The Art of CPU Lapping: A Complicated Story to Tell

When it comes to CPU lapping, the process can be as tricky as it is fascinating. As the speaker explained, "lapping" refers to the practice of adjusting the CPU's position within its socket to achieve optimal heat transfer and performance. This technique requires a deep understanding of the intricacies involved in heat dissipation and the specific nuances of each CPU cooler design.

To illustrate this concept, let's consider an analogy: lapping one's CPU is similar to lapping oneself around a track or circle. Just as a driver must navigate through a complex course with precision and control, a CPU enthusiast must carefully calibrate their CPU's position within its socket to achieve the optimal temperature distribution.

Assuming a spherical cow (a hypothetical scenario where all variables are accounted for), we can begin to explore the various factors that influence CPU lapping. For instance, if one were to lap their CPU in socket, the goal would be to flatten out any convexities or irregularities on the surface of the die, thereby creating a more uniform heat transfer surface.

However, as the speaker pointed out, this is not always a straightforward process. Custom heat spreaders, for example, can provide a flat surface for cooling, but only if properly manufactured by a CNC machine. Similarly, using offset mounting or custom washers can also facilitate optimal lapping results.

On the Intel side of things, we find a similarly complex story to tell. With full ILM pressure, there is hardly any contact between the CPU and its socket in the critical die area. However, when using 1mm washers, this issue begins to dissipate. The standard middle-based convexity version provides decent contact, but it still falls short of perfection.

In contrast, the HP version offers excellent contact, even with or without washers. However, combining HBC (high-performance base) with washers can result in slightly too much pressure, while using HPC (high-performance contact) with washers may not provide enough pressure for heavily deformed CPUs.

Ultimately, the speaker concluded that the situation is complicated and does require a deep understanding of CPU cooling dynamics. The development of custom coolers and innovative mounting solutions has helped to mitigate some of these challenges, but it remains essential to approach lapping with caution and precision.

As reviewers, our goal is to simplify this complex topic for our readers. While we acknowledge that comprehending the intricacies involved in CPU lapping can be overwhelming, we aim to provide clear guidance and recommendations through our review and benchmarking process.

In fact, we were debating whether or not to offer multiple skus (different versions) of a particular cooler. While this option provides more choices for customers, it also requires tighter tolerance control in manufacturing and can be less economically viable. Despite these challenges, we believe that offering the best possible solution is essential, especially when our target audience prioritizes performance above all else.

In conclusion, CPU lapping is an intricate process that demands attention to detail and a deep understanding of heat dissipation principles. While this article has provided a glimpse into the complexities involved, there is still much to be explored in the world of CPU cooling. Stay tuned for our upcoming engineering videos and reviews, where we will delve deeper into these topics and provide expert guidance on all things CPU-related.

The Importance of Custom Coolers

One potential solution to the challenges posed by standard coolers is the development of custom heat spreaders. These bespoke solutions can provide a flat surface for cooling, which can be especially beneficial in situations where optimal contact between the CPU and its socket is crucial.

In theory, using a custom cooler could simplify the lapping process, as it would eliminate any irregularities or convexities on the die's surface. However, this approach also requires careful consideration of the cooler's design and manufacturing processes.

For instance, a well-designed custom cooler should be fabricated using advanced CNC machines to ensure precise tolerances and optimal heat transfer. In practice, however, this level of precision can be difficult to achieve, especially when working with complex geometries or unconventional materials.

Another approach is to utilize existing coolers and modify them for improved performance. This might involve adding custom washers or contact frames to enhance the cooling experience.

Legacy Sockets: A New Perspective

For users who still rely on legacy sockets like LGA 2066, Am4, or 2011, there are alternative solutions available that can simplify the lapping process.

In this case, using an LBC (low-profile) cooler can be a more straightforward option. By designing coolers with specific geometries and materials in mind, manufacturers can create products that offer improved performance without requiring extensive tweaking or modification.

Of course, even with these alternatives, there is still some degree of complexity involved in achieving optimal lapping results. However, by offering solutions tailored to specific needs and preferences, we aim to make the process more accessible and user-friendly for our readers.

The Future of CPU Cooling

As we continue to explore the world of CPU cooling, it's clear that innovation will play a critical role in driving progress forward. By developing custom coolers, optimizing existing designs, and embracing new technologies like advanced materials and 3D printing, manufacturers can create products that truly push the boundaries of what is possible.

In our upcoming engineering videos and reviews, we'll be examining some of these cutting-edge solutions up close. From novel cooling technologies to expert advice on lapping techniques, we're committed to providing in-depth guidance and insights that will help you navigate the complex world of CPU cooling with confidence.

"WEBVTTKind: captionsLanguage: enI promise it's actually real this time so the nocta nhd15 G2 just to clarify a few things uh this is not the video from 2023 you're in the right video please don't click away this is the new one all right so now we've cleared that up so the Gen 2 D15 is actually happening I have even requested a proof of life photo uh and what I've been told is that such photo does exist and it will be dispatched to me post haste and the cooler is actually in the process of shipping so it sounds like June is I'm getting the nod so it sounds like June is when the D15 G2 is happening uh there is a lot to talk about today so we're just split this video uh into kind of two halves of the same video I'm going to focus on product stuff I will be joined by Jacob off camera here uh in a little bit to talk about some of the specifics but to get into the details straight away on the G2 there are a couple things you need to know the first one is that this is the cooler uh but we're going to be covering some familiar details to people who saw our cold plate and IHS flatness discussion there's some really specific detail on fan uh mounting and structure as well and then uh they also have a a setup here just showing the original D15 that um later might be running some heat loads on it all right the details that matter the most right away uh $150 and is going to be an expensive air cooler a lot of that comes in from the fan before that this video is brought to you by thermal Grizzlies aeronaut and hydronaut thermal pastes aeronaut is thermal Grizzly's entry-level thermal solution marketed as resistant to curing and for long-term endurance hydronaut is thermal Grizzly's Next Step Up targeted for overclocking and higher performance applications we've used hydronaut on a lot of our systems internally over the years you can learn more at the link in the description below so for the $150 kit what they are shipping is the cooler itself the two fans and then this small piece of paper down here has four washers on it so these are 1 mm washers it's for an ilm washer mod we're going to get into detail on that uh with some of these pressure Maps too so noct is at a point now where as we've seen with air coolers in general to further tune to get any additional performance out of it you have to start doing tricks uh that may be outside of the cooler itself so as an example where we see the best coolers in our testing right now there's kind of like a 3° cluster of plus or minus and they're all basically in that range if they're the best coolers fighting for an extra 1 to 2° is very difficult at this point and where KNX was trying to do that with the g2 is going to be in the three variations so they have uh an LBC or low based convexity HPC or high based convexity and then standard standard is represented by the uh green line over here and one of the things we'll be going over is some of the temperature differences that nocta is seen based on uh what sort of CPU you're pairing with the cold plate and so the problem comes in with actually looking at flatness again uh this is an am5 solution this CPU only this CPU in socket and then we've got the basically the deviation and flatness so the problem we saw with intel was that when you clamp the ilm down it has that really deep concavity uh and that gets worse noct was saying over time which is not an aspect that we explored in our testing so that's incredibly interesting because it causes all kinds of new and wonderful problems for our test benches that uh we just deployed so I'll get to talk about that more soon back to the G2 though so other differences for the cooler include the heat pipes they're going up to eight heat pipes from six these are centered heat pipes they uh say they've done some heat pipe tuning and in addition to that uh the fin pitch and density has changed so if you look in the side by sides we have of the original versus the new one you'll see that difference in the actual fin stack the fans are different that's a large contributor to the performance change as well there's the cold plate tuning that comes from the three variations uh depending on which you want for the uh convexity and that'll cover kind of the key differences for the G2 so we need to test it still in a fixed heat load like for like scenario so they've got 180 watt 750x 250 wat 13900 K over here uh in these scenarios about 2 to 3 degrees different depending on what you're comparing so so I think ideally if you're trying to be perfect comparison you would look at G1 of course that's the gry line so that's the original that everyone knows and then standard is going to be probably the most typical or most comparable to the way that no used to do it as opposed to now where they have three different skews for the cold plate alone uh so 2 to 3 degrees at that heat load um as we always say temperature is not a fire strike score so it's going to change depending on what you're running but that gives you an idea for where it falls as for what two to 3° means in say one of our Benchmark charts if it were to materialize in such a way in one of our own tests that's the difference basically between kind of the middle of the pack the best coolers and the top so it matters it's a big difference uh but we'll test that later for a couple of other products quickly before we get into some of the engineering discussion with Jacob so I'm not going to spend any time over here I know nocta uh cares a lot about it but we don't do much with power supplies so this is a nocta addition power supply uh and we're going to Breeze past that but it's something they're working on there's also a super prototype two-phase thermos siphon which you see here so this is a pumpless in a sense a there's no mechanical moving parts to a pump uh for the thermos siphon we have covered thermos siphons in the past the fact that nocta is trying something like this is honestly super cool just because uh my entire experience with nocta has been air so it'll be something different this is there's no presence on a road map right now they're not making any promises for dates uh no no information about price so purely from a hey look at the thing we're doing standpoint the solution basically is uh it looks like about a a 240 standard-ish radiator with uh you got two tubes running down there's an evaporator there's condenser as the liquid or as the liquid heats up refrigerant actually uh it'll evaporate it travels through the thermos siphon and will recondense so in some ways some Concepts that are shared with just a heat pipe and that you have an evaporator you have a condenser the liquid does evaporate does move uh but the heat carrying capacity will be far greater and what no is shooting for is comparable to an AO solution because uh if you're going for liquid cooling levels of performance but you don't want the extra physical mechanical point of failure in a pump then there aren't many options so uh Roman shown a few in the past for example with his Novac fluid that's a whole different thing so hey Theros was pretty cool there's really no uh concrete product level detail on that for now the last thing on knock was tail before we move into the engineering is just uh some server side stuff so over here for example there is a Grace Hopper cooler we just saw some Blackwell servers uh with Grace in them in one of our other videos and otherwise there's a 120 mm fan that is also in the Gen 2 lineup so this uh it doesn't have a name so this this is the NF A14 Gen 2 G2 this is unnamed but otherwise like a smaller version of the nfa1 14 G2 you might say it's a NFA 12 G2 if I were to pick a name you can have that one for free I've coined it right here uh that's going to be that's my name of the product whether not to use it or not all right let's jump over to some engineering discussion now we are joined by Jacob for the engineering part of the discussion Jacob and I spoke last year at this booth and had an awesome engineering topic we actually have more coming up after the show there was something you wanted to clear up though yeah absolutely you mentioned that you want to see that photo of the D15 G2 coolers uh being packed uh I took this one yesterday at our Factory you can see it I zoom in a little bit so did you can count the heat pipe number 1 2 3 4 5 6 7 eight that's it's it is the D15 G2 they are being produced as we speak um uh being packed being shipped to warehouses around the world and uh yeah we uh aim to have them available on the market by end of June so it's happening so we have some really cool engineering topics uh that we'll get into kind of at least overview here so the fan uh is one where you were talking about these there's this little rubber bump here and that actually ends up being something somewhat significant so we're going to talk about that and then we've also got the cold blate curvature where do you want to start uh let's uh let's talk about the fans first okay uh because this is something we've discussed at length uh at last year's computex uh back then um our plan was to use uh LCP not only for the impeller but also for the frame to further strengthen it to avoid that deformation we saw um when using the fans on water cooling radiators but also with heating fan Clips um sadly um the the long-term tests we've been running back then didn't show the desired results so um even the stronger LCP um over longer periods of time uh will eventually have those deformation issues is basically just shifting the point of time where we uh come into critical territory it's basically whenever you have a um a plastic part bolted to a metal part eventually even the hardest plas will give in uh so uh yeah this is why uh the um targeted start of production date of end of last year didn't hold and uh while we're still um not having the fan out on the market um however um we have uh done some uh adjustments on the fan frame as such we've struct uh strengthened the structure of the fan frame as such and at the same time reduced the amount of force that goes into the fan and these little uh silicon pads are um part of the story uh you can see those little rings around the mounting holes uh we call them load relief rings and what they do is that they um distribute the counter pressure the fan gets when it's uh bolted down onto a radiator more evenly so without those load relief pads we have a compressive force from screw growing uh going down onto the fan and counter pressure from the radiator just at the very edge of the frame um and this enables uh the part around the mounting hole to go further down and torsional War being introduced into the frame and that torsional warp is uh causing the long-term deformation we've been battling with for so long so how how much of this is a a performance is it all performance oriented you're trying to prevent performance Problems by having gaps is it structural is you know what are the key parts of it I mean I would say it goes hand inand we need that supertight tip clearance to achieve the desired performance levels so we don't want to compromise on that um but it's basically a lifetime um issue uh lifespan issue um that uh requires us to avoid the deformation because if you would have a fan running for let's say 68 years at relatively high temperatures this is where you could run into that type of issue um and this is where the new load relief pads and the strength and frame step in to pre to prevent that type of issue because we want our fans to run for at least a decade or even longer if possible one thing to keep in mind with this is that we still don't want want you to torque the fans uh excessively um we Advocate uh 2 Newton met maximum torque which may sound like uh very low torque to apply um but with your average radiator screw that will amount to a total compressive load on the entire fan of around 30 kilograms so uh it's not like the fan is fragile it's uh just that it's quite easy to apply a lot of force with screws especially because some of the screw heads are very small so you have very little friction between the screw head and the fan frame also sometime lubricants are used to prevent the black uh black paint from wearing too quickly and those lubricants also reduce the friction between the the screw hat and the fan frame was there anything else you know in this topic you want to go over uh before moving over to the IH stuff yeah I mean one thing we could um explain as well is uh uh timelines and why the timelines differ for the round frame and for the for the square frame in both cases we initially Target using LCP for the frame um and in both Cas we ended up um thankfully getting away with um just structurally um modifying the frame to be stronger and reducing the amount of force that goes into the frame when mounting um so in both cases we could now stick to ABS PBT uh which enables us to stay at the the 40 EUR 40 USD price point uh rather than going even higher which would have been necessary if you used uh LCP the modification we had to do on the round frame uh was slightly um uh slightly smaller than what we had had to do on the square frame to make it work H and this is why the round frame fan is uh coming with the cooler uh end of June and uh the square frame uh is going to come uh in September um one more note on Square uh frame uh that is for the beijan Brown version Black Version we know people want it it's in the works Uh current Target is q1 um but we've talked at length last year about how difficult it is to create injection molding toolings and uh to make the to validate that everything is working perfect so before we want to get going with the black one we need to make sure that the regular one is going okay and then we'll uh do our best to get the the black one out for people as as quickly as we can and I forgot to say this in the first part but it's already publicly known the white fan is just kind of on hold right now with with no target date so uh so these are the two that you guys are working on with this one immediate future um anything else here before moving over I mean we could talk a little bit about uh what sort of performance metrics to expect um will it be worth the upgrade over the existing A14 uh we've got some performance data here um noise normalized results on a uh on a water cooling radiator um we're seeing a roughly at 3° Improvement which is which is massive um we have a little bit less when the resistance is slightly lower like on a heat SN but this is still more than two Dee Improvement uh in The Benchmark we're showing here one little explain before people complain about the graphs um some people may notice uh that the temperatures are very similar for the 200 watt uh results on the water cooling radiator and 320 watt results for the cooler the reason is that we used the hotspot um centered heating element for the water cooling tests and the uniform uh heating element for the cooler tests and that's why the the Delta temperatures are similar despite the the different heat loads it's it's like watching when it's I feel like I I feel like you would do very well running a uh a hardware review YouTube channel cuz you're already caveat everything which is what I have to do I know that that comment would come and it's a it's a valid point to make I mean if I was looking at that data I would say hey there's something wrong here and uh yeah there's something wrong we forgot to add that the one test was uh hot spot and the other was one was uniform but the data is fine it's just the no and clater so um um let's move over here so so quick primary we have a a video on the channel for those of you who haven't seen it it gets you some laser scans and looking at IHS curvature and especially Intel with the ilm so by pure coincidence uh we were both putting together some presentation at the same time probably you've been running the test a little longer uh but some of the we had some similar conclusions one of the comments I made in that video was you know for a manufacturer to really take advantage of this they'd have to make multiple SKS of the same thing which is what you guys are doing and why it was so awesome that that coincidence lined up cuz I'm actually primed to talk about this um I I think I want to since you gave me the run through I'll just Define kind of the grid we have and then I'll let you take it away all right so key points to look at when vital shows the uh the footage on the screen there's this block over here where the heat sink is Illustrated with its in its curvature uh for over here the basically axis label if you want to look at it that way then there's CPU uh they have the CCD presentence for AMD and then it's uh the different so the HBC LBC and standard there's some thermal numbers at the bottom up top there's the actual uh deviation that we'll talk about and then there's also offset mounting so basically if you look at it like a matrix it's standard mounting offset mounting HBC standard LBC and then the presence of or the the actual CPU Cur uh heat SN curvature let's run through um I want to just start with this gigantic deviation we're seeing here and I'll let you explain it yeah uh so I think it's become quite common knowledge you've shown it again in your uh recent video that Intel LGA 1700 CPUs deform heavily Under ilm Pressure so uh once you close the internal loading mechanism you have those two tabs pressing down on the side of the uh of the integrated heat spread better and that will cause it uh to uh warp uh quite heavily uh it's actually uh a complex warp so you're seeing a slight uh uh convexity in east to west Direction but that's very slight but you're seeing a huge concavity in north to south Direction and that's what really impacts uh cooler contact quality uh now the interesting thing is that um when you take a fresh a factory new Intel CPU uh scan it it's actually going to be a convex so uh we're north to south that is uh we're talking about something like around 15 Micron range uh convexity and the first time you put it into the socket and close the ilm it will go to concave around uh 25 to 35 uh microns uh yeah it's important to keep in mind that the the deformation you see we see is partly plastic partly elastic so when you take the CPU out of the socket it will go back but it will not go back all the way and that will depend on multiple factors like how much pressure your ilm actually applied their tolerances to that um how much um uh high heat load testing you did how long you had it in the socket so one thing we definitely see is uh uh a increasing deformation over time so it's not like you you close the ilm for the first time and that's it uh but the deformation will progress and uh progress to different levels depending on uh the parameters I've I've already mentioned heat sync mounting pressure is also parameter that uh that comes into play uh the bottom line is that we have seen CPUs that go up to around 55 Micron uh concavity and that means that um if you put a flat heat SN on that uh on that CPU you will have basically no contact whatsoever over the over it's it's like a lake you know like it's basically forming a a lake so um it the the key problem is that it it introduces a lot of complexity uh not only for customers but also for um Heat Sy reviewers because it means that uh the results May look different between what you're getting on a fresh CPU and what you're getting I don't know let's say three months down the line uh if the CPU has been used heavily for for cooler testing just because it has deformed much a lot and the the the uh contact situation for each heat heat sink has changed now how about for that specific comment how about for AMD what's your observation been there well the AMD IHS is much thicker uh and this is why we see much much less deformation for am uh am5 um when you look at pure numbers alone uh it's actually not so dissimilar to what we see from intel if you do the washer mod we're going to come to that in a minute I guess or or using contact frames um but when you look at the at the surface SC we have up there uh you can see that the deformation we get on am5 is actually very limited to the areas where the socket actuation mechanism presses onto the heat spreader um and the main contact area for the heat SN is staying relatively flat and this is why we generally see very good results with relatively flat um heat sinks on AMD so uh then there's these we call them pseudocolor in our lab but these pressure scans I guess uh well are you is it a pressure scan or is it a are you trying to show a hot spot here pressure this is a pressure scan okay this is purely pressure so pressure scans uh walk me through some of these for the the different skew so maybe we should start with AMD first because it's perhaps a little bit easier um we have the you already mentioned the HBC the high based convexity version the standard regular version that has the same medium base convexity like all our other multi-socket coolers do as well and the low base convexity LBC version and when you look at the um at the pressure SCS you can see that the HBC version has a very concentrated pressure pattern on am5 and it's uh like straight on the iodi with without the offset mounting and shift it towards the ccds with the offset mounting but still the contact area is relatively small and if you move on to the um uh to the regular version you can see a slightly bigger contact patch that's nicely spreading across the ccds with the offset mounting and that's why we see excellent results with the regular version on AMD once you use the offset in case however you don't want to use that offset mounting the LBC version is going to give you excellent contact quality even without the offset so you can see that it's got very nice contact both over the ccds and over the iodi so uh if you want to have the best temperatures on both the ccds and the iodi the LBC version is what you want to get so we should we should probably I didn't talk about this in the first part of the video I probably should have uh so for the I addressing confusion right I asked you what are the reasons you wouldn't just always do this some of them are kind of obvious in terms of making three different variations some are obvious the biggest one is definitely customer knowledge so we should briefly go over that just as simply as possible how does nocta see each of these three uh most optimally being deployed precisely I mean uh that's why I was so happy when I saw your video coming out uh because uh customer education is going to be key here um I think the the first and perhaps most important message is to get out is you cannot really go wrong with the standard version The Standard Version is the best allrounder by far it's going to give you the best possible results on am5 with the offset mounting and uh we're including shim washers to do uh to reduce ilm pressure on Intel LGA 1700 and this way it's going to give you the best possible results on a fresh Intel 1700 CPU as well the only case where the standard version uh is going to lag behind a little bit is if your Intel CPU has already been in Ed and has been heavily used so that it has become heavily deformed this is when you definitely want to go with the with the HBC version so um I think uh for customers uh who feel confused about this uh or say I just don't want to bother with all that just go for the standard version uh you're going to be happy with it uh but it's really uh those customers who uh know their exact use case and they want to Edge out that extra one or two degrees that's where you might want to go go with the LBC or HBC version I guess example to is like lapping or something where the LBC if I understand it right if you were to lap your CPU uh that would get you the the closest to Flat I guess of the three I mean La lapping is a tricky bu business you you would have to know what you do uh and you would probably have to lap it in socket to actually get it flat in socket otherwise you may up end up with with weird results um but so this is this is where we have to assume a spherical cow again I was afraid I was afraid to bring up laughing cuz I was thinking like through variables as well but yes no I agree s sadly it's not as straightforward as it seems um but there are also custom heat spreaders so you could replace your heat spreader with a custom one that should be actually flat because it's properly made hopefully by a CNC machine and that's definitely a case where you would want to go with an LBC cooler another case would be if you're running Legacy sockets like if you're still on LGA 2066 uh on am4 on 2011 uh or if you just for some reason don't want to use the offset mounting we offer then the LBC is going to be a great option for you is there uh anything you want to go over on the Intel side over here so yeah I mean Intel um it's basically uh the other way around so if we start with the LBC again um with we see that with full ilm pressure there's hardly any contact in the critical uh die area um if you use the um 1 mm washers we Supply you actually starting to get decent contact even with the low base convexity version and uh if we move up to the standard um Middle based convexity version you can see that with the washers on a fresh um or relatively fresh CPUs you're getting the perfect nice concentrated uh contact you want to have exactly over your die you can also see that the standard version without washers uh is lacking a little bit of contact in the in the um Central Area so it's not the end of the world it's going to do okay uh but there's room for improvement so if you don't want to use those washers or a contact frame for example then you would want to go for the HP version that has excellent contact UM even uh if you don't use the washers on a fresh CPU if you put in the washers with HPC you're actually ending up with a little bit too concentrated pressure um but if you have a heavily really heavily deformed CPU like uh we've had some that were uh used for heating testing for months then actually combining HBC with the washers can give you the best results so uh yeah I understand it's a it's a complicated story to tell uh but sadly the situation is complicated so it's uh I'm afraid it doesn't get much more straightforward than that and also I mean frankly so it's it's a complicated story to tell and the testing of course but ultimately from uh we'll look at all this in our benchmarking and probably do like a matrix I guess we'll end up with like a nine test Matrix or something but it's going to be a big yeah so ultimately from a uh a someone building a computer perspective hopefully you know as reviewers our job is to just give the answer as much as we can so it should simplify it a bit if this is overwhelming uh I I don't think it'll be necessary for people to necessarily have a full comprehension of all this which um we'll we'll talk about in the review so um Jacob why don't you just make something easier have you ever thought about so you talk about like how different color fans is different difficulties do you ever think about just not caring you know put some LEDs in it maybe $9.99 uh maybe it moves air maybe it doesn't yeah what do you think I guess it's just not our cup of tea sorry I mean for for this one honestly we were debating if we should go for the three different skus because uh it's difficult for from a customer education perspective it's uh difficult more difficult at least from an economic of scale uh perspective it uh requires very tight uh tolerance control in manufacturing but we do have that anyway we did control base convexity with every heat syn that that leaves the factory anyway so that is doable for us and then in the end what we want to do is offer the best possible uh solution for um our target audience our target audience is caring about performance first and foremost and that's what we're going for so check back for the review and for more discussions with Jacob we have a couple engineering videos we already shot they'll be up after the show and thank you again for joining me my pleasure thank you very much for coming absolutely and we'll see you all next timeI promise it's actually real this time so the nocta nhd15 G2 just to clarify a few things uh this is not the video from 2023 you're in the right video please don't click away this is the new one all right so now we've cleared that up so the Gen 2 D15 is actually happening I have even requested a proof of life photo uh and what I've been told is that such photo does exist and it will be dispatched to me post haste and the cooler is actually in the process of shipping so it sounds like June is I'm getting the nod so it sounds like June is when the D15 G2 is happening uh there is a lot to talk about today so we're just split this video uh into kind of two halves of the same video I'm going to focus on product stuff I will be joined by Jacob off camera here uh in a little bit to talk about some of the specifics but to get into the details straight away on the G2 there are a couple things you need to know the first one is that this is the cooler uh but we're going to be covering some familiar details to people who saw our cold plate and IHS flatness discussion there's some really specific detail on fan uh mounting and structure as well and then uh they also have a a setup here just showing the original D15 that um later might be running some heat loads on it all right the details that matter the most right away uh $150 and is going to be an expensive air cooler a lot of that comes in from the fan before that this video is brought to you by thermal Grizzlies aeronaut and hydronaut thermal pastes aeronaut is thermal Grizzly's entry-level thermal solution marketed as resistant to curing and for long-term endurance hydronaut is thermal Grizzly's Next Step Up targeted for overclocking and higher performance applications we've used hydronaut on a lot of our systems internally over the years you can learn more at the link in the description below so for the $150 kit what they are shipping is the cooler itself the two fans and then this small piece of paper down here has four washers on it so these are 1 mm washers it's for an ilm washer mod we're going to get into detail on that uh with some of these pressure Maps too so noct is at a point now where as we've seen with air coolers in general to further tune to get any additional performance out of it you have to start doing tricks uh that may be outside of the cooler itself so as an example where we see the best coolers in our testing right now there's kind of like a 3° cluster of plus or minus and they're all basically in that range if they're the best coolers fighting for an extra 1 to 2° is very difficult at this point and where KNX was trying to do that with the g2 is going to be in the three variations so they have uh an LBC or low based convexity HPC or high based convexity and then standard standard is represented by the uh green line over here and one of the things we'll be going over is some of the temperature differences that nocta is seen based on uh what sort of CPU you're pairing with the cold plate and so the problem comes in with actually looking at flatness again uh this is an am5 solution this CPU only this CPU in socket and then we've got the basically the deviation and flatness so the problem we saw with intel was that when you clamp the ilm down it has that really deep concavity uh and that gets worse noct was saying over time which is not an aspect that we explored in our testing so that's incredibly interesting because it causes all kinds of new and wonderful problems for our test benches that uh we just deployed so I'll get to talk about that more soon back to the G2 though so other differences for the cooler include the heat pipes they're going up to eight heat pipes from six these are centered heat pipes they uh say they've done some heat pipe tuning and in addition to that uh the fin pitch and density has changed so if you look in the side by sides we have of the original versus the new one you'll see that difference in the actual fin stack the fans are different that's a large contributor to the performance change as well there's the cold plate tuning that comes from the three variations uh depending on which you want for the uh convexity and that'll cover kind of the key differences for the G2 so we need to test it still in a fixed heat load like for like scenario so they've got 180 watt 750x 250 wat 13900 K over here uh in these scenarios about 2 to 3 degrees different depending on what you're comparing so so I think ideally if you're trying to be perfect comparison you would look at G1 of course that's the gry line so that's the original that everyone knows and then standard is going to be probably the most typical or most comparable to the way that no used to do it as opposed to now where they have three different skews for the cold plate alone uh so 2 to 3 degrees at that heat load um as we always say temperature is not a fire strike score so it's going to change depending on what you're running but that gives you an idea for where it falls as for what two to 3° means in say one of our Benchmark charts if it were to materialize in such a way in one of our own tests that's the difference basically between kind of the middle of the pack the best coolers and the top so it matters it's a big difference uh but we'll test that later for a couple of other products quickly before we get into some of the engineering discussion with Jacob so I'm not going to spend any time over here I know nocta uh cares a lot about it but we don't do much with power supplies so this is a nocta addition power supply uh and we're going to Breeze past that but it's something they're working on there's also a super prototype two-phase thermos siphon which you see here so this is a pumpless in a sense a there's no mechanical moving parts to a pump uh for the thermos siphon we have covered thermos siphons in the past the fact that nocta is trying something like this is honestly super cool just because uh my entire experience with nocta has been air so it'll be something different this is there's no presence on a road map right now they're not making any promises for dates uh no no information about price so purely from a hey look at the thing we're doing standpoint the solution basically is uh it looks like about a a 240 standard-ish radiator with uh you got two tubes running down there's an evaporator there's condenser as the liquid or as the liquid heats up refrigerant actually uh it'll evaporate it travels through the thermos siphon and will recondense so in some ways some Concepts that are shared with just a heat pipe and that you have an evaporator you have a condenser the liquid does evaporate does move uh but the heat carrying capacity will be far greater and what no is shooting for is comparable to an AO solution because uh if you're going for liquid cooling levels of performance but you don't want the extra physical mechanical point of failure in a pump then there aren't many options so uh Roman shown a few in the past for example with his Novac fluid that's a whole different thing so hey Theros was pretty cool there's really no uh concrete product level detail on that for now the last thing on knock was tail before we move into the engineering is just uh some server side stuff so over here for example there is a Grace Hopper cooler we just saw some Blackwell servers uh with Grace in them in one of our other videos and otherwise there's a 120 mm fan that is also in the Gen 2 lineup so this uh it doesn't have a name so this this is the NF A14 Gen 2 G2 this is unnamed but otherwise like a smaller version of the nfa1 14 G2 you might say it's a NFA 12 G2 if I were to pick a name you can have that one for free I've coined it right here uh that's going to be that's my name of the product whether not to use it or not all right let's jump over to some engineering discussion now we are joined by Jacob for the engineering part of the discussion Jacob and I spoke last year at this booth and had an awesome engineering topic we actually have more coming up after the show there was something you wanted to clear up though yeah absolutely you mentioned that you want to see that photo of the D15 G2 coolers uh being packed uh I took this one yesterday at our Factory you can see it I zoom in a little bit so did you can count the heat pipe number 1 2 3 4 5 6 7 eight that's it's it is the D15 G2 they are being produced as we speak um uh being packed being shipped to warehouses around the world and uh yeah we uh aim to have them available on the market by end of June so it's happening so we have some really cool engineering topics uh that we'll get into kind of at least overview here so the fan uh is one where you were talking about these there's this little rubber bump here and that actually ends up being something somewhat significant so we're going to talk about that and then we've also got the cold blate curvature where do you want to start uh let's uh let's talk about the fans first okay uh because this is something we've discussed at length uh at last year's computex uh back then um our plan was to use uh LCP not only for the impeller but also for the frame to further strengthen it to avoid that deformation we saw um when using the fans on water cooling radiators but also with heating fan Clips um sadly um the the long-term tests we've been running back then didn't show the desired results so um even the stronger LCP um over longer periods of time uh will eventually have those deformation issues is basically just shifting the point of time where we uh come into critical territory it's basically whenever you have a um a plastic part bolted to a metal part eventually even the hardest plas will give in uh so uh yeah this is why uh the um targeted start of production date of end of last year didn't hold and uh while we're still um not having the fan out on the market um however um we have uh done some uh adjustments on the fan frame as such we've struct uh strengthened the structure of the fan frame as such and at the same time reduced the amount of force that goes into the fan and these little uh silicon pads are um part of the story uh you can see those little rings around the mounting holes uh we call them load relief rings and what they do is that they um distribute the counter pressure the fan gets when it's uh bolted down onto a radiator more evenly so without those load relief pads we have a compressive force from screw growing uh going down onto the fan and counter pressure from the radiator just at the very edge of the frame um and this enables uh the part around the mounting hole to go further down and torsional War being introduced into the frame and that torsional warp is uh causing the long-term deformation we've been battling with for so long so how how much of this is a a performance is it all performance oriented you're trying to prevent performance Problems by having gaps is it structural is you know what are the key parts of it I mean I would say it goes hand inand we need that supertight tip clearance to achieve the desired performance levels so we don't want to compromise on that um but it's basically a lifetime um issue uh lifespan issue um that uh requires us to avoid the deformation because if you would have a fan running for let's say 68 years at relatively high temperatures this is where you could run into that type of issue um and this is where the new load relief pads and the strength and frame step in to pre to prevent that type of issue because we want our fans to run for at least a decade or even longer if possible one thing to keep in mind with this is that we still don't want want you to torque the fans uh excessively um we Advocate uh 2 Newton met maximum torque which may sound like uh very low torque to apply um but with your average radiator screw that will amount to a total compressive load on the entire fan of around 30 kilograms so uh it's not like the fan is fragile it's uh just that it's quite easy to apply a lot of force with screws especially because some of the screw heads are very small so you have very little friction between the screw head and the fan frame also sometime lubricants are used to prevent the black uh black paint from wearing too quickly and those lubricants also reduce the friction between the the screw hat and the fan frame was there anything else you know in this topic you want to go over uh before moving over to the IH stuff yeah I mean one thing we could um explain as well is uh uh timelines and why the timelines differ for the round frame and for the for the square frame in both cases we initially Target using LCP for the frame um and in both Cas we ended up um thankfully getting away with um just structurally um modifying the frame to be stronger and reducing the amount of force that goes into the frame when mounting um so in both cases we could now stick to ABS PBT uh which enables us to stay at the the 40 EUR 40 USD price point uh rather than going even higher which would have been necessary if you used uh LCP the modification we had to do on the round frame uh was slightly um uh slightly smaller than what we had had to do on the square frame to make it work H and this is why the round frame fan is uh coming with the cooler uh end of June and uh the square frame uh is going to come uh in September um one more note on Square uh frame uh that is for the beijan Brown version Black Version we know people want it it's in the works Uh current Target is q1 um but we've talked at length last year about how difficult it is to create injection molding toolings and uh to make the to validate that everything is working perfect so before we want to get going with the black one we need to make sure that the regular one is going okay and then we'll uh do our best to get the the black one out for people as as quickly as we can and I forgot to say this in the first part but it's already publicly known the white fan is just kind of on hold right now with with no target date so uh so these are the two that you guys are working on with this one immediate future um anything else here before moving over I mean we could talk a little bit about uh what sort of performance metrics to expect um will it be worth the upgrade over the existing A14 uh we've got some performance data here um noise normalized results on a uh on a water cooling radiator um we're seeing a roughly at 3° Improvement which is which is massive um we have a little bit less when the resistance is slightly lower like on a heat SN but this is still more than two Dee Improvement uh in The Benchmark we're showing here one little explain before people complain about the graphs um some people may notice uh that the temperatures are very similar for the 200 watt uh results on the water cooling radiator and 320 watt results for the cooler the reason is that we used the hotspot um centered heating element for the water cooling tests and the uniform uh heating element for the cooler tests and that's why the the Delta temperatures are similar despite the the different heat loads it's it's like watching when it's I feel like I I feel like you would do very well running a uh a hardware review YouTube channel cuz you're already caveat everything which is what I have to do I know that that comment would come and it's a it's a valid point to make I mean if I was looking at that data I would say hey there's something wrong here and uh yeah there's something wrong we forgot to add that the one test was uh hot spot and the other was one was uniform but the data is fine it's just the no and clater so um um let's move over here so so quick primary we have a a video on the channel for those of you who haven't seen it it gets you some laser scans and looking at IHS curvature and especially Intel with the ilm so by pure coincidence uh we were both putting together some presentation at the same time probably you've been running the test a little longer uh but some of the we had some similar conclusions one of the comments I made in that video was you know for a manufacturer to really take advantage of this they'd have to make multiple SKS of the same thing which is what you guys are doing and why it was so awesome that that coincidence lined up cuz I'm actually primed to talk about this um I I think I want to since you gave me the run through I'll just Define kind of the grid we have and then I'll let you take it away all right so key points to look at when vital shows the uh the footage on the screen there's this block over here where the heat sink is Illustrated with its in its curvature uh for over here the basically axis label if you want to look at it that way then there's CPU uh they have the CCD presentence for AMD and then it's uh the different so the HBC LBC and standard there's some thermal numbers at the bottom up top there's the actual uh deviation that we'll talk about and then there's also offset mounting so basically if you look at it like a matrix it's standard mounting offset mounting HBC standard LBC and then the presence of or the the actual CPU Cur uh heat SN curvature let's run through um I want to just start with this gigantic deviation we're seeing here and I'll let you explain it yeah uh so I think it's become quite common knowledge you've shown it again in your uh recent video that Intel LGA 1700 CPUs deform heavily Under ilm Pressure so uh once you close the internal loading mechanism you have those two tabs pressing down on the side of the uh of the integrated heat spread better and that will cause it uh to uh warp uh quite heavily uh it's actually uh a complex warp so you're seeing a slight uh uh convexity in east to west Direction but that's very slight but you're seeing a huge concavity in north to south Direction and that's what really impacts uh cooler contact quality uh now the interesting thing is that um when you take a fresh a factory new Intel CPU uh scan it it's actually going to be a convex so uh we're north to south that is uh we're talking about something like around 15 Micron range uh convexity and the first time you put it into the socket and close the ilm it will go to concave around uh 25 to 35 uh microns uh yeah it's important to keep in mind that the the deformation you see we see is partly plastic partly elastic so when you take the CPU out of the socket it will go back but it will not go back all the way and that will depend on multiple factors like how much pressure your ilm actually applied their tolerances to that um how much um uh high heat load testing you did how long you had it in the socket so one thing we definitely see is uh uh a increasing deformation over time so it's not like you you close the ilm for the first time and that's it uh but the deformation will progress and uh progress to different levels depending on uh the parameters I've I've already mentioned heat sync mounting pressure is also parameter that uh that comes into play uh the bottom line is that we have seen CPUs that go up to around 55 Micron uh concavity and that means that um if you put a flat heat SN on that uh on that CPU you will have basically no contact whatsoever over the over it's it's like a lake you know like it's basically forming a a lake so um it the the key problem is that it it introduces a lot of complexity uh not only for customers but also for um Heat Sy reviewers because it means that uh the results May look different between what you're getting on a fresh CPU and what you're getting I don't know let's say three months down the line uh if the CPU has been used heavily for for cooler testing just because it has deformed much a lot and the the the uh contact situation for each heat heat sink has changed now how about for that specific comment how about for AMD what's your observation been there well the AMD IHS is much thicker uh and this is why we see much much less deformation for am uh am5 um when you look at pure numbers alone uh it's actually not so dissimilar to what we see from intel if you do the washer mod we're going to come to that in a minute I guess or or using contact frames um but when you look at the at the surface SC we have up there uh you can see that the deformation we get on am5 is actually very limited to the areas where the socket actuation mechanism presses onto the heat spreader um and the main contact area for the heat SN is staying relatively flat and this is why we generally see very good results with relatively flat um heat sinks on AMD so uh then there's these we call them pseudocolor in our lab but these pressure scans I guess uh well are you is it a pressure scan or is it a are you trying to show a hot spot here pressure this is a pressure scan okay this is purely pressure so pressure scans uh walk me through some of these for the the different skew so maybe we should start with AMD first because it's perhaps a little bit easier um we have the you already mentioned the HBC the high based convexity version the standard regular version that has the same medium base convexity like all our other multi-socket coolers do as well and the low base convexity LBC version and when you look at the um at the pressure SCS you can see that the HBC version has a very concentrated pressure pattern on am5 and it's uh like straight on the iodi with without the offset mounting and shift it towards the ccds with the offset mounting but still the contact area is relatively small and if you move on to the um uh to the regular version you can see a slightly bigger contact patch that's nicely spreading across the ccds with the offset mounting and that's why we see excellent results with the regular version on AMD once you use the offset in case however you don't want to use that offset mounting the LBC version is going to give you excellent contact quality even without the offset so you can see that it's got very nice contact both over the ccds and over the iodi so uh if you want to have the best temperatures on both the ccds and the iodi the LBC version is what you want to get so we should we should probably I didn't talk about this in the first part of the video I probably should have uh so for the I addressing confusion right I asked you what are the reasons you wouldn't just always do this some of them are kind of obvious in terms of making three different variations some are obvious the biggest one is definitely customer knowledge so we should briefly go over that just as simply as possible how does nocta see each of these three uh most optimally being deployed precisely I mean uh that's why I was so happy when I saw your video coming out uh because uh customer education is going to be key here um I think the the first and perhaps most important message is to get out is you cannot really go wrong with the standard version The Standard Version is the best allrounder by far it's going to give you the best possible results on am5 with the offset mounting and uh we're including shim washers to do uh to reduce ilm pressure on Intel LGA 1700 and this way it's going to give you the best possible results on a fresh Intel 1700 CPU as well the only case where the standard version uh is going to lag behind a little bit is if your Intel CPU has already been in Ed and has been heavily used so that it has become heavily deformed this is when you definitely want to go with the with the HBC version so um I think uh for customers uh who feel confused about this uh or say I just don't want to bother with all that just go for the standard version uh you're going to be happy with it uh but it's really uh those customers who uh know their exact use case and they want to Edge out that extra one or two degrees that's where you might want to go go with the LBC or HBC version I guess example to is like lapping or something where the LBC if I understand it right if you were to lap your CPU uh that would get you the the closest to Flat I guess of the three I mean La lapping is a tricky bu business you you would have to know what you do uh and you would probably have to lap it in socket to actually get it flat in socket otherwise you may up end up with with weird results um but so this is this is where we have to assume a spherical cow again I was afraid I was afraid to bring up laughing cuz I was thinking like through variables as well but yes no I agree s sadly it's not as straightforward as it seems um but there are also custom heat spreaders so you could replace your heat spreader with a custom one that should be actually flat because it's properly made hopefully by a CNC machine and that's definitely a case where you would want to go with an LBC cooler another case would be if you're running Legacy sockets like if you're still on LGA 2066 uh on am4 on 2011 uh or if you just for some reason don't want to use the offset mounting we offer then the LBC is going to be a great option for you is there uh anything you want to go over on the Intel side over here so yeah I mean Intel um it's basically uh the other way around so if we start with the LBC again um with we see that with full ilm pressure there's hardly any contact in the critical uh die area um if you use the um 1 mm washers we Supply you actually starting to get decent contact even with the low base convexity version and uh if we move up to the standard um Middle based convexity version you can see that with the washers on a fresh um or relatively fresh CPUs you're getting the perfect nice concentrated uh contact you want to have exactly over your die you can also see that the standard version without washers uh is lacking a little bit of contact in the in the um Central Area so it's not the end of the world it's going to do okay uh but there's room for improvement so if you don't want to use those washers or a contact frame for example then you would want to go for the HP version that has excellent contact UM even uh if you don't use the washers on a fresh CPU if you put in the washers with HPC you're actually ending up with a little bit too concentrated pressure um but if you have a heavily really heavily deformed CPU like uh we've had some that were uh used for heating testing for months then actually combining HBC with the washers can give you the best results so uh yeah I understand it's a it's a complicated story to tell uh but sadly the situation is complicated so it's uh I'm afraid it doesn't get much more straightforward than that and also I mean frankly so it's it's a complicated story to tell and the testing of course but ultimately from uh we'll look at all this in our benchmarking and probably do like a matrix I guess we'll end up with like a nine test Matrix or something but it's going to be a big yeah so ultimately from a uh a someone building a computer perspective hopefully you know as reviewers our job is to just give the answer as much as we can so it should simplify it a bit if this is overwhelming uh I I don't think it'll be necessary for people to necessarily have a full comprehension of all this which um we'll we'll talk about in the review so um Jacob why don't you just make something easier have you ever thought about so you talk about like how different color fans is different difficulties do you ever think about just not caring you know put some LEDs in it maybe $9.99 uh maybe it moves air maybe it doesn't yeah what do you think I guess it's just not our cup of tea sorry I mean for for this one honestly we were debating if we should go for the three different skus because uh it's difficult for from a customer education perspective it's uh difficult more difficult at least from an economic of scale uh perspective it uh requires very tight uh tolerance control in manufacturing but we do have that anyway we did control base convexity with every heat syn that that leaves the factory anyway so that is doable for us and then in the end what we want to do is offer the best possible uh solution for um our target audience our target audience is caring about performance first and foremost and that's what we're going for so check back for the review and for more discussions with Jacob we have a couple engineering videos we already shot they'll be up after the show and thank you again for joining me my pleasure thank you very much for coming absolutely and we'll see you all next time\n"