPU first, before applying the delidding tool? It seems like the Indium releases quite easily once heated, and then it wouldn't take so much force with the tool. Is it because of the risk of the die separating from the substrate? Also same for removing the Indium from the die. Why not heat it and remove it when still hot?
You first need to loosen the silicone from the IHS/PCB, which is what he has done when he put it in his delid-die-mate. I don't think he would have much luck with a razor blade due to the transistors and the fact that soldered cpus won't flex and make way for the blade, as the solder still hold it tightly in place.
@@pavel2058 Adhesive always releases better with heat. Oven first would make it a lot easier since the glue will expand, and when it contracts again it loses some of it's integrity. However, when heated, all the other components on the cpu are in danger of falling off, which is probably why he didn't heat it up first.
Beautiful cat!
It's good to see Sheik is doing well after that fall!
Gigabyte shooting out auto voltages that can almost kill new CPUs. Awesome.
Yeah, is that deliberate?
Well those cpus are doa anyway lol. Ideal voltage for those few months stop gap until 12th gen.
@@Acenis Socket 1700 will be a thing before the end of the year. 11th gen seems like a bit of a pointless endeavour to me.
@@Outland9000 Yeah that's what I meant. Voltage gonna kill them in time for 12th gen.
The 11900k only has to last 6 month until alder lake is release, so using CPU killing voltages is just fine ;)
Back when I had a 7700k Delidding took me from 87-89°C highs to 65-67°C after going from stock TIM to Conductonaut. I even have the results logged it's insane.
As always, an interesting video not mere tests like the other channels do) thank you!
and so?
happy with the show? useless experiments.
fine and good for fun and entertainment,
Intel should have spent all this 11th Gen porting/marketing money on cutting the prices of the 10th Gen and then bring Alder Lake as quickly as humanly possible.
Rocket lake was more an R&D expense for Intel, the knowledge from the engineering side is probably worth it and selling the product would recuperate a little.
I am a simple man with simple tastes, I see a cat in the tumbnail and I click.
Very interesting, thank you for posting this. The tool you used to measure the die width is called a micrometer in English
The fact that delidding has such a huge impact tells you all you need to know about intels current state. I do hope they get their shit together next Gen.
Intel got complacent, and AMD took advantage. Intel is still being complacent, though, and it seems that they havent REALLY learned from being usurped. Not that I am complaining.
Yeah, it seems like this gen is just being sold as a "well, we have it so might as well". Probably next gen as well, but I'm hopeful it Intel will be competitive after 2022. (It takes much longer than a year or two for their investments and new hires to play out in production.)
I couldn't watch that part.
I closed my eyes when you cleaned the solder with the razor, man one slip of the blade and your thumb and index finger would have battle scars.😄
from the enginiers of the intel termal paste, now on the 11th cpu generation the indium alloy. Feel the emotion of change it for liquid metal and get 12c° cooler again
I just made a conclusion tech tuber who have a cat are the only tech tuber you can trust.
I didn't even notice the cat until the end of the intro lol that's awesome
Ha. The cat was the very first thing I saw.
I have never seen anything like this before, and didn't know that it could even be done! Really interesting, and I enjoyed the running technical comentary on the process. I wish my German was as good as this guy's English!
Really appreciate you doing these videos in English as well, danke from AU
12c is a massive improvement... I wonder is the indium thicker? Maybe lower quality??
Omg insane amount of force needed for delidding
And this is the reason why I love to watch his videos. Even tought something looks awesome, always question your results. Even better by asking to see more results. Top one Roman!
Roman, that tool is called a Micrometer. My-Crom-E-tr
Not Mi-Cro-mE-Ter? (I assume the capital letters are for emphasis)
@@GLITCH_-.- my-CROM-itr
@@mdd1963 my-chroME-iter?
@@GLITCH_-.- the mot common American pronunciation is my-KROM-itr...(emphasis on KROM)
@@mdd1963 Well, americans also pronounce meter as "three feet" and kilometer as ".7 miles". They're just always wrong with their pronunciations.
The word micrometer is clearly, logically and intended to be a combination of "micro" and "meter".
I want to try this for the 11900k. This would be my first time doing something like this. If i follow directions to a tea should i run into any problems?
Sad part is as much as these flagship CPUs cost, this should already be done at the factory. You shouldn't be able to tell a bit of difference by doing this. Thank you for the video as always you're smarter than the manufacturer. 👍
Well, what's the fun in that!?!?
for my 9900k the vccsa and vccio where 1.39v and i clocked it back to 1.17 and its stable just fine. and im running 4266 cl16-16-16-32-400. ym motherboard voltage was 1.49v and my vcore was 1.395v which i both lowerer to 1.25v
I dont know whats up with boards screwing up voltages these days. temps also went down 30c...
Haave you tried to see how low voltages can go? 1.38v still feels stupid high. and what did you cant up putting the vccsa and vccio on.
Board voltages are getting crazy. I had an Asrock board trying to push 1.46v through a 3600x and temps were way too high. Set my own voltage to 1.275v and it was much cooler and sustained max clocks without issue.
@@dracer35 Yeah its nuts i get they try to set a voltage that works for every chip but frankly 0.07 volts over the minimum should work for any chip or i would freaking rma that thing anyway. if my intel would need ore then 1.29 volts i would not have kept it at the time seeing most people get even lower voltages then me i think thats not to much to ask. no need to make it 0.17 higher.
He had to raise it up to get the stable overclock
Honey, what do we have for today's dinner.
Something special baby, 11900K!
xD
It's a Micrometer, in english.
i am here for the cat
Max 70°C @ CPU Package, 67°C avarage (the same test as in this video @17:17)
11900k, 5000MHz, 1.29V, 12k non AVX (10 min warmup), 212W CPU power;
Asus Maximus 12 Extreme, Cooler Master ML240L RGB;
CPU die is milled down to about 0.3 mm thickness.
Non AVX and CPU package temps basically mean nothing.
Wouldn't it be better to pre tention the lid and put the whole delidder and cpu into the oven? It would have to be a 2. step process. First time to offsett the glue and the 2.nd time to get rid of the lid. The benefit would be a lot less pressure would be needed. Perhaps it could be tested with a dead cpu?
Love the cat
What about long-term perspective? Is this temperatures still so low after months, years?
I scalped 4790K and use it with the Intel turbo boost only (
This oven has seen more hardware than pizza.... I died xD
Pure quality content as always. Cheers. Looking forward to the next video. 12 degrees is like the original de-lidding from 7th gen.
What load is it under? 6:38 It doesn't look like cinebench is currently running & 80-90+ degrees under watercooling doesn't seem that great, of course the only thing I have to compare to is my 49x 7700k overclock at 1.34v
Knowing Gigabyte, that's probably a chipset or VRM cooling fan, they like to put those little screamers on their high-end motherboards for that kind of purpose lol
Happy midweek Roman
awesome!....... So happy my 5600x only uses 65watt tho
Happy with my 5800x aswell, marginally better r20 scores but less than half the wattage of this 11900k... Intel needs to do better!
Thanks for taking the risk to show these results. Look forward to the direct die test.
Roman it might be worth sending the indium to a testing lab to see its thermal conductivity the alloy might be able to handle less heat the than Liquid metal
it's kinda surprising to see just how much biggeer the 11900k die is vs the 10900k. so much more room but for 2 more cores. but that also somewhat explains the temps. bigger die surface area plus the bigger ihs surface area means better cooling. thats probably the one thing intel did rite with rocket lake.
I was like man this guy is so confident in this process. Then I realized it's Der8auer himself
well, taking in account the difficulty of the delidding process itself. If you have good soldering equipment, smd removal alloy and good soldering skills. It would be very easy to first remove the smd components, usually just some capacitors, one by one, recoding location of each component, then delid without a worry and place the components back into the original location afterwards.
8:30: Me- "good God, you're going to break it."
Well, that's kinda the point
"this oven has seen more hardware than pizza" lmao
it will be more interesting to see the results on 11600k and how much advantages it can offer for an itx buld and low profile cooler.
What you used today to measure the height difference is a micrometer. Great video!
Could you maybe unsolder those capacitors on the outside of it, remove the IHS, and then solder them back on?
Also, the term for that measuring tool in English is "micrometer," pronounced "My-KROM-eter."
or just calipers would suffice even though that is the more accurate term.
@@gannonbest7251 No, that tool is a micrometer. Calipers are MUCH different and less accurate
Sleepy cat.. sleepy cat.. sleepy 😺.... 😍😍🥰
300W?!!!!! That's like Threadripper 3990X territory!!!!
Let's go for 1KW, run some AVX512 workloads on it.
Little fan is for Aquantia 10Gbit Lan, but it also cools VRMs cause it's right beside them.
That's a frightening delid process, have you tried a fluxed solder ribbon on an iron to clean up the In instead of risking a chunk of your thumb to a razor? Does the LM dissolve the indium residue? I'm really curious what the story is with the larger die despite dropping two cores. Does correlate with your results though, better thermal transfer through a larger surface area and less cores adding heat to begin with. Maybe Intel just wanted to send off that legacy silicon with a robust physical design.
Using a razor across the die? Any surface scratches is bad. Couldn’t you just spot heat it up and use a solder debraid to remove excess material? Thanks for sharing you are a braver man.
How fast, and also for the average user safely and stable can you overclock the EPYC 7763
my 10900k with rockitcool ihs, lapped, with LM on both sides does 5.3all core at 1.385v and doesnt exceed 90c with arctic 420aio. around 325-330w. also, instead of using a gasket maker to glue it down i used kingpin thermal grease to grab heat from the pcb as well.
just heat up the ihs with soldering bolt BEFORE putting it in to the delid tool... those pre soldered die's could easly get destroyed if you place them cold in the delid tool...
Is the distance between the lid and the core smaller because the black iNtel glue is thicker then your glue ?
I'm half a year late but finally retiring my 5820k today for one of these with the MSI Z590-A Pro, hoping for a huge upgrade, but can't decide on doing a delid
Perhaps since it's not using the 14nm Intel like the 10th gen one, the die is thicker? See if die lapping could have any performance impact.
Hello. When I delided my 11700k, I lost one of this elements 9:07. Someone know what is this, or when I can find? So, I will try solder a new.
Just to let you know, it should be rather easy to resolder on those surface mount components.
I don't even need to read anything, and the very first words tell me, that you are a native German speaker XD
And das is sehr good, weil I mag the Klang of deinem accent XD
P. S. Super spannender Prozess, vielen Dank!
If you're curious he also has a channel where he edits these videos but speaking German.
13:51 It's called a micrometer or fully a "micrometer screw gauge".
I love watching Deliding videos. I delid all my cpus. And love to watch Roman delid special cpus.
So, how many are died of delidding and how many survived?
@@mojeimja 4 Went perfect, 1 Died. It was 7900x, so yeah, expensive mistake.
@@Delistd ouch!
Hi Roman! Can you please tell me how you delidded 11900k?
Definitely surprising, looking forward to seeing what it can do on the EK TEC :D
Why Gigabyte motherboard? The thing doesn't seem very impressive and I've only worked with one before and didn't have a good experience.
Probably a stupid question, but why do you remote the indium with a blade rather than heating it up and wiping it off ? Is it because of the risk of it spilling around ?
That's what I was wondering too, why not just use a temperature controlled soldering iron and some copper solder wick to absorb and wipe it off?
it's hard to work with molten Metal, yes.
also less heating is always nice.
@@taiiat0 yea but he is heating it up enough to melt it anyway in the delidding process. As for working with molten metal, it should be no different than removing solder with soldering wick unless the indium compound doesn't wet as well to copper solder wick for some reason.
@@abinnixon8511
i don't think it's worth risking the molten Metal dripping onto the Chip while trying to Wick it.
Razer Blade works fine!
Why not always use the oven method more often?
I just built a 10900kf a couple weeks ago, and I am fully stable for multiple runs of cinebench with 5ghz on all cores using a 240mm radiator and voltage exactly where yours is at. I was hoping for a bit faster but the stability doesn't seem to be there, maybe if I delided like you did I could actually push it faster, but im too scared to do it x.x
So I have to ask... the TG conformal coating .... it’s looks like nail polish ...?
I'll be honest, I only clicked on the video because I saw the cat
I think reason is the die area, it is much bigger , you applied much better conductor on much larger space .
Wow very interesting, and super thanks for the video in multiple languages!
I think it was because he put quite a bit of liquid metal on the die so it made better contact and had better thermal transfer
Actual content!!!!!!
My Ryzen 7 5800X makes 5998 pts (multicore) without delidding ;)
My temp was 88*C, and power draw was 151W during test (highest values).
Anyway, very good job and improvement of temperatures by delidding an i9!
Well done. You really need to take that uhd750 iGpu for a spin now if your overclock (or somesuch mod) that 11900k cpu and its uhd750. Can you make it defeat an nvidia1030 at 720p and 1080p in Red-Dead-Redemption dx12 at 24fps or higher? Likewise Vulkan in MSWindows and Linux, especially games like Doom-Eternal. You could even try a machine learning software in a battle vs the cuda-cores of the 1030.
Not buying a nvidia1030 for that Mboard saves £75 to £100 and 30watts and space and a fan cooling (e.g. redirecting airflow in a case the 1030 has obstructed)... Whereas the 11900k can be an upgrade cpu which still uses a watercooler the pc owner already had with no need for a new fan.
Thank you Roman! Always appreciate your English videos. The temperature reduction is impressive, but the chip’s performance is almost laughable. My 5900hs low voltage laptop chip is getting close to 90% of cinebench score shown here.
Wouldn't it be easier and more effective to do away with the lid and use a copper shim of a suitable thickness in its place with thermal compound?
Could it be you kind of attached the cooler wrong the first time? I mean there was a temp difference of 10 degrees between cores, and that setup is the only thing you didn't double check, right?
@Der8auer your "measuring device" is a micrometer.
Bizarre results, now we need more delids for verification :D
Before apply protective layer please use alcohol to clean the board.
Is the delid plus liquid metal stable long term to leave in a PC? Why wouldn't Intel just use this instead of indium solder? Is it less reliable to manufacture?
80W/mK at a die size of 2.76e-4 m2 would cause 12K at 300W if the indium solder was t=2.76e-4*80*12/300=0.8mm. So if the indium solder was that thick, it could explain the temperature drop. But it seems like it was originally much thinner, so this is hard to explain.
Very nice video keep good work videos buddy
This man deserves a medal. Also, Intel, seems like your solder isn't much better than thermal paste...
The der8aulls on this guy. Hefty clackers, still very interesting content.
Damn! Was thinking you've had the two cores growing back!!
I wish Intel would release a special CPU SKU that either has liquid metal pre-applied, or leave the IHS unglued and unsoldered, so it is easy for overclockers and enthusiasts to liquid metal the CPU. You'd think it'd be a core focus for Intel, given their advantage over AMD being that monolithic die being so good for liquid metal and extreme OC...
What's the reason the manufacturer glued the lid so hard on the chip?