Видео

The noise levels are not super loud, no where near old workstation or datacenter cards but 4500 rpm does make a good woosh noise when its pulling 350+ watts. I would say its no where near as quiet under load as a typical 3 fan card but its not distractingly loud.

Sorry I thought I responded to this. I would definitely go with the 4070 Ti Super. I am not a fan boy of any products, I have owned most cards from both sides and Intel. For a daily driver user experience NVIDIA is the way to go, way more stable, creation tools are all compatible with cuda for accelleration, runs much lower wattage, amazing undervolting

I have many opinions on cooling and cooling issues with the 7900 XTX. The 7900 XTX die size is so large that even at high wattage its thermal density in mm/sq is roughly the same as an RTX 4070. However 7900 XTX's in general do have issues with hotspot temps and I think this has mostly to do with how much pressure the cooler is pressing on the die and how flat the cooler/die is. I have noticed that all the 7900 XTX than have come my way that ran awesome temps on die/hotspot had insanely stiff spring bar mounts. And the 7900 XTX's that did not have good hotspot temps did not have high tension mounts. In the end there is so much die surface area between all the chiplets to get the heat into the cooler I think that a better solution is honestly the factory paste which on most every 7900 XTX is going to be something similar to PTM 7950 phase change material. You never know until you try though right :)

@The_Thermal_Channel that's interesting to know. Thankfully I don't have any hotspot issues on my 7900 XTX but I'm not cranking the power limit up either. I just don't think the performance gain is worth all the extra wattage and heat.

@@jinxPad What I have learned from tearing down probably a dozen or so 7900 XTX's from ASRock, MSI, Asus, XFX and a couple 7900 GRE's is if your hotspot temp delta is good just leave it be. Don't modify, change paste, take it apart. Be happy that its running nice and cool and because most all 7900 XTX's have phase change material from the factory it will not need maintenance for many years. And you are correct, if anything for raster performance these are so overkill unless you are maxing out AAA games at 4k and even then its a lot. You could probably even reduce power 10% and not notice any performance hit.

The hardware is legit good, it's still just the software and drivers holding back more performance. I run all 3 brands (AMD, NVIDIA, Intel) and rotate cards pretty much weekly just for fun. Nvidia is definitely so far ahead it becomes boring. My 4070 Super undervolted plays everything better than a 3090 Ti at max 170watts (with the undervolt) but it's somewhat boring if that makes sense. AMD scratches that itch for turning knobs and fooling around with things, stacking Hyper-X with AFMF and skipping in-game upscaling you can get ridiculous FPS. ARC though, arc gives me that new feeling and I am impressed whenever I install it. I have had the ASrock PG, Sparkle and this Acer a770 along with an a310 as a backup card. Every month that goes by it seems that games get better performing, I really hope Battlemage gets a full release and isn't dropped by Intel due to financial issues.

Yeah i'm excited for how battlemage'll go

And yeah i get what you mean, my brother got a nvidia gpu and it's just too good and stable but that doesn't mean their gpus are cheap. I'm going to get an arc for my first gpu because it's going to be a good budget card at least that's what i heard and that's their whole thing.

No worries about measurement, no issues with pad density requirements. Putty conforms to just about anything and in most cases has higher performance than pads. Its mostly reusable. The putty that is very messy in this video is actually a mix of a thick putty and a near paste like putty which is why it appears so goopy. I have been using putty for almost 2 years and have not looked back after maybe 75 cards or so. You see in this video that the blue putty is actually AMD factory, and I reused it for other components. Pads will have their place, but I pretty this stuff 10/10 times.

A couple of things sir, if that card is on the original paste, it maybe time for a refresh. For my 220w baseline test I am doing nothing special and even locking the fans at a fairly low speed. I do keep my testing area at 20c or 68-69f ish. Also those numbers are above ambient temps, so if the temp shows 45c it means 45c over 20c ambient which is most where this room stays at. Still a good deal cooler than than 77c. The 5700xt is still a good viable card so if you don't plan on upgrading and want to get some more life out of the card I would do a couple things: - At 220w, any name brand paste will do. If you are in the USA order a small tube of Noctua NT-H1 or H2 from amazon as it lasts a long time and is easy to work with. If you buy a 3g tube you could also re-paste your CPU with the left over paste. - Order either the appropriate thickness and size thermal pads or what I do is use thermal putty, Amazon has Upsiren thermal putty available, you will need at least 50g. I have some videos on my channel of applying thermal putty, you don't need to be that precise which makes it very easy compared to getting the right pad thickness and density. Upsiren LMTG-100 is extremely overkill for 220w and requires prep-work to safely use on any gpu. Unless you plan on running the card at 300+ watts for only a few % more FPS I would advise to just refresh the paste and pads/putty. The lower thermals will give you a slight boost in clock speed. If you are in the Central Texas area I can refresh your card for you for free. Takes about 15 minutes :)

@@The_Thermal_Channel first of all thank you very much for the detailed and great explanation! My card is actually limited to 180W instead of 220. I can move that limit to 270W but Im guessing anything above that would require bios flash and I deem it unnecesary since im yet to find FPS difference between 180 and 270W(could be my CPU as it never goes above 160 in cyberpunk at 2k for instance) The paste it has on is not the original, I repasted it but then my hotspot temps started to hit 110 degrees instantly. I decided to try GELID GC-extreme after researching a bit and it made my gpu a little cooler than the original state it was in. My vrams never go above 65 degrees Celsius so I guess they are fine. I ordered Honeywell PTM7950 after looking through pastes today for my laptop with 12700H and Rtx4060. It is enough for my laptop and GPU all together. Honestly I don't plan to upgrade my gpu anymore thanks to the new frame gen apps like losless scaling or fsr2fsr3. I am not a citizen of US unfortunately. Again thank you for your answer :) Great videos by the way will be subbing.

I'll try to get those into the hopper, thank you for the feedback. I appreciate the comments.

I think what you are seeing is a good company, that makes good quality products at fair prices. Sure, their fans are more expensive, but they are objectively better performing. It would make sense that their paste is also well tested and engineered for success. The same for Thermal Grizzly, be quiet!, Arctic etc...

That makes sense, the only real benefit is that the paste doesnt crack and break down at extremely low temperature. This paste must be for the 0.001% of users then. Still happy I tested it, it validates that most users should simply skip for something good and available like Noctua or TG.

I will acquire and test because you asked. I will guarantee that it will average in the lower 3rd of pastes performance wise but it has been ordered and should be here in a couple weeks.

Where do you buy that? I cannot find anything about it on google, amazon, aliexpress. Maybe I am spelling it wrong. Would you please clarify?

I have thought about it, but if you are testing on CPU IHS without de-lidding your surface area is just so insanely large, like 1400-1700mm sq vs one of my test cards at 251mm sq and produce less heat. There is the other factoid that the CPU is solder bonded to the IHS or in some cases multiple chiplets are bonded to the IHS creating weird uneven heat zones. A 300 watt overclocked 14900K will still have insanely low thermal density compared to a 300 watt GPU. Another issue with IHS is how much they bend. Some pastes may technically be better in moving heat but may not bond as well as something else rated much lower in thermal conductivity. I am thinking about it, honestly what I would probably do is just pick the most common CPU on the market and test on that. Blind guess is that you would see minimal delta between each paste though. I think you could put Thermal Grizzly, Upsiren, be quiet!, Thermalright or Noctua at random on a cpu and you would just have good performance. However, if you get into extreme overclocking where your CPU is pulling 400-500 watts then I think the thermal interface starts to really matter again, that thermal density comes way up and pumping heat between the die/IHS/cooler efficiently starts to make a difference. Thinking about it. I am testing (or retesting) DC2 Pro right now as I type this, and starting at my new 3g tube of Kingpin. I am looking forward to testing more quality products, I'm over testing the meme pastes :)

Its tiny capsules of liquid metal distributed in a thermal paste. The silicon paste encapsulates the liquid metal to a point. I would be comfortable putting it between my CPU and water block, to be honest right now I have straight liquid metal (DC2 Pro) between my i7 and waterblock and its just fine. The gallium (liquid metal) in LMTG-100 is definitely conductive and will eat aluminum, react with copper but it seems pretty safe compared to straight liquid metal. One thing I noticed testing was that it doesnt stain the cooler like liquid metal does, this is a good indication that the thermal paste is insulating the liquid metal capsules from the surface pretty well. During clean up though you will see little blobs of liquid metal, so it must be treated with respect for sure on application. This stuff is so good/performant and most of all, EASY to apply over liquid metal I ordered 2 more tubes of it for my GPU/CPU on my personal workstation.

I think PCM-1 and the Thermal Grizzly phase change sheet will probably all work and perform around the same levels. But I have not tested these products yet. The benefit from the PTM/PCM etc... sheets is that when it melts it forms an extremely thin bond between the surfaces. This means very low thermal resistance and more heat gets into the cooler faster.

@@The_Thermal_Channel I just ordered the PCM1 and PMT from Alie, let's see how they fair. Not that I am a big tester, will just re-apply on two old builds. And only because I want to change the coolers on them.

@@DragonXDrei Easiest thing you can do, cut it to size, just a hair smaller than the die/IHS. slap your heatsink on there. Hit it with a blow drier, 5 seconds on, 2 seconds off for a good minute to heat up the cooler/cpu. Make sure the screws are snugged up again and you should be good to go with little to no break in period. Let me know how it works out no matter how you install it. I like to hear the peoples different experiences on these more exotic products.

@@The_Thermal_Channel nice one, thanks for the info. Will have to see which one arrives first:)

This is the truth for sure. There should still be a 2 slot options for most new GPU chips.

No problemo, this is how I would attack that problem. The IHS size is roughly 1401mm sq which is quite large to be able to spread heat, of course under that IHS is the actual die which is much smaller. But our min interface size is 1401mm. The die underneath is: 125 mm sq TDP is 105w PPT is 142w Some reports show transient spikes up towards 175w, so let's use this as our target 175w over 200mm (75mm die + 125mm io die) is brutally high between the die and IHS, but its soldered: .875w/per mm sq Oddly this lines up pretty close to my test card at 220w. 175w over the IHS is light work but that heat is highly concentrated in 2 specific spots under the IHS: 0.125w/per mm sq over the IHS Here are the recommendations: In order of performance: 1: Upsiren LMTG100 - liquid metal captured in microcapsules embedded in paste, much safer to use than liquid metal direct. You have to order this from China via aliexpress. 2: Thermal Grizzly 38mm x 38mm Kryosheet - These are set it and forget it for life graphene pads and work amazingly well. This is what I typically run in my rigs. Via Amazon. 3: Thermal Grizzly Hydronaut - I wouldn't expect this to perform so well but it does at this specific heat density. Via Amazon. Kryosheet is the most expensive but honestly its clean, no mess, easy to apply (you just set it down on the cpu and screw your cooler back on. What is your air cooler, that my have the biggest impact on performance if your cooler is already at max heat load.

@@The_Thermal_Channel Currently Im using Corsair H100x Elite liquid cooler but I'm not impressed. Without anything running it suddenly spikes to mid 60's and currently it's at 54 degrees with only this RUclips page loaded in Edge. Since I plan to upgrade to a whole new rig in Jan I just ordered a cheap Air Cooler expecting it to cool my CPU better than this Corsair crap. (I hate iCue) . The air cooler I ordered is the Thlermalright Peerless Assassin 120 SE. I also wanted to thank you for you thorough answer. Much more than I expected, thanks a lot.

@@JaimeLacayoPTY My pleasure. You will find that cheap ThermalRight Peerless Assassin is better than about half of the AIO's on the market. I have an ThermalRight Phantom Evo Spirit which is just slightly better than it keeps up with my 280mm be quiet pure loop 2 and 280mm kraken elite aio's. The AIO's do handle transient temp spikes better but the Thermalright coolers are the best coolers per dollar you can buy right now.

Thank you. I appreciate the comment. I'm getting burned out on all these lower end pastes where it's not really worth the testing. I have 5 more 'low end' pastes but I think I'm just gonna toss them in a drawer for a few months so I can get back to hitting the more premium products. I just ordered Kingpin X, Kryonaut Extreme and a Kryosheet (which I have tested before but not on this channel). I think those products will produce much better numbers, and a much more interesting video. Thank's for watching.

All the paste lore I believe is heavily exaggerated by people who are honestly trying to make the best decisions for their investments they can. Working about which synthetic oil is best for your car, people loose their heads over and send used oil analysis in to Blackstone labs etc... The important part is just to change the oil regularly with a grade/spec that meets the criteria/requirement. I think the same with paste. It's a pseudo consumable item in enthusiast computing. In my experience of repasting maybe 75-100 different cards, some cards used for testing were pasted hundreds of times each, is that it does not matter that much. The internet as a whole puts so much focus on little details, then magnifies those details. Air will be squeezed out. You can have too little paste. And if the paste is thick and there is not any room to squeeze out around the die package you can have too much, but it would have to be a lot of paste and stuff that is very tough like KOLD-01. I just got that card (well both my titan Xps) into a new production plex server last night. They are currenting stacked tight together in a Lenovo p520 workstation tower. So not ideal, not a ton of air flow and the temps are: 28c for the bottom card, 30c for the top card with restricted air flow. Under load they are still pretty similar. I just installed a new AIO and I used the 9 dot pattern on the processor. Again, I think some of this stuff is blown up a bit too much, it's just not that serious for 99% of builds. I do keep the application consistent for testing because I want to try to minimize variables though, hence why I spread each paste roughly the same way on the test card's die in my other videos.

I wish we had bigger blower cards, they are rigid and reliable. Imagine a giant 3-4 slot single fan blower card.

It's the coolest GPU tech, so unfortunate that it costs so much to manufacture and is so complicated to assemble the dies flat. We see these types of dies in the datacenter but of course those cards are 5-500x the price. The vega's will always be cool :)

@@Bl3s5in Its not loud for a blower card but its louder than the normal triple fan coolers for sure. I wear headphones and didnt notice the sound, but fan noise doesnt really bother me. I will say if someone is sensitive to fan noise you probably do not want a blower style card in any form. You will get much better thermal performance and lower noise with typical cooler setups.

Yep I did post the teardown ruclips.net/video/NlejtjR1CZc/видео.html

Are you referring to PyTorch or Stable Diffusion?

I assume you are referring to stable diffusion, my instructions have it running on windows, not WSL2. The WSL2 Ubuntu instance is for running pytorch with rocm.

@@The_Thermal_Channel ooohhh, my misunderstanding!

I don't think it would be very easy to get out of the tube. It's possible but I felt like I would break the tube trying to get it out, then spreading it would be difficult. Its not bad stuff but honestly I would buy the sheet of 7950 over either paste 9/10 times. I think it's just that much easier to work with. You can look at Halnziye HY-P16 (which I have not uploaded that video yet, still working on it) which is also a paste based phase change but it acts much more like a typical paste at room temperature. I should have that video out shortly.

@@The_Thermal_Channel Ah, thanks! Should I spread it myself or just apply a pea-sized amount or a bit larger and then finish it off?

@@Vinprvx If you apply heat with a blow dryer you should definitely spread it out even if just a little bit. It does not have to be perfect and if its warm with heat it will spread easily. Even if you don't it just means you will need to run some hot/cold cycles (gaming) through the card and it will settle in.

Também estou garimpando, se você morar em cidade quente e não tiver ar condicionado é bom ver projetos mais frios

Yes, there are much cheaper options, you can get other 7900 XTX models for $200 USD less but it's still going to perform like a 7900 XTX, so it games very well.

No the card is datacenter grade and very well built. The cooler is exceptional, typically 2 slot blower cards do not enjoy anything above 275-300 watts but this card runs great with a custom fan curve. How much is the price in Brazil? Im very curious because in the USA its $1070 or $R 5,843

​@@The_Thermal_Channel This creator model is R$ 4650 today (851 USD). The ASUS TUF for example is R$ 6300 (1116 USD).

@@josealgarrajr Wow if that is the 7900 XTX model and not the 7900 XT model that is an insanely good price. Awesome.

@@The_Thermal_Channel This is for de 7900 XT only. But this is the cheapest 7900 XT available here, if the XTX runs fine the XT probably also will.

​@@josealgarrajreu também estou namorando uma RX 7900 XT, mas essa placa de vídeo é feiosa de mais.

Yes, that is how I did it in the past. Use low speed or low heat, and pulse the heat, like 3 seconds on 1 second off. Be patient. Pre-heating the tube make sure to keep the cap off and keep the nozzle of the tube down towards the cpu/gpu die as it may just start flowing out. When it does get to 45c and turn into liquid put a pea size blob and then keep the heat on. You may need someone else to hold the blow drier at that point, but you want heat all the way until you clamp the heatsink down to the die.

@@The_Thermal_Channel mine came in a syringe w but will try. I'm want to repaste a mobile 2080 and plan to use CXH 1300 for the VRAM and VRMs

@@OteimPrincewill Yeah thats perfect and a good stack with CX-H1300 which I have like 3 tubs of. I have used CX-H1300 on many many cards, it works great. In syringe form it will be very difficult to get it to even push out without warming up the tube with a blow drier. Just reach out if you need any help.

@@The_Thermal_Channel I have a baby milk warmer, can I put the syringe in that to warm it up?

This is an excellent question. I assume that there is some additive chemical keeping the PTM 'fresh' and possibly this is just Honeywell CYA because I haven't heard of or seen on discords or other channels where any of the PTMs wear out. I tend to agree that any of their PTM products will most likely last several years.

I didn't take it that way. You likes what you likes. But I will take that feedback and start adding in more 'real voice' content.

Oh man do I sound that robotic? Lol I can go back to the AI voice overs.

@@The_Thermal_Channel sounds like some serious tweaking whether it's reverb, bass or filters of some kind, some would be fine and appear as high quality mic though need to try get a natural sound

@@The_Thermal_Channel could be a bad mic or how you processed the video idk

Maybe its just too sensitive and your manly voice is too much for my pure ears. using my speakers and turning bass all the way down sounds more natural to me.

@@Jumoana I pinged this off the wife and she said it sounds like me. The only filter is the noise canceling in OBS (cpu based). I could pull bass out but I am a fairly large child (6'4-6'5) with a big morning voice sometimes.

@@ElSkeppo please watch one of the latest videos for updated graphs, thank you

The good/bad thing about pastes is that they all perform about the same at "normal" heat density but then you run into outliers that perform much better or provide a ton of value especially if you are a tinkerer. The performance will also change depending on the heat load over an area squared. Some interfaces will transfer heat vertically and horizontally well, some can spread heat out horizontally great but not so well vertically which I think causes a lot of confusion in forums and discord channels. I have about 20 more thermal interfaces on the way. Thanks for checking out my little channel.

@@The_Thermal_Channel I see what you mean. I have a large tube of GD900 that I use for older pc's and I somehow ended up with a big tube of arctic silver that I decided to try out on my 5800x3d two years ago. I wasn't impressed with this paste because it's so thick and it's conductive (capacitive?) but I tried it anyway. 2 years later the temperature of my 5800x3d has remained constant and hasn't risen even 1 degree. I'm learning that things aren't always as they seem and how well they work initially with ease of application and initial performance isn't the same thing as how well it works after a years old application. That's the problem as I see it: how do you test a thermal paste to tell how well it will do after long periods of time. Would leaving the cpu off and heating it through an external means such as a heat gun to temperatures that would kill it if you turned it on and then let it cool and then keep doing that over and over and then test it normally with the cpu on to check for pump-out and temperature consistency? I don't know and really wouldn't want to be in your shoes, but I wish you luck and hope your channel does well.

@@pf100andahalf So a couple things. Yeah, Arctic Silver 5 is an odd ball. It was the go-to XOC (extreme overclockers) choice many years ago. It's not silicone based, it's some sort of synthetic oil blend and in their manual, they denote it has a break in period. I have tested this and can synthetically get it to break in around 30-60 minutes of rapid heat cycling up to 110c and back down to 40c every 30 seconds or so. For your other part I have tested this on my previous personal channel. I took another RX 5700 XT (because they are so brutal in heat density, 300watts on 251mm sq die) and wrote just a little script to heat cycle 2 minutes on, 1 minute off for 3000 cycles, then I would check performance on a 10-minute load test. I was specifically looking for evidence of "pump out" which I could not replicate. But in that testing I did not see any indication of paste drying out or wearing down. The only real way to test is to take an ambient temp reading, run a benchmark, stress test, with X drivers on Y operating system patch level and then check it after so many hours/days/months/years of use. Some pastes do have a shelf life due to separation of ingredients and such. If the paste turns to powder it can become mechanically unstable and break the bond between the die and cooler. Long term testing would be beneficial only if the stakes were high. IE if not repasting some HPC node cost a company 10 million dollars, but it never happens. Pastes are fairly cheap, even the good ones. I would consider a yearly re-paste and re-putty of components just part of PC maintenance. Other solutions like kryosheet and phase change materials are also much longer lasting as more maintenance free.

Welcome aboard. I am waiting on about 20 other thermal pastes and exotics which are in shipping direct from China. As soon as I get those in stock we should be able to add to the data. I also have some components that just came in which I will be using to test thermal resistance of the thermal interfaces because as you can see from my data the W/mK ratings of pastes are non standardized and not even comparable within a manufacture's products. The end goal is to extract the meaningful data and have this channel as a living document of independently validated and tested thermal interfaces.

@@The_Thermal_Channel ill be very interested to see this. im also interested in the ptm7950 thermal pads that melt at 45deg

Yeah some game engines are just optimized for Radeon like the COD series. I have had a number of 7900 xtx's and a couple 4090's and yeah the 4090 is better in just about every way possible except the price. But this card has no direct competitor for what you get in a 2 slot package with a blower fan, 24gb. It handles AI/ML workloads super well even in windows, but especially in linux. If this card doesnt run into any issues I will probably pick up a second one for running larger models with PyTorch. I do wish the video encoding was at least as fast as my arc cards though.

Yes I tested PTM 7950 in sheet form, I also have Honeywell 7950-SP and 7958-SP which are paste forms of their phase change material on hand, just havent gotten it into the test bench yet. If you are asking what PTM is, its a phase change material from honeywell. This means a solid at low temp and a liquid at high temp, it retains coverage very well, performs very well and according to internet lore it lasts for years.

It performs well across all tested loads and yes those loads are significantly higher than the IHS of a R7 7700 can produce in terms of wattage density. The main question I would have is, the cooler you are using, is it nickel plated (shiny and silver)? Or, is it exposed copper/aluminum (dull silver)? If its copper or aluminum, I would skip this paste. If you have a nickel plated low profile cooler then you should be good to go and you will probably see your package temps only slightly lower on average but the hotspot or max per core temps come down as this stuff seems to handle transient spikes way better than most pastes. If I wasn't running a Kraken 280mm AIO that has an unplated copper base I would definitely use this on my CPU without question. I also run an ASRock Creator 7900 XTX GPU which is an uncommon card and when I finish my review/teardown this is the paste I will be utilizing unless its vapor chamber is unplated copper.

Well, the description of Upsiren doesn’t mention anything about corrosion with copper and aluminum, so I’ll go with PTM7950, although I’m concerned about the blow out.

I can say this is all of a matter of choice but i will put that even on 3w chip 😂

@@justrandom4067 I was expecting kold-01 to dominate but it really depends on all the variables. I really didnt expect nth1 to score that well. I am working through building a synthetic test rig and test parameters that will reduce the variables as well. These tests are also at extremely high thermal densitities so its possible thst some of these pastes just cannot absorb and transfer the load fast enough. Keep in mind these results are only valid for X watts at 251mm squared. There isnt anything on the market today that pulls a consistent 300 watts at this small a surface area. More tests will be needed, more videos to come :)

​@The_Thermal_Channel For me I tested 5 different types of thermal paste on an FX 9590 @ 5Ghz pulling approximately 300-350 (9590 doesn't have a sensor to measure power unlike the other fx 8 core cpus) with a 360mm AIO, my results were: 1. Thermal Grizzly Kyronaunt Extreme 2. SYY 157 3. Thermalright TF7 4. Noctua H1 5. HY 510 I wish I remembered the exact temps they got, I know 1st place kept it within the low max temp, same with 2nd, 3rd place was in the high 50s beyond the max temp, 4th and 5th place let the cpu hit the 60s. I also did a test with an elitebook 8740w with an i7 940MX overlooked and set to 120W. HY 510 and TF7 couldn't effectively transfer the heat for the short 15 second test, H1 could but left a lot to be desired, and 157 managed to only let the cpu hit 71C, and TG Kyronaunt Extreme had the CPU at 68C. Of course the laptops cooling couldn't effectively dissipate the heat which is why the test was only 10 seconds after that the heatsink became saturated and it was off to the races.

Yeah, I plan on adding more graphs in the videos since I have all the raw data. I also have a scoring based on value that uses performance and price in a formula to come up with an objective score. Now that I have loaded up all the pastes from my personal horde I am going to work on just 1 paste per video. I started with the combined because I think it's an easier gauge as some pastes are better at filling voids and may be better in hotspot temps while others may have lower thermal resistance but have more materials that thicken the pastes making it less app for deep coverage. On my next video, which will be Upsiren LMTG-100 I will add your suggestion and some other graphs and details sir. Thank you for the feedback I do appreciate it and will integrate and iterate.

New subscriber!

If we go off the data for 90% of this answer and 10% my personal opinion. It definitely goes to the following, which are just so close @ 300 watt loads over 251mm sq: I would say phase change is the best, PTM7950 which is pretty inexpensive from aliexpress. However I could see for newer PC builders/modders it could be a real pain to get applied, a paste might be better which I would go with Kryonaut because it spreads easily and performs well. Most people will buy a single tube, use it once per pc so the price doesn't matter. The data doesn't have a Kryosheet tested yet but I have used them many times personally and those perform extremely well and while expensive you just buy it once and despite what has been said I have remounted one 7 or 8 times before it got to deformed to use. I am wrapping up the last video of all pastes from my personal supply as well that should release soon. We will have a good starting point to start comparing lesser known products. I also have 4 new cool products to test and about 20 other thermal pastes and products shipping in.

@@The_Thermal_Channel Thanks looking forward to more content! I really appreciate the reply!

@@The_Thermal_Channel I've also found Kryonaut to be the best performer out of all top performing pastes, at least if it's used on a CPU IHS. Noctua NT-H2 is usually within margin of error of Kryonaut, and it's usually cheaper, so NT-H2 is worth considering for IHS use as well. However, unfortunately, Kryonaut does really poorly over long periods of time in high power direct die applications with high mounting pressure due to it's severe tendency to pump out (migrate). If you're repasting a modern and power hungry GPU with tendencies for high and concentrated hot spot temperatures (80C+), and you don't want to be repasting it every 6-10 months, you should avoid both Kryonaut and NT-H2, as NT-H2 also has the same severe migration issue as Kryonaut. My pick for best longevity to performance in direct die applications is PTM7950, you just cannot beat it without resorting to conductive solutions (liquid metal). For standard paste? Noctua NT-H1 has the best longevity to performance ratio, but expect performance to be noticeably behind PTM7950.

@@The_Thermal_Channel there is ptm in paste form. You should test that one out.