@@brucepreston3927 Aaaah, I wouldn't say it's anywhere near double the work. The technical testing and setup takes longer than any other part by far. Video prep isn't doubled, recording and editing is the only part doubled. He's smart though, maybe 20% more work to give or take double his views.
I did this in the past with my KS and used Liquid Metal on both sides of the ihs with an aio and contact frame. Great results, with about 18-21C better on hottest cores and I didn't need to grind anything down.
Yep! It's amazing. I've been lucky enough to get my hands on some cheap, obsolete profiling equipment off of eBay. The old stuff I have is all optical and mechanical so it requires a lot of effort and some math to come up with a measurement. The integrated software in his equipment is really where the magic happens. That magic is expensive!
-Sysinfo is the smaller of the loads. OBS is the hog, if 3 to 4 percent CPU load can be called a hog...- Edit: So I looked back on the taskmanager in the video and to my surprise HWInfo is indeed the largest loaf at about 4%. OBS isn't even visible in the list sorted after CPU load, high to low...
Even better would be tapping directly into the SMBUS with an external microcontroller the way out-of-band management controllers (like Dell DRAC or HP iLO) in servers do. A number of motherboards have smbus headers you could read from, and they all have an smbus you could solder to
Thank for the video, der8auer! Just what I needed as I have a EKWB direct die block & just bought a 14900k & I was hesitant to delid it without an expert doing it first.
As someone who has purchased the Delid Die Mate and the Thermal Grizzly Mycro Direct-Die CPU Water Block RGB I can recommend the block and the delid tool. It has made a huge difference to load and idle temps. Outstanding craftsmanship.
From what I understand from both Kingpin and others, running direct die is really not of any benefit for these 1700 socket chips. This solution with liquid metal and reinstalling the IHS provide the same results. I also believe that the 9.1GHz clock done by ASUS was done with an IHS, not direct die. No experience, just what is floating around out there.
I was just asking you about the delidding for the 14900K and found this ONLY video! Thank you for doing this and look forward to doing this with my new Hackintosh setup! I'm really glad you did this video, because I had ALREADY ordered the EK-Quantum Velocity² Direct Die D-RGB and Thermal Grizzly Intel 13th Gen Delid-Die-Mate, before watching this new video. Glad that I ordered those and look forward to seeing even better results with NO lid on that CPU.
Roman, always give your videos a watch/support just because of this sort of thing. You're not someone who will stick to a "point" stubbornly and not admit maybe a new view. It would be interesting to see if this sort of result can also be achieved with the lower end CPUs, if so, it would seem that maybe "14th Gen" is not as bad as first thought and a good value proposition, especially for those who live in hotter climates.
the difference between 13th gen and 14th gen is , 14th gen is a higher IPC and better Quality silica due too improvements in production and experience gained from Intel's new Architecture . better more refined quality Silica with lower temps , higher boost clock out of box and better overclocking potential . most reviewers didn't understand or know really what was going on .
12C drop is good result from any perspective. I was sold on Liquid metal back when I did my i9-7900X and saw a 18C drop in terms on the X299 board. Even the 12700K saw a 10C drop and I done run it as hard as i did the 7900X. I am keen to see this as a last upgrade for the LGA1700 to last me another 3 years. Thanks again Roman good work.
00:04 Testing the compatibility and performance of the 14900K delidding 01:38 Delidding results in higher overclocking potential for improved CPU performance. 02:56 Peak power draw was 293 Watt. 04:45 The video discusses the process of delidding a CPU and modifying the IHS. 06:18 The height profile of the area shows it is at a perfect 0.3mm, indicating no need for adjustment on the heat sink. 07:54 Maximum core temperature after second cinebench run: 93.1°C 09:37 Delidding with IHS and contact frame resulted in significant temperature improvements. 11:03 Delidding and modifying a CPU with liquid metal can provide a temperature improvement of 2 to 6 degrees Celsius.
The fact that you dont do Direct die means to me that the EKWB direct die product is still faulty :S Is there any possibility you revisit the matter and check the new Revised Direct die Cold plate ? Thanks Roman , Happy healthy new year !
I love your Sunday vids. Sunday is the only day off I have from work and I love waking up and seeing a new vid from you every Sunday, really makes my morning! Awesome content as usual and please keep em coming!!!
@@basshead. Nah, Zen+ was legit. 2nd gen Ryzen went from 14nm to 12nm at GF. Added more instruction sets and was way more stable than 1st gen. Granted uplift wasn't great, but it was up to 15% in some cases. It's nothing like 14th gen Intel.
@@alienrenders Agreed, Zen+ didn't offer any earth-shattering improvements but it was far more stable and easier to live with. I would hazard to say that Zen+ was a far more impressive uplift than 13th>14th gen Intel.
It is what it is. Same price, at least where I am - actually the 14900K is about 15 Euro cheaper... And when I bench the 13900K and 14900K I have right here, the 14900K is faster. 14900K scores 40400 in CB and the other 37180. Both clamped at 253 watts, -0.07 offset, LLC 5 and at 83-85 C. The difference may not be huge, but I've yet to see anyone with a 14 i9 K that ran slower than a 13 i9 K.
Here, I got a crappy chip. Slower than the 13900k at any stable settings. Thermal frame, 360aio. I even isolated the hot cores and clocked the. Individually.
My CPU would immediately Thermal when doing any benchmark - just straight to 100C on my Aorus AIO 360mm as soon as you pressed "Start" (eg: Intel Extreme Tuning, OCCT, etc.). I've just de-lidded my i9-14900KF, 7200MT/s DDR5. I went the full "Thermal Grizzly + der8auer route" (de-lid CPU, lap IHS top flat AND shave the skirt, liquid metal CPU->IHS, contact frame IHS, etc.). I had to shave the skirt of my IHS (my OCD really) because when I put the IHS back I didn't feel like the IHS was touching the CPU die, even minimally, and had no way to verify the gap. I wanted to be sure so I kept sanding until I heard a "different tick" sound when I placed the IHS onto the CPU Die board and there was a slight gap (say
For many months I was looking for the DIE height of the processor, I always looked for that name on the internet: on Google and TH-cam but I never found even a video or article about it. And then, casually watching your video without the intention of finding this information: boom. I came across the information I was looking for so long. Just for this information you already deserve my like + subscription.
For this or even for 10c on average - this is a great result. I would delidde my CPu for 10c. I got your delidder for i9 9900K but did not do it due that I hear 10c was hard to achieve. 10c temp that will be lowered from stock is a great result and I would do it. Sell your special delidder for that fancy shnancy for a good price limited edition from Germany.
@@neondemon5137 Oh trust me, if you have a small room it heats pretty fast. When I game I use ~450W and it heats my 20m2 room pretty quick. I haven't used my stove since I upgraded my PC with a 3070TI.
@@neondemon5137 Ever notice the heat setting on those? On lower heat settings they draw less power. Still enough to heat a room though. Think about it. 500 Watts is one third of the highest ratings for a resistive 120 heater you can buy. I think I measured one heating coil to consume 500w, another at 1000w and together the full rating of 1500w.
last thing to do put the his to copper watersand it to 4000 grip mirror finish for better disipation water sand also yr cooling unit to mirror same process for powerfull contact
my take on Delide vs STOCK / minor throttle vs all cores throttling, smoothness with responsiveness / i said take silicon gel make a thin coat on the chip put the heat spreader on as if you are using the silicon gen as thermal compound ... then remove the heat spreader and observe how the silicon makes little stripes and lines and bubbles to show where the high sides as well as edge of chip overall flatness i use paper nail files to level it flat until it's a suction cup to the heat spreader / around the edge and some times in the middle longest sides in the middle
I thought my 98 SP 13900K was bad but can hit 5.8GHz/4.4GHz all core with EK Velocity^2 block and custom loop, though the IMC will only hit 7000 MTs. I'd be interested in a direct die test of at least an average (near 100) 14900K. I'm waiting for the 'good' chips that will release as the KS early next year!
Love the content man! If i don't watch for knowledge or guidance, ill throw you on in the background for the ambiance. Also wanted to say i REALLY love the Kryonaut! I picked some up for my upgrades and noticed a good 6-7c decrease in thermals under full load compared to Noctua paste. Some would say that's mostly marginal, i say its more than enough for me to be completely satisfied. Thank you for all you bring to the space/community.
What about lapping ? I have the lapping tool and it did well on my 13600k but with my 14900k I’m hesitant to do it and void the warranty as I’m having some stability and seemingly a worse imc than my 13600k. Also I lapped my cold plate prior and I’m thinking I’m not getting even contact I think so could using a lapped cold plate be causing me stability issues on a non lapped cpu? 😮
How do you clean off the Indium solder? The last time I de-lided was back when intel CPU's used paste (3rd gen i7's for example) and didn't have capacitors on the PCB that could be sheered off.
I have a torquing screwdriver that ranges from 0.05 nm to 0.6 nm. By using the 13th/14th gen contact frame, what mounting torque would you recommend for a proper CPU contact with the pins in the socket? I know the 12 th gen recommends between 0.03 and 0.06 nm, but the 13th/14th gen one has no recommendation.
14th gen is meant to use less power and have lower temps over 13th gen. 14th gen has implemented something called DLVR, a method to reduce power in 14th gen. Maybe somehow your 14900K messed up with the indium and wrecked the temps. For further evidence see Hardware Canucks vid on the 14th gen release for temperatures.
A further comment: As I said, with the '14th gen' launch I'm waiting for the 14900KS after Intel kept back all the 'good' 13900K chips for the KS release. Hopefully January 2024. Your de-lid tool is obviously still compatible as the die hasn't changed dimensions! Would it be possible for you to do a direct die test of your new 14900K using LM, Kryonaut and graphene sheet? I'm sure a lot of people would be interested in the results and it would potentially remove the need to re-apply liquid metal every year or two. Also, any chance of you offering binned delidded 14900K chips with likely KS level of SP and IMC?
@@Pavasoz Yes, probably only 100MHz on paper but Intel kept the best IMC and Ring chips back for the KS release as well and will probably do so again so that's another reason I'm waiting (and hoping).
@@Pavasoz I got my 14900KS a day after release. 117 on the P Cores and 86 on the MC SP. Dropped to 106 overall by mediocre E Cores at 76 but not complaining. Runs 5.9GHz all core with 6.2GHz on 2 cores and 4.5Ghz on the E Cores out of the box on my Asus Z790 Hero on custom loop. MoRa arriving in 4 days for delid and see if I can get at least 6GHz all core in R23.
I'd like to see you take it a little further. Sand down the IHS, etc and so forth. Do 'all the things' to it, aside from direct die; then do a direct die comparison as well to top it off. Let us see just how much can be done to make the thermals better. I suspect there is another 6-9 degrees that can be dropped off.
First thing I did with my 14900k was under volt it. Lost a lot of heat and saved about 80 watts in Cinebench R23 without losing any points on the final score.
On a different subject, I purchased your new ryzen delidding tool and after removing the IHS and cleaning the silicon realized that the chiplets are not parallel. It took a significant amount of lapping to create a single plane surface for my waterblock. After some thought I decided to remachine the waterblock to alter the water flow by 90 deg trying to improve the temps across the cores. Have you tried this ? It seems like all waterblocks are primarily designed around intel chip layout. EK has offset their inlet but are still cooling the cores along the long axis.
That was surprising at first, but after thinking the difference makes sense. Probably one of the reasons why Indium is as thick as it is. Allows for manufacturing tolerances and means that they don't have to make sure that every wafer is exactly the same. It's easy to forget just how small 0.1mm is, and how expensive it can be to get everything perfect.
I used 14900kf with supercool direct die for a couple of days while testing and comparing it with 13900ks, these are two absolutely identical processors, there are no differences in appearance. But the IMC of the 14900kf turned out to be so garbage that even 8000 on the 790 apex were unstable, so I sold it immediately after the tests. This "new generation" reminds me of the 11th gen))) no improvements, just replaced one digit on the lid and pulled up the turbo boost.
I have a 14900k and just bought delidder but sketched out because the thermal grizzly delider outer packaging different material from kryonaut and contact frame.
I’ve had two CPUs on a Liquid Freezer II and can’t get remotely anything like these temps. Always have the P5 core at 100+ under load if I’m not limiting something.
I am currently working on a similar project, but with the Z690 Extreme Glacial Motherboard and a delidded i9-14900K. When comparing our systems in BIOS (7:51), I notice that we have the same frequency of 5700 MHz. While that appears fine, your Core Voltage is sitting much lower at [1.376 V], and mine is sitting at [1.447 V (Stock)]. Additionally, I am using a slim 240mm custom EKWB Loop for testing, which I realize might not be sufficient for this system. Please let me know if you have any suggestions as to how I could improve or "fix" this setup (Anyone's help would be greatly appreciated), Thank you!
Hi, I don't understand what you did. It seems to me you just replaced indium with Gallium. By cleaning the heatspreader you tightened the fit, the die ends up closer to the heatspreader ? is that it ? otherwise I don't see where the improvement came from. And you're cpu isn't safe anymore, the heatspreader is now loose.
9 หลายเดือนก่อน
I really would like to see a 14900k or 13900k with direct die and OC for daily use. What would be the clock and performance increase? There is no point in just showing the temperature. Same thing goes for Ryzen, they scale worst with more voltage.
I'm curious if Intel change something with their manufacturing process. Could you try this with a different 14th gen CPU too, to see if this result is consistent? Being able to delid, apply liquid metal, and then putting the IHS back on for a 10C reduction would be great. Also, it's been years (maybe a decade..) since I've used any liquid metal paste, what is the expected longevity in a case like this, where it's under the IHS? Is reapplication required every few years, or no?
Very nice to see your content, thank you Roman. Would your contact frame allow direct die option ? I convinced a friend to go direct die and we bought the whole set of yours but for a 13900KF and now i am stuck with my question ... Thanks again for what you provide us 🙂
It’s really surpassing Intel 14th Gen has no real performance gain over 13th gen. I am running 13900k and without OC and delidding my highest C23 score was over 40k. This was achieved by switching process to “real time” in the process manager. If I don’t tweak the processes my score drops to 39k. Hoping gen 15 is going to be better.
The fact that der8auer records in German and English for every video is some crazy dedication. Huge respect.
Yea, almost double the work! Besides prepping for the video, he does everything twice, and has to edit 2 videos everytime!
der8auer will be greatly benefit from the help of HeyGen AI, if you dont know what that is, google it. you will be impressed
Yes, it's good that he does this.
Was is das? 😳😱🤗
@@brucepreston3927 Aaaah, I wouldn't say it's anywhere near double the work. The technical testing and setup takes longer than any other part by far. Video prep isn't doubled, recording and editing is the only part doubled. He's smart though, maybe 20% more work to give or take double his views.
A leap in performance that can't be measured
We're talking upper mid single digits! I think I even saw 6% at one point in a game benchmark!
Hahaha 🤣
Literally can't measure it
Immeasurable
"You can literally see it ! " Intel Computex 2021 opening keynote meme.
Ah good to see DerCat is also present again. ps: DerCat is not sleeping - DerCat is in energy safe mode ;)
derMeower
DerKater
@@pneumaticFrequency😂
derKatze
DeFeline
What a week indeed. The gains from 13th-14th really had me going up a wall my dude.
hahaha
I did this in the past with my KS and used Liquid Metal on both sides of the ihs with an aio and contact frame. Great results, with about 18-21C better on hottest cores and I didn't need to grind anything down.
awesome! Thanks for the info
Can you explain the point of delidding CPUs that already come with liquid metal
@@MoonSon298 These don't come with liquid metal, they come with solder. While solder is better than thermal paste, liquid metal is better than solder.
@@fearthechad I think it's more about the thickness of that solder compared to a really thin layer of LM.
@@fearthechad I see thanks for the clarification thought indium was liquid metal
That height profiling equipment is pretty cool
Yeah, I wonder if it could be used in watchmaking
@@vincent67239 it is used in watchmaking, but its Keyance so this setup is most likely somewhere around $80k
Yep! It's amazing. I've been lucky enough to get my hands on some cheap, obsolete profiling equipment off of eBay. The old stuff I have is all optical and mechanical so it requires a lot of effort and some math to come up with a measurement. The integrated software in his equipment is really where the magic happens. That magic is expensive!
Wonder how far you can push it with direct die cooling.
Direct die is the way to go. I’m doing 6.1 all-core and 4.7 on the e-cores but I think I can get a better chip that can do a bit better.
@@tony_bauer @400Watts
You could disable all the unused HWiNFO sensors to drop its CPU usage, as well as greatly improve readability.
-Sysinfo is the smaller of the loads. OBS is the hog, if 3 to 4 percent CPU load can be called a hog...-
Edit: So I looked back on the taskmanager in the video and to my surprise HWInfo is indeed the largest loaf at about 4%. OBS isn't even visible in the list sorted after CPU load, high to low...
Even better would be tapping directly into the SMBUS with an external microcontroller the way out-of-band management controllers (like Dell DRAC or HP iLO) in servers do. A number of motherboards have smbus headers you could read from, and they all have an smbus you could solder to
@@mycosys SM BUS, thanks, it's on a header.
Thank for the video, der8auer! Just what I needed as I have a EKWB direct die block & just bought a 14900k & I was hesitant to delid it without an expert doing it first.
I'm not sure how seriously I can take der8auer now that I've seen he doesn't use the Dark Mode theme in Windows.
As someone who has purchased the Delid Die Mate and the Thermal Grizzly Mycro Direct-Die CPU Water Block RGB I can recommend the block and the delid tool. It has made a huge difference to load and idle temps. Outstanding craftsmanship.
Is that Mycro direct-die cpu water block for 13th/14th gen? I thought he only made it for AMD.
@@Spinelli__ Only for AM5. It works flawlessly.
@@thomaswilliams Thanks. Darn, I wish there was one for 13th / 14th gen.
@Spinelli__ EKWB Has one for i914900k.
From what I understand from both Kingpin and others, running direct die is really not of any benefit for these 1700 socket chips. This solution with liquid metal and reinstalling the IHS provide the same results. I also believe that the 9.1GHz clock done by ASUS was done with an IHS, not direct die. No experience, just what is floating around out there.
I was just asking you about the delidding for the 14900K and found this ONLY video! Thank you for doing this and look forward to doing this with my new Hackintosh setup! I'm really glad you did this video, because I had ALREADY ordered the EK-Quantum Velocity² Direct Die D-RGB and Thermal Grizzly Intel 13th Gen Delid-Die-Mate, before watching this new video. Glad that I ordered those and look forward to seeing even better results with NO lid on that CPU.
Roman, always give your videos a watch/support just because of this sort of thing. You're not someone who will stick to a "point" stubbornly and not admit maybe a new view. It would be interesting to see if this sort of result can also be achieved with the lower end CPUs, if so, it would seem that maybe "14th Gen" is not as bad as first thought and a good value proposition, especially for those who live in hotter climates.
Perhaps 14700K,,,
the difference between 13th gen and 14th gen is , 14th gen is a higher IPC and better Quality silica due too improvements in production and experience gained from Intel's new Architecture . better more refined quality Silica with lower temps , higher boost clock out of box and better overclocking potential . most reviewers didn't understand or know really what was going on .
12C drop is good result from any perspective. I was sold on Liquid metal back when I did my i9-7900X and saw a 18C drop in terms on the X299 board. Even the 12700K saw a 10C drop and I done run it as hard as i did the 7900X. I am keen to see this as a last upgrade for the LGA1700 to last me another 3 years. Thanks again Roman good work.
00:04 Testing the compatibility and performance of the 14900K delidding
01:38 Delidding results in higher overclocking potential for improved CPU performance.
02:56 Peak power draw was 293 Watt.
04:45 The video discusses the process of delidding a CPU and modifying the IHS.
06:18 The height profile of the area shows it is at a perfect 0.3mm, indicating no need for adjustment on the heat sink.
07:54 Maximum core temperature after second cinebench run: 93.1°C
09:37 Delidding with IHS and contact frame resulted in significant temperature improvements.
11:03 Delidding and modifying a CPU with liquid metal can provide a temperature improvement of 2 to 6 degrees Celsius.
Great use of the Derlidder on this -Core i9 13980K- oops, I mean 14900K
eagerly waiting for your coverage of 7000WX and pro series chips!
The fact that you dont do Direct die means to me that the EKWB direct die product is still faulty :S
Is there any possibility you revisit the matter and check the new Revised Direct die Cold plate ?
Thanks Roman , Happy healthy new year !
I noticed you used a cpu with rog bios rating of SP 82 which I guess makes it more commonplace as opposed to golden. Interesting experiment, thanks!
Just realized that Roman's a Stargate fan. I knew this guy was awesome, but good to have confirmation!
I love your Sunday vids. Sunday is the only day off I have from work and I love waking up and seeing a new vid from you every Sunday, really makes my morning! Awesome content as usual and please keep em coming!!!
13900kf, direct die block, CB23 around 44000
we can soon start measuring the power draw in horsepower of these things..
that is a fantastic result, i’m impressed!😊
I've just recovered from the mind blowing experience of 14th gen myself, what a wildly innovative and completely revolutionary series of products.
Zen+ all over again.
@@basshead. Nah, Zen+ was legit. 2nd gen Ryzen went from 14nm to 12nm at GF. Added more instruction sets and was way more stable than 1st gen. Granted uplift wasn't great, but it was up to 15% in some cases. It's nothing like 14th gen Intel.
@@alienrenders Agreed, Zen+ didn't offer any earth-shattering improvements but it was far more stable and easier to live with. I would hazard to say that Zen+ was a far more impressive uplift than 13th>14th gen Intel.
It is what it is. Same price, at least where I am - actually the 14900K is about 15 Euro cheaper... And when I bench the 13900K and 14900K I have right here, the 14900K is faster.
14900K scores 40400 in CB and the other 37180. Both clamped at 253 watts, -0.07 offset, LLC 5 and at 83-85 C.
The difference may not be huge, but I've yet to see anyone with a 14 i9 K that ran slower than a 13 i9 K.
Here, I got a crappy chip. Slower than the 13900k at any stable settings. Thermal frame, 360aio. I even isolated the hot cores and clocked the. Individually.
Can you make a video with max oc vs stock ?
13th gen didn't sell, so they put a 14 on the box.
My CPU would immediately Thermal when doing any benchmark - just straight to 100C on my Aorus AIO 360mm as soon as you pressed "Start" (eg: Intel Extreme Tuning, OCCT, etc.). I've just de-lidded my i9-14900KF, 7200MT/s DDR5. I went the full "Thermal Grizzly + der8auer route" (de-lid CPU, lap IHS top flat AND shave the skirt, liquid metal CPU->IHS, contact frame IHS, etc.). I had to shave the skirt of my IHS (my OCD really) because when I put the IHS back I didn't feel like the IHS was touching the CPU die, even minimally, and had no way to verify the gap. I wanted to be sure so I kept sanding until I heard a "different tick" sound when I placed the IHS onto the CPU Die board and there was a slight gap (say
Honestly if I were Intel, I'd have put a proper lid and thermal paste for the 14900k, so that it can at least have a better cooling than a 13900K.
Too much cost, they’ve got fanboys to profit of!
rreiter
Same as cars, standard BMW M5 Good enough, or you need better ?
Born too late to meet Nikola Tesla, Too early to see Flying cars but the right time to be born in Germany and be able to watch both of his videos
good video. thank you for the info and the work you do.
For many months I was looking for the DIE height of the processor, I always looked for that name on the internet: on Google and TH-cam but I never found even a video or article about it. And then, casually watching your video without the intention of finding this information: boom. I came across the information I was looking for so long. Just for this information you already deserve my like + subscription.
Fyi- there is one capacitor on the board that is slightly taller than the core processor. If you're planning on direct die cooling, be aware of that.
ferric chloride on a Q-tip will dissolve the solder and leave a factory smooth finish on the die.
For this or even for 10c on average - this is a great result. I would delidde my CPu for 10c. I got your delidder for i9 9900K but did not do it due that I hear 10c was hard to achieve. 10c temp that will be lowered from stock is a great result and I would do it. Sell your special delidder for that fancy shnancy for a good price limited edition from Germany.
It worked out well!
The fact that this cpu uses the same amount of energy and one of of the space heaters in my bedroom is nuts.
Its a 2-in-1, a space heater and a CPU in 1! Dont see anything wrong with that, I dont know why they dont market it like that.
That's a very weak heater, most are at least 1500 watts. Is this like a little tabletop heater from Wish.
@@neondemon5137 Oh trust me, if you have a small room it heats pretty fast. When I game I use ~450W and it heats my 20m2 room pretty quick. I haven't used my stove since I upgraded my PC with a 3070TI.
@@neondemon5137 Ever notice the heat setting on those? On lower heat settings they draw less power. Still enough to heat a room though. Think about it. 500 Watts is one third of the highest ratings for a resistive 120 heater you can buy. I think I measured one heating coil to consume 500w, another at 1000w and together the full rating of 1500w.
Damn you must live in a shoebox. I have a tiny flat and I have 2 x 2000W space heaters as radiators.
DerKat ❤
could extra extra precious bin cpus use edm machined heat spreader be any usefull
im wondering if kryosheet is safe to put on laptops due to vibrations while moving the machine
last thing to do put the his to copper watersand it to 4000 grip mirror finish for better disipation water sand also yr cooling unit to mirror same process for powerfull contact
the video i was waiting for. now i can replace my direct die cooled 12700kf, with a 14900k.
my take on Delide vs STOCK / minor throttle vs all cores throttling, smoothness with responsiveness / i said take silicon gel make a thin coat on the chip put the heat spreader on as if you are using the silicon gen as thermal compound ... then remove the heat spreader and observe how the silicon makes little stripes and lines and bubbles to show where the high sides as well as edge of chip overall flatness i use paper nail files to level it flat until it's a suction cup to the heat spreader / around the edge and some times in the middle longest sides in the middle
I got 13C drop by deliding and using direct die from EK on 13900ks. Glad it matches the pros results.
Next time, can you please review performance with an aftermarket IHS (e.g. Rockit)?
I thought my 98 SP 13900K was bad but can hit 5.8GHz/4.4GHz all core with EK Velocity^2 block and custom loop, though the IMC will only hit 7000 MTs. I'd be interested in a direct die test of at least an average (near 100) 14900K. I'm waiting for the 'good' chips that will release as the KS early next year!
My 14900k has an sp of 98 also and the imc sucks compared to my sp80 13600k lol 😂 I don’t know why an sp98 doesn’t run stable on 7200 Xmp
@@cppctek 68 SP IMC 😢 It will run 7200 but gets a single bit memory error about every 2 hours on heavy memory testing load.
Always der-bleeding edge from der8auer! Who you gonna call? (we know!) All give a big thx to Roman ...
Love the content man! If i don't watch for knowledge or guidance, ill throw you on in the background for the ambiance. Also wanted to say i REALLY love the Kryonaut! I picked some up for my upgrades and noticed a good 6-7c decrease in thermals under full load compared to Noctua paste. Some would say that's mostly marginal, i say its more than enough for me to be completely satisfied. Thank you for all you bring to the space/community.
What about lapping ? I have the lapping tool and it did well on my 13600k but with my 14900k I’m hesitant to do it and void the warranty as I’m having some stability and seemingly a worse imc than my 13600k.
Also I lapped my cold plate prior and I’m thinking I’m not getting even contact I think so could using a lapped cold plate be causing me stability issues on a non lapped cpu? 😮
Do follow-up video how much better it overclocks now...
just like the old days of delidding the 8700k
That intro sarcasm is 10/10
Nice video, well done, thanks for sharing it :)
How do you clean off the Indium solder?
The last time I de-lided was back when intel CPU's used paste (3rd gen i7's for example) and didn't have capacitors on the PCB that could be sheered off.
finally, an actually measurable improvment in production! kudos Intel! 🎉
Thanks, i wasn't sure about what i should do vs my 13900KS. I have it OC to 5.8P core and 4.4E core, no issues scores around 419k on r23.
Regardless of Intel's choices thank you Roman for the interesting amazing vids!
This is how intel is producing it's CPU's, only 0.3 % need 5.8 Ghz for gaming gear, only 0.0036 % of them do this. Not a market for them.
It’s crazy that you can optimize this better than the engineers at intel. Makes you wonder what else they’re missing over there.
Impressive
I have a torquing screwdriver that ranges from 0.05 nm to 0.6 nm. By using the 13th/14th gen contact frame, what mounting torque would you recommend for a proper CPU contact with the pins in the socket? I know the 12 th gen recommends between 0.03 and 0.06 nm, but the 13th/14th gen one has no recommendation.
TY, About what I thought it would be.
14th gen is meant to use less power and have lower temps over 13th gen. 14th gen has implemented something called DLVR, a method to reduce power in 14th gen.
Maybe somehow your 14900K messed up with the indium and wrecked the temps. For further evidence see Hardware Canucks vid on the 14th gen release for temperatures.
DLVR on 14th Gen? Are you sure? :)
That looks like our confocal 😂
I`d really love to see the EK direct die waterblock on that delidded 14900k, and with a chiller hooked up to it (-20c water).
The EK direct die block is actually really bad. Supercool or Iceman Direct Die blocks perform much better.
@@jjlw2378can you elaborate on this? I’m in the market but don’t really know where to start for reliable research
It would be interesting to see the numbers with KryoSheet under the IHS.
A further comment: As I said, with the '14th gen' launch I'm waiting for the 14900KS after Intel kept back all the 'good' 13900K chips for the KS release. Hopefully January 2024. Your de-lid tool is obviously still compatible as the die hasn't changed dimensions! Would it be possible for you to do a direct die test of your new 14900K using LM, Kryonaut and graphene sheet? I'm sure a lot of people would be interested in the results and it would potentially remove the need to re-apply liquid metal every year or two. Also, any chance of you offering binned delidded 14900K chips with likely KS level of SP and IMC?
Would the ks be worth waiting for probably 100mhz more the like its stock version from 13
@@Pavasoz Yes, probably only 100MHz on paper but Intel kept the best IMC and Ring chips back for the KS release as well and will probably do so again so that's another reason I'm waiting (and hoping).
@@Pavasoz I got my 14900KS a day after release. 117 on the P Cores and 86 on the MC SP. Dropped to 106 overall by mediocre E Cores at 76 but not complaining. Runs 5.9GHz all core with 6.2GHz on 2 cores and 4.5Ghz on the E Cores out of the box on my Asus Z790 Hero on custom loop. MoRa arriving in 4 days for delid and see if I can get at least 6GHz all core in R23.
It's the same thing and Intel is pissing on you, while saying it's raining.
So no Intel Direct Die water block? We’re your plans just scrapped? If so that really sucks I really wanted one.
we're still working on it but we had some problems in the design unfortunately
I'd like to see you take it a little further. Sand down the IHS, etc and so forth. Do 'all the things' to it, aside from direct die; then do a direct die comparison as well to top it off.
Let us see just how much can be done to make the thermals better.
I suspect there is another 6-9 degrees that can be dropped off.
Probably right but for the risk is it worth it
With these 2 steps, delid, liquid metal, and contact frame, you can get it from 93'c to 81'c, that's fucking insane
Thank you!! Doing mine today with your delidder :))
First thing I did with my 14900k was under volt it. Lost a lot of heat and saved about 80 watts in Cinebench R23 without losing any points on the final score.
Just did this to make Jay's build take an extra day.
Yea, if i delidded and needed height, i think i would get copper shims. Putting the heat spreader back on is just a waste.
On a different subject, I purchased your new ryzen delidding tool and after removing the IHS and cleaning the silicon realized that the chiplets are not parallel. It took a significant amount of lapping to create a single plane surface for my waterblock. After some thought I decided to remachine the waterblock to alter the water flow by 90 deg trying to improve the temps across the cores. Have you tried this ? It seems like all waterblocks are primarily designed around intel chip layout. EK has offset their inlet but are still cooling the cores along the long axis.
That was surprising at first, but after thinking the difference makes sense. Probably one of the reasons why Indium is as thick as it is. Allows for manufacturing tolerances and means that they don't have to make sure that every wafer is exactly the same.
It's easy to forget just how small 0.1mm is, and how expensive it can be to get everything perfect.
13900ks STOCK SCORES 42k. 14900k OC does 39k???? WTF are you doing mate?????
@5:50 What is this underfill material on the sides of the die? Is it the same silicon, or something else?
Intel clearly missed the opportunity to call this a "refresh", with CPUs being named Core i7 13770K, Core i9 13980K, etc...
Definitely in my top five list of best Germans. Keep up the great work
Who's number 1
Can I buy one of these from you ready to go?
I used 14900kf with supercool direct die for a couple of days while testing and comparing it with 13900ks, these are two absolutely identical processors, there are no differences in appearance. But the IMC of the 14900kf turned out to be so garbage that even 8000 on the 790 apex were unstable, so I sold it immediately after the tests. This "new generation" reminds me of the 11th gen))) no improvements, just replaced one digit on the lid and pulled up the turbo boost.
does that extra thermal headroom allow for a better overclock since you arent thermal throttling?
Your cat doesn't seem surprised...
I have a 14900k and just bought delidder but sketched out because the thermal grizzly delider outer packaging different material from kryonaut and contact frame.
I’ve had two CPUs on a Liquid Freezer II and can’t get remotely anything like these temps. Always have the P5 core at 100+ under load if I’m not limiting something.
This channel is now Green: DerCat is sustainable energy 🔋 😌
Wonderful video.
It is like this on 10 or 5 other similar chips? Mabye the improvements are bigger than intel can "realize" or make effective.
wow... still playing on my PC with integrated graphic, and here we are:) delidding 14th gen intel cpu like that haha
I am currently working on a similar project, but with the Z690 Extreme Glacial Motherboard and a delidded i9-14900K. When comparing our systems in BIOS (7:51), I notice that we have the same frequency of 5700 MHz. While that appears fine, your Core Voltage is sitting much lower at [1.376 V], and mine is sitting at [1.447 V (Stock)]. Additionally, I am using a slim 240mm custom EKWB Loop for testing, which I realize might not be sufficient for this system. Please let me know if you have any suggestions as to how I could improve or "fix" this setup (Anyone's help would be greatly appreciated), Thank you!
Hi, I don't understand what you did. It seems to me you just replaced indium with Gallium.
By cleaning the heatspreader you tightened the fit, the die ends up closer to the heatspreader ? is that it ? otherwise I don't see where the improvement came from. And you're cpu isn't safe anymore, the heatspreader is now loose.
I really would like to see a 14900k or 13900k with direct die and OC for daily use. What would be the clock and performance increase? There is no point in just showing the temperature. Same thing goes for Ryzen, they scale worst with more voltage.
I'm curious if Intel change something with their manufacturing process. Could you try this with a different 14th gen CPU too, to see if this result is consistent? Being able to delid, apply liquid metal, and then putting the IHS back on for a 10C reduction would be great. Also, it's been years (maybe a decade..) since I've used any liquid metal paste, what is the expected longevity in a case like this, where it's under the IHS? Is reapplication required every few years, or no?
I would like to get answers for all of those questions as well ;)
some people recommend reapplying every 6-7 months but thats just what ive heard.
I found the answer 3 comments below yours, every 1-2 years it has to be reapplied
@@InsomniaReal Thanks dude, appreciate it.
Such a nice cat..................😉
Very nice to see your content, thank you Roman. Would your contact frame allow direct die option ? I convinced a friend to go direct die and we bought the whole set of yours but for a 13900KF and now i am stuck with my question ... Thanks again for what you provide us 🙂
CPU crystal height 400 um, CPU cover height 300 um, so there is no space between CPU and cover in total ? and CPU cover hanging on the chip ?
It’s really surpassing Intel 14th Gen has no real performance gain over 13th gen. I am running 13900k and without OC and delidding my highest C23 score was over 40k. This was achieved by switching process to “real time” in the process manager. If I don’t tweak the processes my score drops to 39k. Hoping gen 15 is going to be better.
Now this is content
Finally!
Testing with Windows Defender service active doesn't seem a great idea from a method standpoint...