thank you so much for the hard work. i applied ptm7950 from watching your videos. i have waited 3 years for something that could replace liquid meta . My laptop always been on balance mode now i can easily change it to performance mode on my laptop that was using tradition thermal paste. thanks again
@ahmed-mujtaba That's great to hear. I feel like Liquid Metal should be considered an Enthusiast solution and not in so many devices straight from the factory. Considering all the drama around warranties and RMA's, LM just doesn't feel like the right fit. I'm glad Phase Change Materials are on the scene now, and we're seeing nearly every brand bring offerings under their brands. How they all stack up with each other still remains to be seen. It will be interesting to see if any of them can beat Honeywell PTM7950. Matching is okay too, but only when their is a cost savings.
This videos tests are now obsolete, but it's still a good rough idea. Now I'm using Testbench 2.0 and log ambient intake temperatures in real time using HWINFO64 and an AquaComputer Quadro and a few sensors. I test for 2 hours and use data from the 2nd full hour and enter that data into the chart, located at the Google Drive link.
TY Snark for the extensive testing. Have you ever taken a look at the copper shims offered by coolmygpu? The one thing I like about their products is that they wrap around the bracket for the GPU die (with some fairly liberal tolerances) which means that once set in place they can only slide so far in any direction before catching on the bracket. I think for a lot of people this eases worries about the shims sliding around due to inadequate mounting pressure. Another interesting thing I noticed while looking for reviews of their products is that some users are getting good results sandwiching the shims with paste instead of putty. I have the impression that you prefer putty to paste for shims but have you ever actually tested the paste + shim combo in your test bench? I think it would be interesting to see how MX-4 or Kryonaut compare in performance and especially the impact on core temps since paste and putty will compress differently. Thanks again for these great videos, I've been getting a lot of inspiration here for my next build.
Thanks for checking out my videos. I have seen those Coolmygpu copper plates. It seems like a great product to me as long as the thickness is less than 80% of gap size. If you were wanting to do work on 1 GPU and cost is no issue, and you typically avoid DIY type stuff I think it's a good way to go. Do loose shims safely takes more effort than popping in a cool my gpu plate. If I had access to a water jet I would probably cut my own for myself. I have not tested a paste and shim combo yet, so everything I say about it comes from things I've read so far. Thermal Paste, being more less viscous than putty can pump out easier during hot/cold cycles, dry and crack, and not grip a shim as firmly. It's also not designed to be a gap filler, but rather an imperfection filler for direct metal to metal contact. Realistically a shim would need to be ~99% of gap (not pad size) for paste to be long lasting and effective, but this could compromise core contact if the measurement is off at all. I do plan to test this out on a card at some point to check performance so we can directly compare it to a shim+putty combo.
What's interesting is the copper and th949-1. i'm surprised the core temp was higher than the rest while the vram was a lot cooler than most. Great video! I only found out about TG-PP10 a few weeks ago as I was researching on how to cool down a 3090 I picked up. For my CPU the stock AMD 3600 CPU cooler was heating up to low 90s. I bought SYY thermal paste and upgraded the cooler to a scythe fuma 2. Unfortunately I didn't do a test on each. Overall the cooling dropped down to maintaining 70 +/- 3 degs. I will be trying the SYY thermal paste on the GPU at a later time. Now that you've done some testing on other putty, I might go for one of the new ones you mentioned.
As the VRAM is allowed to dump more of it's heat into the cooler, the Core will go up. In fact the coolest temp on the core was achieved with the stock pads being used, where the core didn't have to share nearly as much of the cooling capacity. It it's like a 4 lane highway, the core was hogging all the lanes and forcing the VRAM to drive on the shoulder with stock pads. Now it's giving up one of the lanes to the VRAM. Tha analogy isn't perfect because each GDDR6X chip only generates around 3.4 watts, so that's around 27 watts combined. Makes the VRAM like a motorcycle in size for the highway analogy.
@@mankybrains I should note that some of the core temps during a couple of the tests were higher due to the pads being fairly stiff, which can compromise the contact between the cold plate and the core. Take the Gelid Ultimate pads as an example. 53 C on the core and 86 on the memory, then we have the core temps going down to 51-52C with softer pads/putty which still achieve cooler memory temps. We don't get back up to 53 C on the core until we get into shim territory with Memory temps of 72 C. To properly use Gelid Ultimate I believe they need to be rolled with a rolling pin (between 2 sheets of parchment paper or similar), until they are close to the correct thickness, if just slightly thicker so they can get their final compression during install, for best contact. This is why most high end pads are hard to get a perfect install with, especially when you use them on 3+ different components with sometimes 4-5 different thicknesses. I guess that's why I like Thermal Putty. It makes it easy to replace all the pads without stressing over pad thicknesses. It also works great as a supportive product if you want to put good pads on your VRAM and putty everywhete else, and is the perfect match for shims. The Upsiren 3D graphite pads also do well with being easily compressed. If anyone's does try the Upsiren pads, just be sure you have everything lined up perfectly and use a quality paste. Ideally you don't want to take the card back apart until it needs a repaste.
Kudos for your efforts testing all those pads and shims! I found out about your channel from one of your Reddit posts ragarding "ASUS TUF 3080 Thermal Pads Gelid Extreme". I was thinking about changing the pads and thermal paste on a 2y old Asus TUF 3090. I'd rather go with the Gelid GP Extreme pads, do your pads measurements from your Reddit post still apply? Thanks in advance!
Glad you made it here from Reddit. That was where I initially posted stuff prior to making a UA-cam channel. I haven't worked on a TUF 3090 yet. I've done a Strix 3090. This guide is probably worth a read: www.reddit.com/r/nvidia/comments/qxwlv3/asus_tuf_3090_vram_thermal_pad_thickness_findings/?
I find it's not too bad if I scoop up most of it with the tiny paste spatula, then either Q-tips and Isopropyl, or sometimes a dry soft bristle toothbrush can work really well for crevices and valleys. I always finish cleaning with Q-tips and IPA. The Kapton tape can definitely make some of the cleanup easier.
@@snarksdomain i scoop and reuse my thermal putty, but because of all the small SMD's, i had to use isopropyl and a tooth brush to clean it, and it was not nice; took the better half of a day (to be honest, this was my first time using thermal putty and smear a butt load on the card's vrm and vram)..
5:00 its better to use double sided thermal glue tape for this :) Super easy to use and keeps the shims in place. the tape is super thin and conducts heat really well to the shim. if u want to remove, make sure to heat the card properly. otherwise u could damage parts, if the glue tape is cold and hard.
You can. I still prefer putty on both sodes of the shims myself. My 3090 had Copper Shims and TG-PP10 in it for 2.5 years and the putty was still soft when I took it apart.
Thanks for the info It was a great video I'm thinking of changing my GPU 3070ti pads but I haven found the correct thickness so I believe putty is the best solution
My pleasure. Be sure to check out the latest charts with a lot more putties tested, and some pads. My current testing method is more accurate than what I was doing for this video. And I'm planning to further improve the accuracy soon, and retest everything again :) Link is in the description (Test Chart Repository)
Nice video, lots of testing must have taken a lot of your time, thank you for this. I have a question, why not use copper shims as thick as the block tells you that you need? so all you'd need is a skim coating of puddy? btw, I somehow never heard of using shims even though I've been water cooling with custom loops for close to two decades. I am sure I was dating a girl during this & my PC was collecting dust & loop was growing mold & algae.
Thanks for checking out my videos. The gap between VRAM and cold plate is 1.8mm. Putty won't squish down like a paste would, but has the benefit of not pumping out or baking like a paste would. Using 1.6mm shims and 0.1mm on putty either side means there's a bit of wiggle room for the tolerances. 0.1mm is also about as thin as you can use putty effectively. A lot of putties have minimum bond line thicknesses of 0.12mm so I'm already pushing it with 0.1mm each side.
Can guess you are using kapton tape since this can be removed later (non permanent solution). For something more permanent I would recommend using liquid Electrical Tape, you can apply with a syringe and the layer is much more finner. Problem is that you can remove it but at the risk of taking off some SMDs. People usually use LET for masking video cards to use with LN2 POTs. Regards
@@snarksdomain LET is a plastic/rubber solution meant for similar use as as black electric tape but in a liquid form. Is dielectric, UV resistant and waterproof. Some electronic parts come with it and is called as "conformal coating". Regards
Thanks Joey! I plan to add in gaming benchmarks using some of these shim/pad/putty solutions in the future. ....Thanks for the extra work @GhostMotley 😉 lol.
Thanks for the great info !! :) But i have to say that I'm personally a bit ""disappointed"" in how badly it seems that some manufacturers are at cooling their products. I mean they "work" as delivered but they most certainly are not "built to last" considering that "High temps" are a sure fire way to if not "out right kill" at least speed the aging of a chip... So with that in mind the GDDR6 and GDDR6X setups of many cards seems to risk making a whole lot of cards die "prematurely". Seeing how most "normal users" likely are not even aware if their cards are throttling due to "risky temps"... Which of course suits the manufacturers, but it really stinks for everyone else.. Best regards.
It is disappointing for sure. It's also possible that the manufacturers weren't fully prepared to deal with GDDR6X VRAM on the 30 series, the worst culprit being the 3090 and it's 24 memory chips. It will be interesting to see how the 40 series cards are doing in this regard. GhostMotley mentioned that his MSI 4090 has very respectable memory Temps while gaming. So it's quite likely some improvements have been made. Whether it's improved pad materials, smaller gaps, hardware tweaks, or even lower memory voltages(?) remains to be established.
@@snarksdomain Yea it could be that the "OEM's" were "caught of guard"... in which case they should really be "annoyed" with the "GreedyGreenGiant" (aka Nvidia) for not giving them good enough info ahead of time... Then again it is also "kind of their own job" to make sure that their own cards "work right". And not run on the VERY Ragged edge, like many seem to do... So Yea I really hope that the 40-series are "better engineered" when it comes to this aspect of things. Being a "normal enthusiast user" I usually upgraded every generation, but I stopped with that after the 10-series due to "not liking" the "overhype" of the 20-series launch. And then came the "Scalpocalypse"... I considered getting a used "30-series" but with their "weaknesses" I didn't feel comfortable doing that. So I'm "holding high hopes" for the "40's", my 1080 is getting really "long in the tooth" by now :) ... Best regards.
Your videos have inspired me to try my hand at shimming my 3090ti. I used 0.8mm shims (stock ones for my waterblock were 1.0mm) and have gotten amazing, but mixes results. My memory temps are running at around 42c under load (ambient around 70f) but my hot spot temps have jumped almost 15c (106c under load). Gpu temps stayed pretty much the same though i went from liquid metal to thermal grizzly hydronaut paste on the die. Do you think this could be a result of reduced contact between the block and the gpu? Any thoughts would be helpful.
Did you use putty in conjunction with the shims. If stock pad thickness was 1mm then 0.8mm is definitely the max you'd want to go. If your Core temps are still the same you should be good, but it might be worthwhile going with Honeywell PTM7950 (sheet) for the core if you ever decide to repaste. I have occasionally seen slightly increased hotspot temps even f4om putty jobs. My guess would be slightly inadequate cooling of VRMs maybe.
Thanks for the reply. Yes all pads were replaced with putty and the shims were installed with putty on either side. I will look into the honeywell pad but I may just liquid metal the core again. Keep up the great work sir. I will update this post with any relevant findings.
Putty should last you a very long time. Some folks misinterpret the "use before" dates as an expiry date. In reality, these dates represent how long one can expect the viscosity to stay the same. This is more important for automated applications as it will be programmed for a specific air pressure, nozzle size, and "X" amount of time to get the perfect application. If the viscosity increases then the flow rate will be lower and not enough putty will be applied. This would require a tweak to the programming until a fresher batch of putty could be sourced. For manual applications it isn't nearly as critical. The performance should be fairly consistent for years, perhaps even the life of the device. Items you will find Thermal Putty in: -Sony PS5 -Modern EVGA GPU's -Modern Lenovo gaming laptops -Some AMD 7000 series GPU's -Some mobile phones
@@snarksdomain Thanks! I think I'll go with the U6 Pro in my laptop. I saw you applied to the cpu core in other video I wouldn't mind having 4 or 5 C higher than thermal paste as long as I don't have to be replacing the putty frequently
@@elvioguerrero5914 U6 is a solid choice. I would recommend considering a Phase Change Material for the core. Either Honeywell PTM7950 sheet or Upsiren PCM-1
@@snarksdomain Got it, I'll try both Upsiren solutions as soon as I can, i got a Dell G5 and this is difficult to keep cool let's see how it performs, I can use the putty on the pipes to make sure they make contact with the copper side of the case right? that should also help with cooling
Hey! sorry for the late update but I applied the U6 Pro and the PCM-1 to my laptop and the temps are night and day, is like 10-20 C degrees difference!, I just want to know when to change the PCM-1?
They are 1.60mm thick. The gap is 1.8mm, this leaves 0.1mm putty on either side which basically ma5ch3s uo with the minimum bondline thickness of most putties. You can check out some of my other shim videos where I list the shim thickness.
I see people using shims with thermal paste! Isn’t it better transferring heat than Putty? I’m waiting to get my shim and my putty from ali express, but I think I will go with the Thermal Paste.
Hard to say for sure. I've still got an RTX 3080 with TG-PP10 on the core and it was marginally warmer (2-3 degrees). While TG-PP10 is an okay putty, it is nowhere near the performance level of UX Pro or many other putties. In general, I don't think pastes automatically perform better than putty. In my opinion, pastes are limited in how much heat conducting additives they can use before they max out their usable viscosity. A putty can load up with higher density and likely achieve equal or better thermal transfer. I should probably try using some UX pro on a core and see what the temps are compared to MX-4, just to get a half decent comparison.
@@snarksdomain that could work!! I recently bought a second hand RTX 3080 TUF I am the 4th owner. The middle fan makes noises and the GPU is clean! After dissembling it and noticing oh the first owner might gave a bath to the GPU to clean it, because the heat sink is Corrostted(I don’t know if this is a word, simply corrosion) as hell so I gave it a bath in vinegar and I ruined the nickel plating didn’t know nickel get eaten by acids. So I tried to fix everything and gave it multiple baths in distilled water every time I take it out to oil it and protects it it get rusted and I am confused later on I said fu** it if it rust at least it is better and protected. I realized the corrosion is made of salt and salt was filling the distilled water and now it is regular tab water hahaha why I didn’t think of that! So I gave it one more bath and no longer rusting I tried to clean it the best I can as I don’t have a sonic cleaner. Now I ordered a one piece memory Shim and UX putty just in case I was going to use it, but after what you said it could make sense to use the putty. If you need any help or check my results I can start making an excel sheet for you Delta tamp and my thermal paste is GD900 I believe I don’t know if it help corrosions between the shims and the heatsink. I no longer care if anything happened to my bad looking heat sink I would switch to a water block
@@cheradaninezakalwe2851 That's likely not a good idea. One thing you need to have with shims is a method of keeping them in place, partially to prevent the shims shorting something out on the card, but also to ensure no VRAM goes without adequate cooling. There are 2 common methods of keeping the shims in place. One is by using a fairly viscous substance such as paste or putty (I prefer Putty). The other is to have compressive force on the shim using the mounting pressure of the cooler to do so. This 2nd method places too much force on the VRAM and also means it would be less likely to have adequate core contact and mounting force applied to the core. If the shims were undersized with Liquid Metal they would easier shift out of place, especially if vertically mounted. The current best combination with shims is Thermal Putty and an undersized shim (ideally 0.3mm or more of the gap being filled with putty, 0.15mm on each side of the shim).
I want to replace the pad/putty on my PS5 problem is that i dont even know if they was originaly pads OR drop of putty ... what you would recommend and where to buy the good qty without spending too much for not used stock/remain ?
I'd love to test things out on a PS5 at some point. I believe they used a Thermal Putty. For the core they use Liquid Metal. I would probably try using PTM instead and see how it does in comparison.
I won't know that for some time. It does seem to have a very similar consistency to TG-PP10 and even TH930. I ordered a bunch of the tiny containers of Jeyi and plan to load up an Icing Piping Bag of it to use on GPU's.
@@svenvanderzwaag1012 I have not. So far none of the putties I've used leak oil like pads do. I find pads like Gelid Ultimates often start leaking after a couple months. My reasoning is that putties are already better than pads for longevity so I'm full speed ahead on using putties instead of pads.
At that point it would make more sense to just use shims I think. With tape you would still need a thermal interface material on either tside of that. Some folks might use thermal paste, but I would opt for a putty in that scenario for the longevity. That would make quite the putty-tape-pad-tape-putty sandwich. If using 3D Graphite pads I think the best thing to do would be to use a Phase Change Material on the core itself (Honeywell PTM7950 or Upsiren PCM-1). That way you wouldn't need to repaste a couple years later and could just keep it assembled until there were any thermal issues.
@@snarksdomain "That way you wouldn't need to repaste a couple years later and could just keep it assembled until there were any thermal issues" That us why I ordered PTM7950SP to be able to apply it and not to have to buy 2 sheets for 2 CPU'S like E5-2697v2. On GPU vram I still think it is wise to use 3D Graphite pads then a piece of aluminum foil on top then artic silver 5 that never seems to harden. That way you can replace GPU thermal paste and other thermal interface without braking graphite pads. The Graphite pads brakes because they are coated with PTM7950SP that glues them dead to heatsink. It's that simple.
@juliusvalentinas definitely one way to go. I do like PTM7950. I still lean towards using UX Pro on VRAM, VRM, etc, and PTM7950 on the core. Or for shims I'd likely currently do $5 of aluminum shims and U6 Pro. I'm hoping PTM7958 sheet becomes available soon. I didn't have much luck with the syringe style application.
That is odd. Did you get a confirmation email for the order or any cancellation update from the seller? If there is still an issue let me know and I will contact the seller.
@@snarksdomain didnt get anything but just pop in to my aliexpress account i can see the items in my basket and they have red writing saying the seller can not send items to this address. And seen £101 payment at 5th august from 司順貿易有限公司 this chinese name, just open a case on it, as i dont recotnise this company. Anyway surely if its paid and i didnt get the item i willl get my money back but i am kind of 90% sure i ordered and paid with paypal thank god !
I will be retesting this all soon with better logging of ambient temperature. So the verdict is still out, in my opinion. But I believe it is quite possible that in this scenario, the weak links are the thermal conductivity of the putty and the Silicon. Since Silicon is around 90-160 w/mk, it isn't nearly as fast at moving heat as Aluminium is. That's my best guess as to why I ended up with the same results for both.
@@snarksdomain that’s makes sense didn’t come to my mind! I was saying maybe it was the putty! But then I said ok but the heat already passed through the shims! Slightly confused
I'm not sure how accurate the ratings are, but I'm using some 21w/mk pads in a 3070ti, just on the VRAM with putty everywhere else. So far they seem decent performance. They aren't the softest pads, so it's important to get the sizing right, or one could run into core contact issues. We should have a better idea of their performance once I retest everything.
GDDR6 doesn't run too hot so you'd likely be okay with K5 Pro, although I personally won't use it. For anything with GDDR6X I strongly recommend against using K5 Pro. It boils, pumps out, and off gasses at higher temps.
@@snarksdomain It's so funny. Some people love K5 Pro, others highly recommend not using it. The vendor says it's okay to use up till 250C. Some people on Reddit also said it "boils". Not sure what to make of it! My goal is to have my GPU as maintainance free as possible. Ideally I want to just blow out the dust from the cooler every 1-2 years and not have to change the thermal pads ever. I will use Honeywell 7950 for the GPU itself. What would you recommend as thermal pads?
thank you so much for the hard work. i applied ptm7950 from watching your videos. i have waited 3 years for something that could replace liquid meta . My laptop always been on balance mode now i can easily change it to performance mode on my laptop that was using tradition thermal paste.
thanks again
@ahmed-mujtaba That's great to hear. I feel like Liquid Metal should be considered an Enthusiast solution and not in so many devices straight from the factory. Considering all the drama around warranties and RMA's, LM just doesn't feel like the right fit. I'm glad Phase Change Materials are on the scene now, and we're seeing nearly every brand bring offerings under their brands. How they all stack up with each other still remains to be seen. It will be interesting to see if any of them can beat Honeywell PTM7950. Matching is okay too, but only when their is a cost savings.
Damn excellent videos.
This videos tests are now obsolete, but it's still a good rough idea.
Now I'm using Testbench 2.0 and log ambient intake temperatures in real time using HWINFO64 and an AquaComputer Quadro and a few sensors.
I test for 2 hours and use data from the 2nd full hour and enter that data into the chart, located at the Google Drive link.
TY Snark for the extensive testing. Have you ever taken a look at the copper shims offered by coolmygpu? The one thing I like about their products is that they wrap around the bracket for the GPU die (with some fairly liberal tolerances) which means that once set in place they can only slide so far in any direction before catching on the bracket. I think for a lot of people this eases worries about the shims sliding around due to inadequate mounting pressure.
Another interesting thing I noticed while looking for reviews of their products is that some users are getting good results sandwiching the shims with paste instead of putty. I have the impression that you prefer putty to paste for shims but have you ever actually tested the paste + shim combo in your test bench? I think it would be interesting to see how MX-4 or Kryonaut compare in performance and especially the impact on core temps since paste and putty will compress differently. Thanks again for these great videos, I've been getting a lot of inspiration here for my next build.
Thanks for checking out my videos.
I have seen those Coolmygpu copper plates. It seems like a great product to me as long as the thickness is less than 80% of gap size.
If you were wanting to do work on 1 GPU and cost is no issue, and you typically avoid DIY type stuff I think it's a good way to go. Do loose shims safely takes more effort than popping in a cool my gpu plate. If I had access to a water jet I would probably cut my own for myself.
I have not tested a paste and shim combo yet, so everything I say about it comes from things I've read so far. Thermal Paste, being more less viscous than putty can pump out easier during hot/cold cycles, dry and crack, and not grip a shim as firmly. It's also not designed to be a gap filler, but rather an imperfection filler for direct metal to metal contact.
Realistically a shim would need to be ~99% of gap (not pad size) for paste to be long lasting and effective, but this could compromise core contact if the measurement is off at all.
I do plan to test this out on a card at some point to check performance so we can directly compare it to a shim+putty combo.
What's interesting is the copper and th949-1. i'm surprised the core temp was higher than the rest while the vram was a lot cooler than most. Great video! I only found out about TG-PP10 a few weeks ago as I was researching on how to cool down a 3090 I picked up.
For my CPU the stock AMD 3600 CPU cooler was heating up to low 90s. I bought SYY thermal paste and upgraded the cooler to a scythe fuma 2. Unfortunately I didn't do a test on each. Overall the cooling dropped down to maintaining 70 +/- 3 degs. I will be trying the SYY thermal paste on the GPU at a later time.
Now that you've done some testing on other putty, I might go for one of the new ones you mentioned.
As the VRAM is allowed to dump more of it's heat into the cooler, the Core will go up.
In fact the coolest temp on the core was achieved with the stock pads being used, where the core didn't have to share nearly as much of the cooling capacity.
It it's like a 4 lane highway, the core was hogging all the lanes and forcing the VRAM to drive on the shoulder with stock pads. Now it's giving up one of the lanes to the VRAM.
Tha analogy isn't perfect because each GDDR6X chip only generates around 3.4 watts, so that's around 27 watts combined. Makes the VRAM like a motorcycle in size for the highway analogy.
@@snarksdomain Even with that analogy, I got it. Thanks!
@@mankybrains I should note that some of the core temps during a couple of the tests were higher due to the pads being fairly stiff, which can compromise the contact between the cold plate and the core. Take the Gelid Ultimate pads as an example. 53 C on the core and 86 on the memory, then we have the core temps going down to 51-52C with softer pads/putty which still achieve cooler memory temps.
We don't get back up to 53 C on the core until we get into shim territory with Memory temps of 72 C.
To properly use Gelid Ultimate I believe they need to be rolled with a rolling pin (between 2 sheets of parchment paper or similar), until they are close to the correct thickness, if just slightly thicker so they can get their final compression during install, for best contact. This is why most high end pads are hard to get a perfect install with, especially when you use them on 3+ different components with sometimes 4-5 different thicknesses.
I guess that's why I like Thermal Putty. It makes it easy to replace all the pads without stressing over pad thicknesses. It also works great as a supportive product if you want to put good pads on your VRAM and putty everywhete else, and is the perfect match for shims.
The Upsiren 3D graphite pads also do well with being easily compressed. If anyone's does try the Upsiren pads, just be sure you have everything lined up perfectly and use a quality paste. Ideally you don't want to take the card back apart until it needs a repaste.
@@snarksdomain The compression issue is why I went for softer but lower performance pads when I repasted my own card.
Kudos for your efforts testing all those pads and shims! I found out about your channel from one of your Reddit posts ragarding "ASUS TUF 3080 Thermal Pads Gelid Extreme". I was thinking about changing the pads and thermal paste on a 2y old Asus TUF 3090. I'd rather go with the Gelid GP Extreme pads, do your pads measurements from your Reddit post still apply? Thanks in advance!
Glad you made it here from Reddit. That was where I initially posted stuff prior to making a UA-cam channel.
I haven't worked on a TUF 3090 yet. I've done a Strix 3090. This guide is probably worth a read:
www.reddit.com/r/nvidia/comments/qxwlv3/asus_tuf_3090_vram_thermal_pad_thickness_findings/?
@@snarksdomain Thanks a bunch!
i wish i knew that trick with the kapton tape, I had such a "wonderful" time cleaning thermal putty from my card.
I find it's not too bad if I scoop up most of it with the tiny paste spatula, then either Q-tips and Isopropyl, or sometimes a dry soft bristle toothbrush can work really well for crevices and valleys.
I always finish cleaning with Q-tips and IPA.
The Kapton tape can definitely make some of the cleanup easier.
@@snarksdomain i scoop and reuse my thermal putty, but because of all the small SMD's, i had to use isopropyl and a tooth brush to clean it, and it was not nice; took the better half of a day (to be honest, this was my first time using thermal putty and smear a butt load on the card's vrm and vram)..
Great work! 👏🏻👏🏻
5:00 its better to use double sided thermal glue tape for this :)
Super easy to use and keeps the shims in place.
the tape is super thin and conducts heat really well to the shim.
if u want to remove, make sure to heat the card properly. otherwise u could damage parts, if the glue tape is cold and hard.
You can. I still prefer putty on both sodes of the shims myself.
My 3090 had Copper Shims and TG-PP10 in it for 2.5 years and the putty was still soft when I took it apart.
Thanks for the info It was a great video I'm thinking of changing my GPU 3070ti pads but I haven found the correct thickness so I believe putty is the best solution
My pleasure. Be sure to check out the latest charts with a lot more putties tested, and some pads. My current testing method is more accurate than what I was doing for this video. And I'm planning to further improve the accuracy soon, and retest everything again :)
Link is in the description (Test Chart Repository)
Can i use copper shims and putty on laptop?
Nice video, lots of testing must have taken a lot of your time, thank you for this. I have a question, why not use copper shims as thick as the block tells you that you need? so all you'd need is a skim coating of puddy? btw, I somehow never heard of using shims even though I've been water cooling with custom loops for close to two decades. I am sure I was dating a girl during this & my PC was collecting dust & loop was growing mold & algae.
Thanks for checking out my videos. The gap between VRAM and cold plate is 1.8mm. Putty won't squish down like a paste would, but has the benefit of not pumping out or baking like a paste would.
Using 1.6mm shims and 0.1mm on putty either side means there's a bit of wiggle room for the tolerances. 0.1mm is also about as thin as you can use putty effectively. A lot of putties have minimum bond line thicknesses of 0.12mm so I'm already pushing it with 0.1mm each side.
Can guess you are using kapton tape since this can be removed later (non permanent solution). For something more permanent I would recommend using liquid Electrical Tape, you can apply with a syringe and the layer is much more finner. Problem is that you can remove it but at the risk of taking off some SMDs. People usually use LET for masking video cards to use with LN2 POTs. Regards
Is liquid Electrical tape basically liquid Polymide? Or something different?
@@snarksdomain LET is a plastic/rubber solution meant for similar use as as black electric tape but in a liquid form. Is dielectric, UV resistant and waterproof. Some electronic parts come with it and is called as "conformal coating". Regards
Great work! :)
Thanks Joey!
I plan to add in gaming benchmarks using some of these shim/pad/putty solutions in the future.
....Thanks for the extra work @GhostMotley 😉 lol.
Thanks for the great info !! :)
But i have to say that I'm personally a bit ""disappointed"" in how badly it seems that some manufacturers are at cooling their products. I mean they "work" as delivered but they most certainly are not "built to last" considering that "High temps" are a sure fire way to if not "out right kill" at least speed the aging of a chip...
So with that in mind the GDDR6 and GDDR6X setups of many cards seems to risk making a whole lot of cards die "prematurely".
Seeing how most "normal users" likely are not even aware if their cards are throttling due to "risky temps"...
Which of course suits the manufacturers, but it really stinks for everyone else..
Best regards.
It is disappointing for sure. It's also possible that the manufacturers weren't fully prepared to deal with GDDR6X VRAM on the 30 series, the worst culprit being the 3090 and it's 24 memory chips.
It will be interesting to see how the 40 series cards are doing in this regard. GhostMotley mentioned that his MSI 4090 has very respectable memory Temps while gaming. So it's quite likely some improvements have been made. Whether it's improved pad materials, smaller gaps, hardware tweaks, or even lower memory voltages(?) remains to be established.
@@snarksdomain
Yea it could be that the "OEM's" were "caught of guard"... in which case they should really be "annoyed" with the "GreedyGreenGiant" (aka Nvidia) for not giving them good enough info ahead of time...
Then again it is also "kind of their own job" to make sure that their own cards "work right". And not run on the VERY Ragged edge, like many seem to do...
So Yea I really hope that the 40-series are "better engineered" when it comes to this aspect of things.
Being a "normal enthusiast user" I usually upgraded every generation, but I stopped with that after the 10-series due to "not liking" the "overhype" of the 20-series launch.
And then came the "Scalpocalypse"... I considered getting a used "30-series" but with their "weaknesses" I didn't feel comfortable doing that.
So I'm "holding high hopes" for the "40's", my 1080 is getting really "long in the tooth" by now :) ...
Best regards.
Your videos have inspired me to try my hand at shimming my 3090ti. I used 0.8mm shims (stock ones for my waterblock were 1.0mm) and have gotten amazing, but mixes results. My memory temps are running at around 42c under load (ambient around 70f) but my hot spot temps have jumped almost 15c (106c under load). Gpu temps stayed pretty much the same though i went from liquid metal to thermal grizzly hydronaut paste on the die. Do you think this could be a result of reduced contact between the block and the gpu? Any thoughts would be helpful.
Did you use putty in conjunction with the shims. If stock pad thickness was 1mm then 0.8mm is definitely the max you'd want to go. If your Core temps are still the same you should be good, but it might be worthwhile going with Honeywell PTM7950 (sheet) for the core if you ever decide to repaste.
I have occasionally seen slightly increased hotspot temps even f4om putty jobs. My guess would be slightly inadequate cooling of VRMs maybe.
Thanks for the reply. Yes all pads were replaced with putty and the shims were installed with putty on either side. I will look into the honeywell pad but I may just liquid metal the core again. Keep up the great work sir. I will update this post with any relevant findings.
Hey awesome videos you got! now I want to put putty on my laptop, but for how long can I use the putty before needing to be replaced??
Putty should last you a very long time. Some folks misinterpret the "use before" dates as an expiry date. In reality, these dates represent how long one can expect the viscosity to stay the same. This is more important for automated applications as it will be programmed for a specific air pressure, nozzle size, and "X" amount of time to get the perfect application. If the viscosity increases then the flow rate will be lower and not enough putty will be applied. This would require a tweak to the programming until a fresher batch of putty could be sourced. For manual applications it isn't nearly as critical.
The performance should be fairly consistent for years, perhaps even the life of the device.
Items you will find Thermal Putty in:
-Sony PS5
-Modern EVGA GPU's
-Modern Lenovo gaming laptops
-Some AMD 7000 series GPU's
-Some mobile phones
@@snarksdomain Thanks! I think I'll go with the U6 Pro in my laptop. I saw you applied to the cpu core in other video I wouldn't mind having 4 or 5 C higher than thermal paste as long as I don't have to be replacing the putty frequently
@@elvioguerrero5914 U6 is a solid choice.
I would recommend considering a Phase Change Material for the core. Either Honeywell PTM7950 sheet or Upsiren PCM-1
@@snarksdomain Got it, I'll try both Upsiren solutions as soon as I can, i got a Dell G5 and this is difficult to keep cool let's see how it performs, I can use the putty on the pipes to make sure they make contact with the copper side of the case right? that should also help with cooling
Hey! sorry for the late update but I applied the U6 Pro and the PCM-1 to my laptop and the temps are night and day, is like 10-20 C degrees difference!, I just want to know when to change the PCM-1?
how come people dont use thin thermal pads with shims. is it not very effective?
Difficult to find thermal pads that measure 0.1mm
@@snarksdomain i thougt instead of using 2mm pads they would use .5 mm on each side and a 1mm shim in the middle (2mm thick in total)
Solder the copper shims to the cooling block first. This way you only need the putty one sided and the shims won´t move by accident.
That'd be a cool mod.
any idea how thick the shims you used were? maybe there would be a better result using a softer putty and 1mm shims.
They are 1.60mm thick. The gap is 1.8mm, this leaves 0.1mm putty on either side which basically ma5ch3s uo with the minimum bondline thickness of most putties.
You can check out some of my other shim videos where I list the shim thickness.
I see people using shims with thermal paste! Isn’t it better transferring heat than Putty? I’m waiting to get my shim and my putty from ali express, but I think I will go with the Thermal Paste.
Hard to say for sure. I've still got an RTX 3080 with TG-PP10 on the core and it was marginally warmer (2-3 degrees). While TG-PP10 is an okay putty, it is nowhere near the performance level of UX Pro or many other putties. In general, I don't think pastes automatically perform better than putty.
In my opinion, pastes are limited in how much heat conducting additives they can use before they max out their usable viscosity. A putty can load up with higher density and likely achieve equal or better thermal transfer. I should probably try using some UX pro on a core and see what the temps are compared to MX-4, just to get a half decent comparison.
@@snarksdomain that could work!!
I recently bought a second hand RTX 3080 TUF I am the 4th owner. The middle fan makes noises and the GPU is clean! After dissembling it and noticing oh the first owner might gave a bath to the GPU to clean it, because the heat sink is Corrostted(I don’t know if this is a word, simply corrosion) as hell so I gave it a bath in vinegar and I ruined the nickel plating didn’t know nickel get eaten by acids. So I tried to fix everything and gave it multiple baths in distilled water every time I take it out to oil it and protects it it get rusted and I am confused later on I said fu** it if it rust at least it is better and protected. I realized the corrosion is made of salt and salt was filling the distilled water and now it is regular tab water hahaha why I didn’t think of that! So I gave it one more bath and no longer rusting I tried to clean it the best I can as I don’t have a sonic cleaner.
Now I ordered a one piece memory Shim and UX putty just in case I was going to use it, but after what you said it could make sense to use the putty.
If you need any help or check my results I can start making an excel sheet for you Delta tamp and my thermal paste is GD900 I believe I don’t know if it help corrosions between the shims and the heatsink. I no longer care if anything happened to my bad looking heat sink I would switch to a water block
@@snarksdomain what about copper shims with liquid metal on either side?
@@cheradaninezakalwe2851 That's likely not a good idea. One thing you need to have with shims is a method of keeping them in place, partially to prevent the shims shorting something out on the card, but also to ensure no VRAM goes without adequate cooling.
There are 2 common methods of keeping the shims in place. One is by using a fairly viscous substance such as paste or putty (I prefer Putty). The other is to have compressive force on the shim using the mounting pressure of the cooler to do so. This 2nd method places too much force on the VRAM and also means it would be less likely to have adequate core contact and mounting force applied to the core.
If the shims were undersized with Liquid Metal they would easier shift out of place, especially if vertically mounted.
The current best combination with shims is Thermal Putty and an undersized shim (ideally 0.3mm or more of the gap being filled with putty, 0.15mm on each side of the shim).
I want to replace the pad/putty on my PS5
problem is that i dont even know if they was originaly pads OR drop of putty ...
what you would recommend and where to buy the good qty without spending too much for not used stock/remain ?
I'd love to test things out on a PS5 at some point. I believe they used a Thermal Putty. For the core they use Liquid Metal. I would probably try using PTM instead and see how it does in comparison.
I think is better to use thermal paste between vram and shims, and putty between shims and cooler.
Reddit brought me here. I would love to see a recommended putty to replace the one that comes on the ps5. Thanks
I will see if I can borrow a PS5.
52C on core and 64C on Memory with 100% fan speed with Copper shims + TH949-1
Awesome man i need to get my temps down on my 3090. The gelid pads giving up on me after almost 9 months
Im using k5 pro putty on my 6800xt and it works wonders
Do you know if jeyi 8100 dries out faster compared to for example tg-pp-10? Does it have to be replaced more often?
I won't know that for some time. It does seem to have a very similar consistency to TG-PP10 and even TH930.
I ordered a bunch of the tiny containers of Jeyi and plan to load up an Icing Piping Bag of it to use on GPU's.
@@snarksdomain ah okey! I am very curious for the results. Have you experienced any oil leaking until now with jeyi 8100?
@@svenvanderzwaag1012 I have not. So far none of the putties I've used leak oil like pads do.
I find pads like Gelid Ultimates often start leaking after a couple months.
My reasoning is that putties are already better than pads for longevity so I'm full speed ahead on using putties instead of pads.
Why not use aluminum or copper tape glued to upsiren 3d graphite pads to make them not fall apart when disasembling?
At that point it would make more sense to just use shims I think.
With tape you would still need a thermal interface material on either tside of that. Some folks might use thermal paste, but I would opt for a putty in that scenario for the longevity. That would make quite the putty-tape-pad-tape-putty sandwich.
If using 3D Graphite pads I think the best thing to do would be to use a Phase Change Material on the core itself (Honeywell PTM7950 or Upsiren PCM-1). That way you wouldn't need to repaste a couple years later and could just keep it assembled until there were any thermal issues.
@@snarksdomain "That way you wouldn't need to repaste a couple years later and could just keep it assembled until there were any thermal issues" That us why I ordered PTM7950SP to be able to apply it and not to have to buy 2 sheets for 2 CPU'S like E5-2697v2. On GPU vram I still think it is wise to use 3D Graphite pads then a piece of aluminum foil on top then artic silver 5 that never seems to harden. That way you can replace GPU thermal paste and other thermal interface without braking graphite pads. The Graphite pads brakes because they are coated with PTM7950SP that glues them dead to heatsink. It's that simple.
@juliusvalentinas definitely one way to go. I do like PTM7950.
I still lean towards using UX Pro on VRAM, VRM, etc, and PTM7950 on the core.
Or for shims I'd likely currently do $5 of aluminum shims and U6 Pro.
I'm hoping PTM7958 sheet becomes available soon. I didn't have much luck with the syringe style application.
@@snarksdomain "I didn't have much luck with the syringe style application." - did you let it cure for 24hrs?
@juliusvalentinas that part was fine. Had issues spreading it evening onto cold polished Nickel plated surface
the graphite pads i bought never arrived :/ used your link to make sure i buy it correct. strange. oh i see they dont send it to UK, very strange.
That is odd. Did you get a confirmation email for the order or any cancellation update from the seller? If there is still an issue let me know and I will contact the seller.
@@snarksdomain didnt get anything but just pop in to my aliexpress account i can see the items in my basket and they have red writing saying the seller can not send items to this address. And seen £101 payment at 5th august from 司順貿易有限公司 this chinese name, just open a case on it, as i dont recotnise this company. Anyway surely if its paid and i didnt get the item i willl get my money back but i am kind of 90% sure i ordered and paid with paypal thank god !
@@snarksdomain can you recotnize this name 司順貿易有限公司
@@alfiethefeline3311 Was it from "Mlddle999" seller? As that is the only one I'm affiliated with.
@@snarksdomain yes but i dont have any info from him at my paypal, only had that chinese writing ? 司順貿易有限公司
Oh there is something weird the copper and aluminum same temperature!! Isn’t copper better in transferring heat!! I studied those in details haha
I will be retesting this all soon with better logging of ambient temperature.
So the verdict is still out, in my opinion. But I believe it is quite possible that in this scenario, the weak links are the thermal conductivity of the putty and the Silicon. Since Silicon is around 90-160 w/mk, it isn't nearly as fast at moving heat as Aluminium is.
That's my best guess as to why I ended up with the same results for both.
@@snarksdomain that’s makes sense didn’t come to my mind! I was saying maybe it was the putty! But then I said ok but the heat already passed through the shims! Slightly confused
is upsiren pads 21w/18w/16w/ legit?
I'm not sure how accurate the ratings are, but I'm using some 21w/mk pads in a 3070ti, just on the VRAM with putty everywhere else. So far they seem decent performance. They aren't the softest pads, so it's important to get the sizing right, or one could run into core contact issues.
We should have a better idea of their performance once I retest everything.
i really want to make good soft to rap over but i suck in creativeness
You can do it! Follow your dreams. Carpe Diem.
Im using k5 pro putty on my 6800xt and it works wonders
Does it leak oil over time? Did it dry out?
A lot of people seem to love k5 pro, while others hate it.
I don't know what to make of it.
Im using k5 pro putty on my 6800xt and it works wonders
GDDR6 doesn't run too hot so you'd likely be okay with K5 Pro, although I personally won't use it.
For anything with GDDR6X I strongly recommend against using K5 Pro. It boils, pumps out, and off gasses at higher temps.
@@snarksdomain oh wow thanks for sharing this info, what do you recommend for gddr6 then?
@@kiko1764 TH949-1, Jeyi 8100, TH855-5, TH930, TG-PP10 in that order of preference.
@@snarksdomain thanks man, highly appreciated.
@@snarksdomain It's so funny. Some people love K5 Pro, others highly recommend not using it.
The vendor says it's okay to use up till 250C. Some people on Reddit also said it "boils".
Not sure what to make of it!
My goal is to have my GPU as maintainance free as possible.
Ideally I want to just blow out the dust from the cooler every 1-2 years and not have to change the thermal pads ever.
I will use Honeywell 7950 for the GPU itself. What would you recommend as thermal pads?