i did on that cpu few days ago, to play fs 2020 ahahahaha holy shit with liquid metal and rame ihs, -30c o_O 65 70c in game at 4,9ghz at 1,47! I want to see it explode muahahah
Mines still running strong kryonaut under the 4790k on top is a custom copper ihs lapped and liquid metal and a corsair hx100i never seems rk go over mid 60s over locked to 5ghz. Did similar with a 4770k would have been so much easier for this guy if he did this with it heated up or even let run in his system for a bit. Heat guns are marvellous if you don't go to close
🤣🤣🤣 Same I did my 7700k fresh out of the box, back in the day.... Shat my pants when i heard something small falling off once out of the tool but luckily it was a dollop of loose solder from the die 😰😰😰😰😰😰
So, the gain was actually 18 degrees lower on load ( prime95 max power consumption option ), but I tend to concur with you. It is a very risky procedure and I haven't had the opportunity to see if the lower 18 degrees translate to significantly better overclock. Besides, you also need to machine the water block to be able to use it with direct core on block, and that also takes some work. I will upload some videos of the water block mod in the next days. Anyway, I think that the lower temperature will allow even lower voltage in bios and that may lead to better overclock. I have to test that.
@@metalstress1736 yes in any case a nice job and especially a lot of patience. I can't do that on my R9. just sanding Die, it scares me, I did a lot of delid with intel but it was also, much easier
@@metalstress1736 dumb question, it would not have been easier to delid the cpu then use the open cpu, to heat the dough on the die then to clean simply before it cools and then scrape with a blade
@@JLCinelens So, I think the "oven" way may actually be an easy safe way of deliding. Der8auer did it with the 3960x and it seemed quite easy to do, no mechanical stress to the cpu.
@@metalstress1736 I delided old Intel CPUs that had soldered tim with lighter. First cut out all the borders and then just apply heat from the lighter. If you have even just a box cutter under the lid already, it will pop when it's melted. If you have computer running 24/7 like I have, getting CPU much cooler will save you actually pretty nice amount in couple of years. Cooler CPU takes much less electricity to run. And of course getting better performance when needed is why most people would do it.
This is incredible, congratulations for the skill! I would like to do the delid procedure on my ryzen 3900x cpu. What is this tool for removing IHS? Are you using intel IHS removal tool in ryzen?
I would suggest using the oven method instead of deliding using the der8auder tool. Der8auder posted a video about removing the IHS using heat from an oven that seems much less stressful to the CPU That this mechanical method. Regarding the tool I used, it's the der8auder deling tool with the ryzen adapter. www.overclockers.co.uk/der8auer-delid-die-mate-2-amd-kit-hs-007-dr.html
@@metalstress1736 Thanks for the answer. After the delid, did the plastic side of the am4 socket not hinder the fit of the cpu block for perfect contact with the cpu? I saw in a comment that your cpu had drops of 18c which is amazing! Did this drop in temperature allow you to better overclock?
@@evilhf915 When running cinebench or prime95 my cpu runs ( with PBO ) at about 4250mhz on all cores so the lower temperature allows it to maintain a higher performance during more time. I haven't tried to manually overclock it yet, as the asus bios is really buggy. Are you subscribing the channel? I posted a video some few days ago with the cooling setup I am using. I had to modify the copper plate to be able to clear the plastic side of the AM4 socket, as you referred the copper plate hits the socket, so it has to be machined to clear the socket. I am not sure if this happens will all the coolers. Maybe there is one that doesn't hit the socket. I pre-ordered the new ekwb AIO and when it arrives I will try it and see if it fits without modification
@@metalstress1736 Thanks again for your attention. I discovered your channel this week when I was looking for delid, and I automatically subscribed 😎 Looking forward to your tests with new cpu block. Good luck to you!
Awesome video! I'm not sure that those are capacitors at the 9:00 mark, looks like it, but if they were I doubt your CPU would function. Probably chunks of STIM, if you notice all your capacitors are covered in a compound, probably to keep any STIM leaks during manufacturing from leaking on them and killing the CPU. Question, on your dremel, what head did you use to buff the CPU cores and what was the device you kept putting your dremel head on? Wax? Been thinking about deliding my 3950x and applying Liquid Metal.
I recently delidded my 5800x3d AND sanded the silicon down. Overall got a 13 degree reduction with a low profile cooler. Worth it? Yes. Stupid? Absolutely.
@@johnfortnitetoes The nice part is the chiplet and the io dies are the same hight due to the x3d cache. So you won't have to deal with uneven mounting. I'm assuming you know the risks. Whether the risk is worth it to you or not is not something I can help you with. But all I'll say is, if you are going to do it, be very careful. $400 is $400, ya know?
What M board will you use in the mini itx build. I am planning on upgrading my external water loop fx8350 to a 3950x but I can't decide if I should go with the "asrock x570 phantom gaming itx/TB3" for the thunderbolt 3 future proofing.
I am using the Asus crosshair VIII impact. I considered the asrock that you refer, and it actually has a feature that I wanted ( 7.1 sound output on the back, because I have a creative gigaworks S750 7.1 sound system ) that the impact doesn't have ( I have to use one front plug to have 7.1 sound ). Anyway, I choose the impact for two reasons: first, it has two m.2 slots, and I am using 2x2tb nvme ssds. I wanted to eliminate as much cables as possible, this means no sata disks/ssds. Also, all my previous boards were from Asus, and I loved them all, mostly the fact that they keep releasing bios updates even for very old boards. Anyway, regarding the impact, I will also post an unboxing video shortly, but I can tell you this, the boards looks amazing and it has almost perfect specs, but the bios is very very buggy to the point of almost being useless. I am so disappointed with Asus.
I would say to get the one you said, the asrock x570 being that if you decide to upgrade then you could get a ryzen 4th gen chip and not have to buy a new motherboard once again for upgrading
Then it's not direct die. Might as well leave the heatspreader on. Edit: Unless you mean cutouts AROUND the chiplets and IO Die? would be tedious and annoying due to the SMDs. But I get ya.
The die is actually very very hard to break with pressure, but it is very easy to chip away it you hit it on the side. So as long as you place the block with care you can put an insane amount of pressure on it. I will post all this on an upcoming video.
@@jimmah4life I believe @Jackson was suggesting using something like a shrim around the cores so that the cooler block is flat over the cores and not at an angle.
This is more stressful than watching a frog being dissected. Basically a $700 frog. I don't have the balls to delid anything more recent than a 2000 series
Your CPU won't die by removing some caps. Maybe it will never even crash, ever. There is a ton of caps both on the chip and the mobo and each one just stabilizes the voltage a little, preventing interference/noise and thus improving stability to some very high standards, that need to account for less than ideal conditions. Such as a really bad or faulty mobo or other attached component. Also the caps are mounted in parallel to the circuit most likely, so it won't break any circuit either if you remove them. It's just that the closer to the chiplets, the more effective the cap will be. The biggest risk was in removing a soldered lid by force. That is really a lottery in terms of either removing it succesfully, or actually taking along the top (and part) of the chiplet(s) itself. There's quite some pictures on the net of people looking at destroyed chiplets after delidding a Ryzen. Your suggestion on using the oven sounds like a plan but it will liquify the tin on all the caps as well. So they will drift with the slightest movement, possibly shorting something out. Not to mention the risk of solderrunning into places where you can short stuff out. I'd leave the lid on, especially on the newer Ryzen 5000. Achieving an all-core boost with a meaningful result requires nothing more than a really good air cooler to run well within throttling limits, even at a hefty voltage boost. The lower temps are only gonna save a little power and prolong life a little bit but really, at the risk of instantly cutting its life short by delidding.
Are you direct-die cooling with the Enermax? Any issues, e.g. are the chiplets the same height to each other and the IO die? Also did you have to lap the Enermax or did it just for good measure? I really wouldn't mind having a delidded 3950X but I am worried about breaking mine, especially the pins. Delidding TIM-Intel was so much easier..
Not really. The caps are just there to suppress voltage irregularity ("noise"/spikes). There's tons of caps in the entire circuit, it's just that the closer to the to be protected part, the more effective. Losing a few caps won't immediately kill the CPU. In fact, it probably never will. It will just result in stability issues sooner. But, that might still never occur on a well regulated mobo.
Hi, what CPU block are you using and how did you do direct die cooling with the AM4 socket plastic in the way? Did you modify your block coldplate by machining it or something?
Always cut away when using a blade dummy the whole video shows alot of blatant stupidity isopropyl would help soften if up along with heating chip dummmmmmmy
@@metalstress1736 Vera late to the party, but.. Are you not afraid the solder on the cpu pins would melt? The threadripper has a lga, so there are no pins to fall off like on the pga. Idk if the solder on the pins has a higher melting point than indium tho.. Any way, I salute your brave and confident actions.. I would never attempt anything like that :)
@@jakobrazdevsek1595 / IHS is soldered to the die with indium, which has lower melting temperature about 170, way lower than 300+ of that of lead/leadless solder. Indium is used like that to avoid melting the componenets/die away while installing the IHS.
18 degrees sounds way too much as a difference. You prob had a bulging IHS. Did you check the IHS for flatness? Maybe should have lapped it first to see if that worked. A soldered die-IHS should not give that difference. And if it does, something else was the problem.
Just curious - I havent heard anything about your 3950x temps - are you posting any videos about temps, how long it takes to return to idle temps vs with the IHS installed, etc? I''m thinking of delidding or lapping my 3950x - I've done it before (my old 4770k was my first delid) - in fact I just replaced the metal tim on that after five years yesterday which got me thinking about this. My temps are good but not terrific with the IHS and massive water capacity loop. The tricky thing is the temps drop SLOW after load ends - not like any of my other builds. As in like 78-75-73-71-67 etc down to idle over about 15 seconds. As opposed to a 20-30C drop at end of load like all my intels have done. Could be the way this cpu unloads or could be bad contact from irregularity or bad connection between die and IHS - but that seems less likely since its soldered. Anyway - even if it's not a video - I'm curious what your temps and unloading temp decline speeds are like. Did this cpu continue to work long term after losing those two caps?
Hi there, yes, I will post some videos. I made some of the CPU as it came from the factory and others direct die. The temps are overall about 18 degrees lower. Regarding temp drop after load, I just did a quick test with prime95 small fft and it goes from idle ( 50 ) to load ( 89 ) in about one or two seconds and then back to idle ( 50 ) in 22 seconds. Regarding the caps, The CPU was used in stock status ( as it came out of the box ) for maybe two or three hours just to make a few videos and then the rest of it's life ( the past four months ) without the caps 24/7, overclocked, 100% stable. So I haven't seen any drawback which is kind of curious. Anyway, I plan on soldering the two caps again. I got hold of an old Opteron CPU that has similar caps and will use it as a donor. I need a USB microscope first to be able to solder then as they are so tiny that it would be quite a feat to do it without any kind of augmentation. Before I solder them back I will try to manually overclock the CPU to the maximum possible ( as opposed of using PBO as I am using now ) to see if adding back the caps changes the overclock capabilities of the CPU.
metal stress gotcha - so the slow cool down of the die after load appears normal and by design - it’s just totally different from what I’m used to. Your 50C idle temps seem really high - mine is in the mid to high 30’s at idle with liquid metal or kryonaut . No lapping no delidding. It drops slow from load too. It spikes under pbo+200 up to about 75 folding at home and about 80 small fft- so our peak temps aren’t that far off but I would expect your idle to be lower - mine can spike up to about 48 for a second and trickle back down. I don’t like how pbo behaves I’m thinking of delidding kine to see if I can get core temps down 10C or more under load. But I wonder if the die height is equal and if I’d have to modify my block mounts or factory mounts
@@gqneon Regarding my idle temps, the computer wasn't really idle, like I have about 100 Firefox tabs open plus a vm sitting idle on the background and some torrent downloads, but because of the 32 threads it's really doing very little. So not really idle but almost idle. I should have stated that, sorry :) Regarding the "small FFT" test with prime95 v29.8, what is happening is this: I found out that if I change some bios in settings like for instance Asus core optimization or whatever they call it, the temps are much lower but performance is quite lower too and the CPU doesn't boost to the same level as without those options. For instance with the core optimization thing on it only boosts to 3.9 on prime or something like that. with the bios fine tuned for performance ( by measuring the level of boost all the cores reach in prime ) it boosts to 4233mhz on all cores and that gives me the 89 degree temperature. With PBO I get consistent values in the 9900 range on cinebench R20. So the difference in temps may be related to your performance being lower or your CPU being better than mine. Regarding your delid questions: the three cores sit at the exact same height. No need to modify the mounts but most likely you will have to modify the CPU block. Check out a video that I put about modding my AIO. I explain the mod I had to do to make it work. One thing that needs to be done is to perfectly level the CPU block with the cores. I did this by setting an all core overclock of 2500mhz at 1v ( so that the temps sat around 60 degrees, run prime with the same 8k fft so as to make sure the heat output is constant, then using hwinfo64 to check the temps of the two core dies. with this setup running I would them proceeded to tweak the four screws of the CPU block bracket until the temps were as low as possible ( and as even as possible between the two dies ). The difference is amazing, in some cases one turn of one one screw would make the one of the dies be like 10 degrees higher. This process took me about an hour to make absolutely sure I had the best contact possible between the dies and the CPU block. I also made a video with my bios settings. check it out. Will also post some more views soon.
gqneon Yeah I mean we haven’t seen any videos of his temps, he broke off 2 transistors in this video so, I’d like to see the hw64 temps with voltages ect. Skeptical we haven’t seen any videos of temps. He said 18 degrees so I’m running 1.34@4523mhz (100.50 multiplyer) and 36c with a 360 clc with LM as the TIM. So he should be in the 20s right ?
Burandon not sure - would like to see him post something of a follow up for sure though. Are you talking about your idle temps being 36C? What are your loads temps like and under what load? I went back to PBO for the single core boost and folding - my temps are usually 62-67C depending on water temp rise as ambient air climbs in the office after days of folding. 4.5 is a lot for these chips. I’m curious about your temps with a CLC because I toyed with going to a CLC next build but wasn’t sure - I like silent cooling and I can hear my wife’s kraken 280 over mine and mine has 13 Corsair fans and two pumps in it currently.
Ok, so I just installed one of these in my new PC here ruclips.net/video/lKQj2t4Yxc4/видео.html What is the highest temperature recommended by AMD for this CPU? 85C? I let mine get up to about 64.1C before stopping it, as I've been told 60C is the Max. Idle temperature has been 100F+. The UFI Bios seems to let it run wild on default, until I set a thermal limit. My CPU Cooler doesn't seem to be getting hot. I'm using MX-4 thermal paste. So is there anything I'm doing wrong? Will a 3rd 150mm fan help? Maybe a 20" box fan in the side of my PC? Or an Air conditioner? Or a Chiller with a water block and Fan? What's the least expensive option that is the most effective?
easier method: take an iron... put some thermal paste on your cpu, put the cpu with the pasted heatspreader on the iron..... wait for about 30 secs.... grab it with a coated grill tong... done xD
@@burandon4674 I will put up a video about the mod I had to make to my Liqtech II to be able to go direct die and also video of my full build. If I was going to do it again, I would try the oven method instead of the delid process I used.
I just got this link from random link to random topic at overclock.net, what im really interested is a video detailing you hacked up custom water cooling system. It must be a sight to behold. :D
it actually looks quite good. from a distance it looks the same, just maybe with a bigger radiator, but the end result is quite clean. I will post a video about it in the next couple of days.
@@burandon4674 Will try to put together the one with the water cooler this weekend. I am sorry, I just haven't had much time. Traveling because of work have kept me away. Yep, directly on the die. No Heatspreader in the middle.
@@CSIG1001 looks like 4 resistors. I don't see the point of delidding 3000 series CPUs. They are soldered. Plus all the dies are at different heights. So this was somewhat pointless. Just lap the IHS and cooler and you are good.
@@anthonyc417 only two capacitors, not 4. The dies are actually at the exact same height. I am currently using the bare cores on the cooler block, without the IHS. Will post a video about this in the near future. 18 degrees improvement versus stock. still I don't think this is a sane thing to do to a CPU. To risky.
2 transistors break off, better is when u get u cpu hot in benchmark for 1 hrs an then u try deliding or in oven heat that shit up, but I don't think it's a good idea with oven and so high temperature
Getting it hot on a benchmark will not work because the melting point of the indium solder is 160 degrees. But an oven will work for sure. Der8auer delided a 3960x that way and it worked very well. I think it is a much better solution than the one I used.
Sure thing! AMD just loves RMA'ing CPU's that have been sliced, popped, and dremelled to hell and back. You actually get bonus points by applying a lot of pressure to the ultra sensitive chiplets by applying a direct contact cold plate. You get your name on the wall of honor in their main office when you run Prime95 on a CPU that's missing some caps too. They love it when people demonstrate just how sturdy their products are and they will gladly send you a new one if you accidentally chop up their sub-microscopic piece of electronical engineering ingenuity, using a flimsy razor knife. Good luck & have fun!
@@relin3062 i saw a few double edge blades on amazon but wasn't sure if it was the same thing. this blade looks thin and flexible. wouldn't a double edge blade mess up your hand?
@@SmoothNobody Yeah, it would, but you can see he taped over the other edge. The blade is very flexible and this video gives me anxiety, if that thing slips it goes straight to the bone 😬
@@relin3062 i saw that little protective grip. i wasn't sure if that was factory or custom. i found it difficult to believe he would use a double edge blade. was convinced this was some kind of blade i never seen before. that protective grip he made just looks like cloth. that blade will go right through that. the level of stupid in this video is alarming.
La pasta lucidante🤦 senza commenti! Usate alcool, magari isopropilico se riuscite a trovarlo. Bisogna pulire, non lucidare! Non serve uno strato di cera
You bought a Der8auer tool but don't know how to properly use it! If you've done your research...Ryzen chips are soldered. Taking a huge risk like this is pointless and shouldn't be on RUclips!
![Eminem - Fuel (feat. JID) [Official Lyric Video]](http://i.ytimg.com/vi/t5H_CewqpKA/mqdefault.jpg)
Brave man! I shat my pants on the 4790k I did back in the day!
i did on that cpu few days ago, to play fs 2020 ahahahaha
holy shit with liquid metal and rame ihs, -30c o_O 65 70c in game at 4,9ghz at 1,47! I want to see it explode muahahah
Mines still running strong kryonaut under the 4790k on top is a custom copper ihs lapped and liquid metal and a corsair hx100i never seems rk go over mid 60s over locked to 5ghz. Did similar with a 4770k would have been so much easier for this guy if he did this with it heated up or even let run in his system for a bit. Heat guns are marvellous if you don't go to close
🤣🤣🤣 Same I did my 7700k fresh out of the box, back in the day.... Shat my pants when i heard something small falling off once out of the tool but luckily it was a dollop of loose solder from the die 😰😰😰😰😰😰
All this risky work, just to gain 10c -15c, I prefer to lower the voltage in the bios.
So, the gain was actually 18 degrees lower on load ( prime95 max power consumption option ), but I tend to concur with you. It is a very risky procedure and I haven't had the opportunity to see if the lower 18 degrees translate to significantly better overclock. Besides, you also need to machine the water block to be able to use it with direct core on block, and that also takes some work. I will upload some videos of the water block mod in the next days. Anyway, I think that the lower temperature will allow even lower voltage in bios and that may lead to better overclock. I have to test that.
@@metalstress1736 yes in any case a nice job and especially a lot of patience. I can't do that on my R9. just sanding Die, it scares me, I did a lot of delid with intel but it was also, much easier
@@metalstress1736 dumb question, it would not have been easier to delid the cpu then use the open cpu, to heat the dough on the die then to clean simply before it cools and then scrape with a blade
@@JLCinelens So, I think the "oven" way may actually be an easy safe way of deliding. Der8auer did it with the 3960x and it seemed quite easy to do, no mechanical stress to the cpu.
@@metalstress1736 I delided old Intel CPUs that had soldered tim with lighter. First cut out all the borders and then just apply heat from the lighter. If you have even just a box cutter under the lid already, it will pop when it's melted.
If you have computer running 24/7 like I have, getting CPU much cooler will save you actually pretty nice amount in couple of years. Cooler CPU takes much less electricity to run. And of course getting better performance when needed is why most people would do it.
This is incredible, congratulations for the skill!
I would like to do the delid procedure on my ryzen 3900x cpu.
What is this tool for removing IHS?
Are you using intel IHS removal tool in ryzen?
I would suggest using the oven method instead of deliding using the der8auder tool. Der8auder posted a video about removing the IHS using heat from an oven that seems much less stressful to the CPU That this mechanical method. Regarding the tool I used, it's the der8auder deling tool with the ryzen adapter. www.overclockers.co.uk/der8auer-delid-die-mate-2-amd-kit-hs-007-dr.html
@@metalstress1736
Thanks for the answer.
After the delid, did the plastic side of the am4 socket not hinder the fit of the cpu block for perfect contact with the cpu?
I saw in a comment that your cpu had drops of 18c which is amazing!
Did this drop in temperature allow you to better overclock?
@@evilhf915 When running cinebench or prime95 my cpu runs ( with PBO ) at about 4250mhz on all cores so the lower temperature allows it to maintain a higher performance during more time. I haven't tried to manually overclock it yet, as the asus bios is really buggy. Are you subscribing the channel? I posted a video some few days ago with the cooling setup I am using. I had to modify the copper plate to be able to clear the plastic side of the AM4 socket, as you referred the copper plate hits the socket, so it has to be machined to clear the socket. I am not sure if this happens will all the coolers. Maybe there is one that doesn't hit the socket. I pre-ordered the new ekwb AIO and when it arrives I will try it and see if it fits without modification
@@metalstress1736 Thanks again for your attention.
I discovered your channel this week when I was looking for delid, and I automatically subscribed 😎
Looking forward to your tests with new cpu block.
Good luck to you!
Che nessuno lo faccia così, come lo sta facendo lui. Sta rischiando di tagliarsi il pollice ad ogni passaggio ! ! !
11:50 - what's the name of the power tool you are using and the material on the spinning disc
@Pew Pow really? So its not like some sort of dremel?
Awesome video! I'm not sure that those are capacitors at the 9:00 mark, looks like it, but if they were I doubt your CPU would function. Probably chunks of STIM, if you notice all your capacitors are covered in a compound, probably to keep any STIM leaks during manufacturing from leaking on them and killing the CPU.
Question, on your dremel, what head did you use to buff the CPU cores and what was the device you kept putting your dremel head on? Wax? Been thinking about deliding my 3950x and applying Liquid Metal.
He was applying a fine grit paste onto the head to “sand down” the surface of the silicon to make it super polished.
I recently delidded my 5800x3d AND sanded the silicon down. Overall got a 13 degree reduction with a low profile cooler. Worth it? Yes. Stupid? Absolutely.
My 5800x3d runs hot when under load lmfao. been thinking about delidding
@@johnfortnitetoes The nice part is the chiplet and the io dies are the same hight due to the x3d cache. So you won't have to deal with uneven mounting. I'm assuming you know the risks. Whether the risk is worth it to you or not is not something I can help you with. But all I'll say is, if you are going to do it, be very careful. $400 is $400, ya know?
What M board will you use in the mini itx build. I am planning on upgrading my external water loop fx8350 to a 3950x but I can't decide if I should go with the "asrock x570 phantom gaming itx/TB3" for the thunderbolt 3 future proofing.
I am using the Asus crosshair VIII impact. I considered the asrock that you refer, and it actually has a feature that I wanted ( 7.1 sound output on the back, because I have a creative gigaworks S750 7.1 sound system ) that the impact doesn't have ( I have to use one front plug to have 7.1 sound ). Anyway, I choose the impact for two reasons: first, it has two m.2 slots, and I am using 2x2tb nvme ssds. I wanted to eliminate as much cables as possible, this means no sata disks/ssds. Also, all my previous boards were from Asus, and I loved them all, mostly the fact that they keep releasing bios updates even for very old boards. Anyway, regarding the impact, I will also post an unboxing video shortly, but I can tell you this, the boards looks amazing and it has almost perfect specs, but the bios is very very buggy to the point of almost being useless. I am so disappointed with Asus.
I would say to get the one you said, the asrock x570 being that if you decide to upgrade then you could get a ryzen 4th gen chip and not have to buy a new motherboard once again for upgrading
Use heat gun to make solder softer, then delid. Easy 3 minutes job.
I think you should make a protector for the die when you direct die it,in case of the mounting presure break the die.
Then it's not direct die. Might as well leave the heatspreader on.
Edit: Unless you mean cutouts AROUND the chiplets and IO Die? would be tedious and annoying due to the SMDs. But I get ya.
The die is actually very very hard to break with pressure, but it is very easy to chip away it you hit it on the side. So as long as you place the block with care you can put an insane amount of pressure on it. I will post all this on an upcoming video.
@@jimmah4life I believe @Jackson was suggesting using something like a shrim around the cores so that the cooler block is flat over the cores and not at an angle.
11:54 what he use on spinning disc to clean cpu?
The best part about this video is all the people who have/are going to see it and brick their cpu trying it.
This is more stressful than watching a frog being dissected. Basically a $700 frog.
I don't have the balls to delid anything more recent than a 2000 series
Very nice work. Would never do that to my 3700X but it's just nice to see how it's possible.
Polishing and polishing paste type designation please 11:49
Your CPU won't die by removing some caps. Maybe it will never even crash, ever. There is a ton of caps both on the chip and the mobo and each one just stabilizes the voltage a little, preventing interference/noise and thus improving stability to some very high standards, that need to account for less than ideal conditions. Such as a really bad or faulty mobo or other attached component. Also the caps are mounted in parallel to the circuit most likely, so it won't break any circuit either if you remove them. It's just that the closer to the chiplets, the more effective the cap will be.
The biggest risk was in removing a soldered lid by force. That is really a lottery in terms of either removing it succesfully, or actually taking along the top (and part) of the chiplet(s) itself. There's quite some pictures on the net of people looking at destroyed chiplets after delidding a Ryzen. Your suggestion on using the oven sounds like a plan but it will liquify the tin on all the caps as well. So they will drift with the slightest movement, possibly shorting something out. Not to mention the risk of solderrunning into places where you can short stuff out.
I'd leave the lid on, especially on the newer Ryzen 5000. Achieving an all-core boost with a meaningful result requires nothing more than a really good air cooler to run well within throttling limits, even at a hefty voltage boost. The lower temps are only gonna save a little power and prolong life a little bit but really, at the risk of instantly cutting its life short by delidding.
What kind of nifty delid tool is that?
I thought there were none for Ryzen 3000/AM4.
Can you do a video on putting the cooler next ? where do you put the thermal paste ? and wich paste ?
I have the same tool at home. One thing I don't understand is the plastic piece that's made for AMD CPUs. What purpose does it serve?
os ryzen ja nao vem com pasta termica de metal dentro de fabrica?
Are you direct-die cooling with the Enermax? Any issues, e.g. are the chiplets the same height to each other and the IO die? Also did you have to lap the Enermax or did it just for good measure?
I really wouldn't mind having a delidded 3950X but I am worried about breaking mine, especially the pins. Delidding TIM-Intel was so much easier..
You can't desoldering them? Without using the blade? Sorry dumb question
3 capacitors ripped off board.. 3950 now is dead
Not really. The caps are just there to suppress voltage irregularity ("noise"/spikes). There's tons of caps in the entire circuit, it's just that the closer to the to be protected part, the more effective. Losing a few caps won't immediately kill the CPU. In fact, it probably never will. It will just result in stability issues sooner. But, that might still never occur on a well regulated mobo.
did you drop 18 degrees with direct cooling? without the IHS attached?
Hi, what CPU block are you using and how did you do direct die cooling with the AM4 socket plastic in the way? Did you modify your block coldplate by machining it or something?
Lol "put it in an oven" :)
I really mean it! @der8auer did it with a 3960X ruclips.net/video/c1wyUMZVv4k/видео.html :)
Always cut away when using a blade dummy the whole video shows alot of blatant stupidity isopropyl would help soften if up along with heating chip dummmmmmmy
@@metalstress1736 Vera late to the party, but.. Are you not afraid the solder on the cpu pins would melt? The threadripper has a lga, so there are no pins to fall off like on the pga. Idk if the solder on the pins has a higher melting point than indium tho.. Any way, I salute your brave and confident actions.. I would never attempt anything like that :)
@@jakobrazdevsek1595 different solder, the solder used on the pins are what we would use in everyday soldering, the IHS is soldered on with indium
@@jakobrazdevsek1595 / IHS is soldered to the die with indium, which has lower melting temperature about 170, way lower than 300+ of that of lead/leadless solder.
Indium is used like that to avoid melting the componenets/die away while installing the IHS.
What's the point of delidding that chip? Also I swear he was going to slice his thumb open.. I was like having an anxiety attack watching this..
Hey whats the tool exactly? Will this work with 5000 series? Thanks
what cpu block you used and did you modify your board for it to lay flat on the core and block
18 degrees sounds way too much as a difference. You prob had a bulging IHS. Did you check the IHS for flatness? Maybe should have lapped it first to see if that worked. A soldered die-IHS should not give that difference. And if it does, something else was the problem.
Just curious - I havent heard anything about your 3950x temps - are you posting any videos about temps, how long it takes to return to idle temps vs with the IHS installed, etc? I''m thinking of delidding or lapping my 3950x - I've done it before (my old 4770k was my first delid) - in fact I just replaced the metal tim on that after five years yesterday which got me thinking about this.
My temps are good but not terrific with the IHS and massive water capacity loop. The tricky thing is the temps drop SLOW after load ends - not like any of my other builds. As in like 78-75-73-71-67 etc down to idle over about 15 seconds. As opposed to a 20-30C drop at end of load like all my intels have done. Could be the way this cpu unloads or could be bad contact from irregularity or bad connection between die and IHS - but that seems less likely since its soldered.
Anyway - even if it's not a video - I'm curious what your temps and unloading temp decline speeds are like. Did this cpu continue to work long term after losing those two caps?
Hi there, yes, I will post some videos. I made some of the CPU as it came from the factory and others direct die. The temps are overall about 18 degrees lower. Regarding temp drop after load, I just did a quick test with prime95 small fft and it goes from idle ( 50 ) to load ( 89 ) in about one or two seconds and then back to idle ( 50 ) in 22 seconds. Regarding the caps, The CPU was used in stock status ( as it came out of the box ) for maybe two or three hours just to make a few videos and then the rest of it's life ( the past four months ) without the caps 24/7, overclocked, 100% stable. So I haven't seen any drawback which is kind of curious. Anyway, I plan on soldering the two caps again. I got hold of an old Opteron CPU that has similar caps and will use it as a donor. I need a USB microscope first to be able to solder then as they are so tiny that it would be quite a feat to do it without any kind of augmentation. Before I solder them back I will try to manually overclock the CPU to the maximum possible ( as opposed of using PBO as I am using now ) to see if adding back the caps changes the overclock capabilities of the CPU.
metal stress gotcha - so the slow cool down of the die after load appears normal and by design - it’s just totally different from what I’m used to. Your 50C idle temps seem really high - mine is in the mid to high 30’s at idle with liquid metal or kryonaut . No lapping no delidding. It drops slow from load too. It spikes under pbo+200 up to about 75 folding at home and about 80 small fft- so our peak temps aren’t that far off but I would expect your idle to be lower - mine can spike up to about 48 for a second and trickle back down. I don’t like how pbo behaves
I’m thinking of delidding kine to see if I can get core temps down 10C or more under load. But I wonder if the die height is equal and if I’d have to modify my block mounts or factory mounts
@@gqneon Regarding my idle temps, the computer wasn't really idle, like I have about 100 Firefox tabs open plus a vm sitting idle on the background and some torrent downloads, but because of the 32 threads it's really doing very little. So not really idle but almost idle. I should have stated that, sorry :) Regarding the "small FFT" test with prime95 v29.8, what is happening is this: I found out that if I change some bios in settings like for instance Asus core optimization or whatever they call it, the temps are much lower but performance is quite lower too and the CPU doesn't boost to the same level as without those options. For instance with the core optimization thing on it only boosts to 3.9 on prime or something like that. with the bios fine tuned for performance ( by measuring the level of boost all the cores reach in prime ) it boosts to 4233mhz on all cores and that gives me the 89 degree temperature. With PBO I get consistent values in the 9900 range on cinebench R20. So the difference in temps may be related to your performance being lower or your CPU being better than mine. Regarding your delid questions: the three cores sit at the exact same height. No need to modify the mounts but most likely you will have to modify the CPU block. Check out a video that I put about modding my AIO. I explain the mod I had to do to make it work. One thing that needs to be done is to perfectly level the CPU block with the cores. I did this by setting an all core overclock of 2500mhz at 1v ( so that the temps sat around 60 degrees, run prime with the same 8k fft so as to make sure the heat output is constant, then using hwinfo64 to check the temps of the two core dies. with this setup running I would them proceeded to tweak the four screws of the CPU block bracket until the temps were as low as possible ( and as even as possible between the two dies ). The difference is amazing, in some cases one turn of one one screw would make the one of the dies be like 10 degrees higher. This process took me about an hour to make absolutely sure I had the best contact possible between the dies and the CPU block. I also made a video with my bios settings. check it out. Will also post some more views soon.
gqneon Yeah I mean we haven’t seen any videos of his temps, he broke off 2 transistors in this video so, I’d like to see the hw64 temps with voltages ect. Skeptical we haven’t seen any videos of temps. He said 18 degrees so I’m running 1.34@4523mhz (100.50 multiplyer) and 36c with a 360 clc with LM as the TIM. So he should be in the 20s right ?
Burandon not sure - would like to see him post something of a follow up for sure though.
Are you talking about your idle temps being 36C? What are your loads temps like and under what load? I went back to PBO for the single core boost and folding - my temps are usually 62-67C depending on water temp rise as ambient air climbs in the office after days of folding.
4.5 is a lot for these chips. I’m curious about your temps with a CLC because I toyed with going to a CLC next build but wasn’t sure - I like silent cooling and I can hear my wife’s kraken 280 over mine and mine has 13 Corsair fans and two pumps in it currently.
Ok, so I just installed one of these in my new PC here ruclips.net/video/lKQj2t4Yxc4/видео.html What is the highest temperature recommended by AMD for this CPU? 85C? I let mine get up to about 64.1C before stopping it, as I've been told 60C is the Max. Idle temperature has been 100F+. The UFI Bios seems to let it run wild on default, until I set a thermal limit. My CPU Cooler doesn't seem to be getting hot. I'm using MX-4 thermal paste. So is there anything I'm doing wrong? Will a 3rd 150mm fan help? Maybe a 20" box fan in the side of my PC? Or an Air conditioner? Or a Chiller with a water block and Fan? What's the least expensive option that is the most effective?
Was it worth delideing it did u notice any difference in temps
Amazing but which aio you use?
You lost a total of three caps, almost certainly due to how deep you drove your razor blade.
easier method:
take an iron... put some thermal paste on your cpu, put the cpu with the pasted heatspreader on the iron..... wait for about 30 secs....
grab it with a coated grill tong... done xD
why there are a few haters commenting this video? If you hate it cuz u can't do it, just switch to another video.
Awsome video,I would so bend a bunch of pins doing this I may try it out.
If you are going to try I would suggest doing the oven method, instead of this one. Der8auer has done it with a 396xx and it seems easier and safer.
@@metalstress1736 Great thanks for the tip.
I am surprised that no pins were bent. I always feel they bend already when you look too long at them.
metal stress You have any more videos or tips with the 3950x? Wanting to go direct die.
@@burandon4674 I will put up a video about the mod I had to make to my Liqtech II to be able to go direct die and also video of my full build. If I was going to do it again, I would try the oven method instead of the delid process I used.
does it still work? (your cpu)
A fantastic result drop temps!
I just got this link from random link to random topic at overclock.net, what im really interested is a video detailing you hacked up custom water cooling system. It must be a sight to behold. :D
it actually looks quite good. from a distance it looks the same, just maybe with a bigger radiator, but the end result is quite clean. I will post a video about it in the next couple of days.
@@metalstress1736 Where are these videos? I want to delid my 3950x. Did you just place the cooler directly on the die? I need to see more vidoes.
@@burandon4674 Will try to put together the one with the water cooler this weekend. I am sorry, I just haven't had much time. Traveling because of work have kept me away. Yep, directly on the die. No Heatspreader in the middle.
@@metalstress1736 I am interested in that video too!
I'll be delidding my 5950x when it arrives than run direct die with custom loop
Stressful video. 1) Man, wear some gloves when using as razor blade. 2) Watching the screw being so close to the pins gave me the chills.
And that's how he killed his 3950x.
better your 3950x than mine.
Per togliere la posta termica dal die usate una spatolina da pittura!
ASSOLUTAMENTE NON FATELO CON LA LAMETTA DA BARBA
God, that was intense...
Imagine if you didnt speed up the video
Great video😄
Thanks!
Un solvente idoneo sarebbe l'ideale
which thermal paste do you use for your 3950x after delid?
Lol, the cpu died. He damaged and throw away two capacitors. 9:03
Is this cpu even boot up?
Golden hands.Well done/
is your camera hanging in a cable or smth
quick tip: don't use youtube stabilization and DONT use youtube stabilization if your camera is still
The first part of your quick tip makes the second part redundant. What a waste of server storage somewhere in the world.
Does it still even function after?
Yes, it works perfectly!
Yeah but you lost a resistor.. look at your video you blew it off the chip
@@CSIG1001 looks like 4 resistors. I don't see the point of delidding 3000 series CPUs. They are soldered. Plus all the dies are at different heights. So this was somewhat pointless. Just lap the IHS and cooler and you are good.
@@CSIG1001 They are not resistors, they are capacitors and I actually lost two. Still the CPU is working perfectly, and that is quite odd!
@@anthonyc417 only two capacitors, not 4. The dies are actually at the exact same height. I am currently using the bare cores on the cooler block, without the IHS. Will post a video about this in the near future. 18 degrees improvement versus stock. still I don't think this is a sane thing to do to a CPU. To risky.
how many tool did you use to delid this cpu????
In the behind the scenes video you can see he had a massive ball peen hammer ready too, in case the delidding tool didn't suffice.
Still alaive?
Temp difference ?
want to de-lid my brand new 3900x? lolz
haha, I wouldn't take the risk with someone's else CPU.
2 transistors break off, better is when u get u cpu hot in benchmark for 1 hrs an then u try deliding or in oven heat that shit up, but I don't think it's a good idea with oven and so high temperature
Getting it hot on a benchmark will not work because the melting point of the indium solder is 160 degrees. But an oven will work for sure. Der8auer delided a 3960x that way and it worked very well. I think it is a much better solution than the one I used.
He cut too deep into the glue, der8auer warned about that in his delid video as well.
so was it worth it ?
Is the warranty still intact?
Sure thing! AMD just loves RMA'ing CPU's that have been sliced, popped, and dremelled to hell and back. You actually get bonus points by applying a lot of pressure to the ultra sensitive chiplets by applying a direct contact cold plate. You get your name on the wall of honor in their main office when you run Prime95 on a CPU that's missing some caps too. They love it when people demonstrate just how sturdy their products are and they will gladly send you a new one if you accidentally chop up their sub-microscopic piece of electronical engineering ingenuity, using a flimsy razor knife. Good luck & have fun!
what kind of razor blade is that?
it is regular double edge safety razor blade.
@@relin3062 i saw a few double edge blades on amazon but wasn't sure if it was the same thing. this blade looks thin and flexible. wouldn't a double edge blade mess up your hand?
@@SmoothNobody Yeah, it would, but you can see he taped over the other edge.
The blade is very flexible and this video gives me anxiety, if that thing slips it goes straight to the bone 😬
@@relin3062 i saw that little protective grip. i wasn't sure if that was factory or custom. i found it difficult to believe he would use a double edge blade. was convinced this was some kind of blade i never seen before. that protective grip he made just looks like cloth. that blade will go right through that. the level of stupid in this video is alarming.
@@SmoothNobody There is certainly a level of enthusiasm in this video that one would call stupid, but I appreciate people who do this
look how shiny she is
Im scared to destroy a 300€ cpu ill let it be like it is
Great video!
Thanks!
Now why would you delid a Ryzen? The IHS is soldered to it. You ain't gonna get better heat transfer than that.
he will use it without the IHS
Oh my god two of the smc choped off those two on the rigth glue at the center u are lucky if that think still works
Não sei como não cortou a mão fora com essa gilete. Coragem demais ! Rsrs
Wauw dude scary
The way you're cutting the solder is super cringy. If it slips it's going into you fingers...
Non eseguitelo in quel modo
Andate semmai più in profondità dall'angolo ma salvatevi le dita
La pasta lucidante🤦 senza commenti! Usate alcool, magari isopropilico se riuscite a trovarlo.
Bisogna pulire, non lucidare! Non serve uno strato di cera
Wo~~ the die is shiny
Waw !
This lad is mad lmao
You bought a Der8auer tool but don't know how to properly use it!
If you've done your research...Ryzen chips are soldered. Taking a huge risk like this is pointless and shouldn't be on RUclips!
careless kkkkkkkkkkkkkkkk
Nope. Fuck that.
Please mute the sound!!!! Or put music over it.
lol i read this as 3950x died XD
Is this even necessary to do?? It looks stupid and dumb
rofl haters gonna hate