I think you’ll find in life that even the most intelligent folks make mistakes. This man is brilliant. He hurt himself trying to teach folks…and turned his own error into a teaching moment for others.
Lets not look a gift horse in the mouth.
Nah. Its “entertainment” as much as “education”. Gotta give him the benefit of the doubt IMO
B
BAZINGA
Radiative is a form of heat transfer, but at these temperatures it is useless.
Why? well radiation goes up by the power of 4, so in high temperatures it has a good effect, but lower temperatures it doesn’t.
When you hold you hand up to a fire and feel the heat, that is radiation, but it works because the fire is more than 800 Deg C. The reason it is yellow is because the frequency changes as the temperature changes, and at some point it comes into our visual spectrum. The same happens with steel that gets red hot.
Radiation can only really be manipulated by different emissive and reflective properties. This means that some materials reflect more and others absorb more. Aluminum reflects Infrared for example.
But even then, radiation only works when the enviroment doesn’t radate it back or is a lower temperature.
In this case a thermal compound that radiates out, will also receive radiation from a heat sink, that is only slightly cooler, so the effect is reduced to a couple of degrees cooling.
Even if you miraculously don’t get any radiation back. the radiation of a cpu at 100C is… Drumroll: about 1W
its a gimmick.
they are pretty much calling there paste quantum thermal paste…
and hoping a physicist doesn’t pull there ad apart.
Lol I had to know. It seemed too fun to pass up!
But now there is MX-6…more numbers more better.
Thanks for doing the testing. MX-4 has been my go to for several years. It’s cheap and I keep a couple of tubes sitting around “just in case.”
If you watched the videos you’d know this is not the case at all. They set up experiments as demonstration and explain the science behind them. They aren’t claiming to have made any new discoveries.
Have you had a chance to test Honeywell PTM7950 TIM? I started buying this about a year ago from a Chinese reseller after seeing it suggested on another forum but you can now buy it on amazon and I can confirm its genuine PTM7950 as I purchased from this seller a few months back when installing my LGA-1700 Plate.
A tip to anyone who hasn’t worked with PTM7950 is to throw it into the freezer at least an hour before you plan to apply it. Once you pull it out you need to work fast to cut and apply it because its only a matter of minutes before it comes back to room temp and goes soft making it harder to work with.
Hi @Susanna. I’ve been interested in thermal pads. Not because I find the goo icky, but I can see why one would, it’s more because every second time I apply thermal paste somewhere I make a mess. Do you have any recommendation on thermal/graphite pads or is one like the other?
I’m also interested in these as well. The PTM7950 gives great thermals but like thermal paste is still messy to clean up. My hope with the PTM7950 was that it would return to a “solid” pad at room temp making it easier to scrape off but that hasn’t been the case in my experience.
I haven’t even heard of that xD might add it to the bucket list
I haven’t used it on CPUs with heat spreaders but it works really well on bare dies like GPUs.
Where it really shines is the ability to not pump out like thermal paste does on bare die applications. I have it on my 3080 for close to 6 months (and counting) and the GPU core hot spot delta remained at a constant 10-11 deg Celsius under load.
Only the stock paste and Honeywell PTM7950 managed this. I tried Thermalright TFX, Artic MX6 and Noctua NT-H1, none lasted more than 2 months before they all eventually pump out.
Yeah, I may have abused that little flag a bit when I had a couple of EXACTLY the same notification in the unread threads…
The Lounge had a small talk about it where it was concluded to be bogus at best, which as shown above, it is.
Different products for different applications. Aftermarket and enthusiast pastes in general are much thinner than the OEM variety so they can spread as thinly as possible and not require any “curing”. This of course can cause the aforementioned “pump-out effect” and require more frequent applications, but it should be noted that pump-out usually only occurs at higher temperatures or from hot spots.
It would make sense that a bare die wound pump-out as the heat is much more concentrated than when spread through an IHS. Your CPU die may be 90C, but the IHS definitely is not. Also I’m pretty sure phase change TiM is the new standard for factory GPU production.
MX-5 was discontinued due to both performance and quality control issues, that’s why MX-6 came out so quickly and why MX-4 is still the bulk standard.
Yeah I agree. When the less viscous pastes (MX6, NT-H1) pumped out, I totally understand. I wasn’t expecting the same for the more viscous TFX since so many people claimed it worked for bare dies.
At least on nvidia FE cards.
Not sure what evga used on my 3080 ftw3 ultra, but it worked really well until I disassembled my card… Could be some sort of phase change thermal interface material.
I suspect they were using much less power hungry silicon than a 3080 FTW3.
I just sent a 1660 non-super out the door with standard Kryonaut on the die, if it comes back in less than a year I’ll use a phase change material on it.
Perhaps it’s more his thumbnails rubbing me the wrong way than his content
It definitely makes sense to use the phase change TIMs on bare dies because it’s easier to get the higher mounting pressure they require to conduct heat well.
PTM7950 needs ~40psi of pressure to perform best, by my calculations you’d need something like 70 pounds of mounting force to get it to perform best on LGA-1700. I’m unsure of how much mounting force sockets normally produce.
I’ve been using Thermal Grizzly KryoSheet, because it’s I can get with next-day delivery. It works fine. Really fragile though, so use a pintset to gently place it on the processor. If it slips out of place, which it probably will, gently push it back in place. Handle it as little as possible basically. Once it’s in place you’ll want to push the cooler on straight down so it doesn’t move, and then screw it in place the same way you would with thermal paste, going between screws diagonally and evenly. Just like the paste it will deform to give good coverage, which is why it’s so fragile and why you want even pressure during the install. It’s electrically conductive so you should also either just buy the right size directly, or cut it to size yourself, so it won’t flop down and short anything.
Thermal pads != Graphite pads. Graphite pads are good for processors and other things that get really very hot really very quickly, they conduct heat as well as paste. Thermal pads are for things that produce much less heat, such as VRMs, SSDs, or memory. They’re thick and squishy, but nowhere near as good at moving heat.
Meh. If I never watched videos with click-baity thumbnails I’d be watching a heck off alot less YouTube videos.
I feel like his thumbnails were less clickbaity a few years back but he probably tried switching to more clickbait style because the views weren’t getting there.
Thanks for taking the time to evaluate our product! We’ve been eagerly awaiting this type of feedback and evaluation as it can dramatically influence our development in the near term.
Let me first assure folks on this thread that radiative cooling is not a gimmick - it is a very new and exciting technology under active development. The Wx-130 product represents a first attempt at commercialization of the technology (specifically, enhanced near-field thermal radiation) for an application we are passionate about - high performance computing. Bare with us as this is our first product release, and while we are thrilled to see it competing with some of the best pastes on the market, our aim is to exceed the capabilities of the best TIMs on the most power-dense CPUs/GPUs available and by much larger margins.
Next, let me point out an important issue raised here in testing thermal compounds. What is the figure of merit in this experiment? If it is a single data point for max temperature over ten minutes, I agree with your summary. If it is the average temps over the run, the list reverses and MX4 wins. If it is temperature relative to the power dissipated (e.g. ~thermal resistance), the data looks like this:
KPX: 77/240.3 = 0.3204 (3.7% better than Wx130)
TF7: 77/235.2 = 0.3274 (1.6% better than Wx130)
WX130: 74/222.4 = 0.3327
MX4: 73/206.6 = 0.3533 (-6.2% worse than Wx130)
This isn’t to be critical of the testing results - just trying to add some color to the discussion of a nuanced subject. It is interesting that the Wx-130 does reach a higher max temperature because that isn’t what we’d expect - and considering the temperatures involved and the materials utilized in the paste, there is likely some “burn-in” occurring still at ~80C (which is a good data point for us to have). Overall, your spread / mount and test setup look superb for all tests you ran - so nice job.
I would love to go into more detail on how the paste is “radiative” and why we are utilizing this mechanism of thermal transport as opposed to adding more and more highly thermally conductive fillers (which is how thermal paste has historically been developed) because the underlying physics of how radiation moves heat around is greatly underutilized (and has fascinating implications for the future). The technology to engineer the materials in the ways required to take full advantage of light in these settings has only recently become available (and yes AI + HPC simulation are critical tools for this!), but we are excited to see folks trying this first generation of Wx130 and humbly await more feedback from users like yourselves. I would implore folks to consider the average temperatures (so long as you’re not hitting thermal throttling events) and average power over benchmark runs, as well as the initial conditions (ambient temp, idle temp) from which the benchmark starts in the analysis.
This is all to say, we are sincerely grateful to you and the folks in this community for taking the time to learn about and evaluate our TIM. Rest assured, Wx130 is transporting heat across the interface differently than any other thermal compound does, and there is much more to come. We would love to send you another sample if you’re interested (DM us or reach out via email).
Thanks again for taking the time to evaluate - hopefully we can continue to engage with you all as we continue to develop and improve our products!
I don’t blame you all for trying, but I doubt you’ll be selling anyone here.
