Відео

@lrmcatspaw1 Yes. They are interesting. It does okay, but can only compress down to 0.1mm (with considerable force applied). I'll test it again once Testbench 3 is complete.

@@snarksdomain Awesome. Looking forward to it.

I'd go Thermal Putty, and a Phase Change Material on the CPU/GPU core.

@@snarksdomain sorry Bro what is meanunf of a phase change material?

@umtman6695 A Phase Change Material (PCM) is one that melts when it gets hot. For CPU/GPU core applications you can now use a PCM instead of a traditional paste. They can melt and squish down to a layer as thin as 0.038mm (very thin). As it gets thinner there us an improvement in heat transfer. Most of them melt around 45 degree Celsius. Laird TPCM7000 series melts between 50-70. Currently the Phase Change Material that most folks buy is Honeywell PTM7950. There is also Upsiren PCM-1, Thermalright Heilos, Material XTP-001, Thermal Grizzly PhaseSheet, Gelid Heatphase Ultra.

@@snarksdomain thanks sir i want to use it on vram while u said that melt on 70 degrees but my vram up to 80 or above degrees is that fine to use?

@umtman6695 no, Phase Change Material is not suitable for VRAM. It's only for between cooler and CPU/GPU core. You want to use Thermal Putty or a traditional thermal pad for VRAM.

I'll have to go back and look, but I've done putty/pad jobs on 3 different Zotac 3070ti's. 2 of the cards are my friends'. If I remember correctly, one had VRAM temps around 106, another was in the 90's. The funny thing is seeing broadly different temps on identical cards. I've seen this happen a couple times where 2 or 3 of the same cards have different temperatures.

The excess overlap? Having extra is no benefit as far as I can tell. I now recommend everyone cut the piece of PCM to be ~3mm short on all 4 sides (1/8" on all 4 sides). Since the PCM starts off at either 0.20mm or 0.25mm, At least 80% of it will squish out ounce it melts and the Silicon will have full coverage.

@@snarksdomainso pcm can pump out?

@@vugiabao7tc425 It needs to initially to get really thin. Most PCM's are around 0.20mm or 0.25mm starting off. They are designed to melt around 45 Celsius, then with the mounting pressure of the cooler it then squishes out. They can get down to around 0.038mm layer between the die and the cooler.

I still need to test Kryonaut, but have put CPU and core testing on pause until I can make improvements to CPU/Core testing methodology and accuracy. I don't feel I can currently differentiate the performance of the various paste/Phase Change Materials within such a small temperature range and have a meaningful comparison.

20-25 grams should be enough for your laptop. I'll be redoing this laptop in the next couple weeks and will be sure to measure the amount of putty I use. As for how much putty did I buy....the only answer I have to that questions is "All of it". lol. Jokes aside, I've spent thousands of dollars of various different Thermal Putties.

CX-H1300 is good. Recently the batches have been firmer than they were previously and perform about 2 degrees worse than it shows on the chart. This still puts it as outperforming U6 Pro slightly, but definitely behind Upsiren UTP-8 and Fehonda LTP81. It will just require a bit more pressure to be applied during reassembly. definitely a good budget option.

@@snarksdomain Hope I get a good one, bc I live in a country where is difficult to get things from internet, but thanks for your help

@@jonnypena7651 Yeah, I hope so too. I tried buying from 4 different sources a couple months ago and they were all stiffer. Hopefully whoever makes CX-H1300 sorts out the batch consistency. The OG batch I have is still soft. It's still usable in it's stiffer form. Just takes a bit more effort to compress it. You can knead it in your hand like dough for a couple minutes to soften it up a little before application.

@@snarksdomain thanks for the advice, also maybe the high avg temp of my country would make it softer, you are under AC at 23C, I'm with a fan with an avg 28c at night

I'm not sure I'll be able to source any as I live in Canada. But perhaps I will find a way to get some.

It may help coil whine a bit. I'm packing putty between the coils at the 20:42 mark. It can't get rid of all the coil whine, since the coil can still vibrate inside it's plastic housing. Some folks go as far as drilling tiny holes into the plastic casing and filling them with stuff like Crazy Glue. I haven't done that myself yet, and doubt I will.

@@snarksdomain I thought the whine is because the coils touch the backplate sometimes. Ok will just put stuff around them like you, maybe it helps a bit. My card is quite ok anyway, but less is always better :)

One potential issue with these Graphite pads on the backplate of your 3090 is that if the backplate flexes a lot when you handle the card, then these pads can become over compressed and lose contact. I mostly recommend folks to use Thermal Putty, instead of pads, as it has proven to have more consistent good results. It also allows for better core contact as Thermal Putty compresses much better than Thermal Pads. If you do go with pads on your Zotac 3090 then it should be 2mm core side. I'm not sure about backside pad thickness.

@@snarksdomain thx for the reply! wouldnt really handle it much since i use it for gaming and not mining so I pretty much put it on the pc case and leave it there until replaced or repasting. Didnt really consider putty, which one would you recommend for the vram vrm's etc?

@brunogoncalves6677 either Upsiren UTP-8 or Fehonda LTP81. If you like you can check out the charts located at the Google Drive link in video description (Test Chart Repository>VRAM TIM>ETCHASH). You'd need around 75 grams of putty for your entire card.

@brupleal This experiment with paste was just for testing as someone had requested I try it out. I use Thermal Putty to replace thermal pads and have found it to be a superior solution that let's the cooler achieve ideal core contact. You may want to consider just using putty instead of paste+pads.

You may be best off rolling by hand as you'll need different diameters of putty for different components. If your laptop currently has pads and you happen to have a set of calipers then you can measure the OEM pads and enter those numbers into the putty calculator and it will tell you log diameter and number of grams used as well. A syringe works okay, but to know diamter needed for each component type you'd first need to measure the OEM pads sizes/thickness. What I recommen most laptops users to do now is apply putty as best you can and then do a test fit of the cooler, apply pressure and tighten screws, then reopen and see how the putty has squished. Then add/remove putty as needed to get the right amount. Once you're happy with the amount, then apply Phase Change Material or paste of your choice and reassemble. Be sure to apply manual pressure by hand once the cooler is on to help squish the putty. If core temps are higher than expected then apply more pressure and/or do a good burn-in to help melt the Phase Change.

@@snarksdomain Thx! The putty calculator is a new one) Where to access it? I will also google it. The best would be to have a syringe with somehow adjustable output diameter ;)

If there is something the putty could transfer heat to it might have a slight effect. The biggest thing with Laptops is that if it managed to lower the CPU/GPU core temps even 2-3 degrees it would be significant. On Videocards, there is usually such a large cooler, that putty on the back of the die has little effect. But with laptops having such a small and heat soaked cooler, there's a better chance of seeing a difference. I'd say go for it.

@@snarksdomain Ahh Okay, I just got a secondhand acer nitro 17 about a year old and temps on the ryzen 7 7735HS can reach over 95 C in spite of having LM The GPU (rtx 4060)can hit 88-90 C as well which I know definitely isn't good, in spite of having a 2 fan air cooler under the laptop and being in a cool environment The ptm7950 and cx-h1300 putty I ordered finally arrived today so I'm pretty excited to see how much of a difference it'll make, I watched your vids regarding application as well, good stuff!

@bobdole930 latest charts are located at the Google Drive link in video description (Test Chart Repository>VRAM TIM>ETCHASH). Upsiren UTP-8 is slightly better performing than Upsiren UX Pro Ultra, and also softer, and much cheaper. Fehonda LTP81 is also good, but a bit pricier. F I think Fehonda will need to lower their price to be more competitive.

@@snarksdomain thanks buddy, appreciate you keep up the good work! 👍

I actually prefer the svrew mounted chipset coolers as they are a bit more robust than those push through palstics. Also, old plastic has the added risk of breaking due to being brittle. So far I've mostly used Thermal Putty on Chipsets as they have typically had a gap large enough to justify putty. If the chipset has direct contact with the cooler, rather than having a pad from factory, then a Phase Change Material should be suitable, I believe. The one caveat being that the Phase Change won't melt and squish if it doesn't get warm enough. This won't be an issue if you control the chipset fan speed (if it has one). I think with a chipset temp of around 60 Celsius you will get it to melt slowly and it should improve over time.

@@snarksdomain thanks for the response I will probably stick with paste for the job but the ptm sounds like it could have potential. As for heatsinks I agree the extra work of gutting my machine in order to repaste/pad heatsinks is worth it when it means far superior durability.

@MrRoccale All test results are on the charts located at the Google Drive link in video deacription (Test Chart Repository). I'm planning on improving the accuracy of CPU/Core tests as the span of temperature that the results fall into is very small. I'm hoping that with improved accuracy the results will be better representative than they are now.

Oh Crazy! They are recommending K5 Pro? that is so nuts. I wouldn't even bother with UX Pro these days for a few reasons. 1) It is firmer than ideal for a putty 2) It's not as high performing as Upsiren UTP-8 (but very close) 3) It is more expensive the UTP-8 4) CSGR also has a version of UX pro, making it potentially confusing, and expensive, for folks to but the wrong one. For latest test charts follow the Google Drive link in video description (Test Chart Repositories>VRAM TIM> ETCHASH)

@@snarksdomain Thanks for the fast reply! And yea I hadn't done too much research before asking for the pads and now I'm kinda shocked they'd even recommend it. I'll go with UTP-8, it's considerably cheaper. Also laughed since they're using your image for advertising on their Aliexpress page. Just want to thank you again for your work. It's shocking that of all the tech subjects it seems proper pad/putty/grease is the least talked about.

@@jacobbetzo8435 Feel free to join the Discord server if you want a good place to see other peoples putty jobs and the things that worked/didn't work for them. It's also a good place to ask for help if you run into any issues when doing a putty job of your own. Also, for the CPU/GGPU core I reocmmend considering a Phase Change Material such as Honeywell PTM7950, Upsiren PCM-1, Thermalright Heilos, or Laird TPCM7200 (Needs higher temperature to melt than the others).

@@snarksdomain I'll totally join! I just got one more question. So I have the pads EVGA sent me. Should I use them in conjunction with UTP-8? Or would it be best just to use UTP-8 on everything and the PTM7950 for the die?

@@jacobbetzo8435 My gut says just use putty and PTM7950 and have the EVGA pads as the backup plan. This way you'll feel safe experimenting with the putty and have a good plan B. Or if you like testing stuff out and don't mind doing it a couple times you could see how putty stacks up against the EVGA pads. I think you will get the best core contact if using putty everywhere else and PTM on the core. When cutting the PTM to size for the die, it's best practice to go smaller by ~3mm on all 4 sides. This will make it a LOT easier to do 2nd peel, saves more PTM in case you need it in the future. Also, ~80% of the PTM will be squishing out anyways and you will most definitely still get full core coverage.

@Omnium94 Thanks for cheacking out the video. Of those 3 the best is UTP-8. It slightly outperforms UX Pro and is softer and also cheaper. Can usually get 100 grams for $35 CAD. Although for a gaming laptop you will likely only need 20-30 grams at most. I'd also recommend using a Phase Change Material on the CPU and GPU core. Fehonda LTP81 is also very good. It's basically tied for first place with Upsiren UTP-8. Fehonda will need to lower their price a bit to be a bit more competitive that way. The latest results are in the charts located at the Google Drive link below (Test Chart Repository>VRAM TIM>ETCHASH).

@niebuhr6197 They are both pretty good. LTP81 roughly matches Upsiren UTP-8. Fehonda will likely need to lower their pricing to be more competitive. Results are in the charts located on the Google Drive link below (Test Chart Repository>VRAM TIM>ETCHASH)

@@snarksdomain I'd use it on a laptop. Specifically 8 vram modules and 14 mosfet + inductors. One 12g syringe wont cut it, will it?

@niebuhr6197 you'll want at least 20g of whichever putty you pick. For my test card, with 8 GDDR6X and a gap of 1.8mm I use 15 grams, just for the 8 VRAM. Laptops typically have smaller gaps so your VRAM will use a bit less. 20-25 grams most likely, for your VRAM, VRM and Mosfets

@@snarksdomain and is upsiren pcm-1 renamed honeywell pad?

@niebuhr6197 I'm not sure if it is or not. My gut tells me some of the PCM's being sold by different brands are rebranded Honeywell stuff, but that's just a hunch.

It's okay but there are much better putties available. For latest putty/pad comparison charts, follow the link in video description (Test Chart Repository>VRAM TIM>ETCHASH). Currenty top two putties are Upsiren UTP-8 and Fehonda LTP81

Most top end putties do in fact spring back. It's because they stick to each surface a little bit. I don't recommend using a paste in place of pads or putty.

@@snarksdomain Sorrry, I meant putty rather than paste. That being said, I just checked the stickiness with some leftover UTP-8 and it seems to have issues filling the gap if compressed-decompressed slightly, detaching from the edges. The thermal conductivity is definitely good but to make the most of putties, heatsinks need to be designed to accomodate them with high pressure throughout. Going to keep putty on memory but long thin bars on VRMs with 3mm+ gap to be filled can't keep an amorphous mateial in place.

@Reirainsong I've been lucky I guess, as I haven't run into that issue yet myself.

@hornsteinhof7592 HY230 basically means you used pink putty. There are at least 2 versions of 230, that being 234 rated at 4 w/mk and 236 rated at 6 w/mk. It should say somewhere on the package which one you used 234 or 236. There is a noticeable difference in performance between the two.

@mark4asp Yeah. All my streams start that way. If I edit out the 3.5 minutes intro it then posts the video in the regular video category. I was hoping to keep the streams categorized separately from the normal videos. I think the countdown shown on the screen is fairly effective in letting folks know how far forward they can skip ahead. Perhaps I could add a timestamp to make it easier to get to the start.

@mark4asp if you like, you can access the test results by following the Google Drive link in video description (Test Chart Repository>VRAM TIM>ETCHASH).

@VigilSerus Thanks for checking out the video. Putty and PCM make a great combo. The putty compresses well, allowing for ideal core contact. Putty takes a bit of getting used too and is easy to use too much the first couple times. I used too much in the past as can been seen in some of my earlier videos.

Phase Change Material. Either Honeywell PTM7950, Upsiren PCM-1, Thermalright Heilos, or even Laird TPCM7200 (requires higher temperature to melt than the other 3)

@@snarksdomain but i was asking only paste and not ptm, since in India there is issue to get anything form AliExpress its banned

@@snarksdomain can u recommend any paste and not ptm. tell me about some good paste, since in india lots of aliexpress does not deliver

There are several good putties to choose from. I'd recommend taking a look at the charts located on the Google Drive link below for the most up to date data (Test Chart Repository > VRAM TIM > ETCHASH). Currently (July 2024) I consider Upsiren UTP-8 and Fehonda LTP81 to be tied for 1st place. Availability and pricing, depending on which country you live, may decide which one is best for you. As for how long those two putties will be good for it is hard to say for sure as they are both fairly new. Good putties should last for years, as has been the case with the TG-PP10 in several of my videocards. There are several putties that perform better than TG-PP10 did, when it was still being sold, and hopefully they will also last as long. Since many manufacturers are now moving to using thermal putties it's a good indication of their expected service life.

It's possible it could be throttling for some reason, but hard to know for sure without testing a few things out. I'd recommend installing HWINFO64 and running "sensors Only" and then recording a log (bottom right). Then make sure your laptop profile is set to a performance mode and playing a game or even running a stress test with Furmark to see what the wattage is. Then you can go back a look at the log, voltages, temperatures, and have more data to go off of to try figuring out what the main issue is. If the VRAM or VRM's are not getting adequate cooling then it could definitely effect performance. If that were the case, then either new pads or Thermal Putty would likely help. I'd personally opt for putty over pads, but that's my personal preference.

@@snarksdomain hw info or any sensor based app does not show temps for gpu memory . i have been doing extensive testing to look for problems but it seems like all is good since benchmark scores are still similar and it can boost core clock to the overclock i used to do. may be in new gpus it does not need to stays at full throttle when cpu is bottlenecking or when it does not need to use lots of powers to get work done. in heaven bench the memory was staying constantly at 6300 6400 6500(with overclock applied ) so may be its not a problem, core temps are like 70.

This is true for 30 series laptops. HWINFO64 can show/log other stuff though like "Reliability Voltage" if it was running unstable. You should also be able to Conditional Format the voltage numbers for each voltage value and see if voltage changes after a while of the laptop warming up. Then it's easier to scroll through the data and see if there is a noticeable trend for any of the values. It's a bunch or work to review the data, but may shed light on issues you're experiencing. Your core temperature seems good at least.

I wouldn't put too much weight/importance on the advertised w/mk rating of any Thermal Interface material. The ratings are used to help market a product. De8auer and Gamers Nexus did a really good video where they talked about the issue. I've even seen 2 products from the same company have a higher and lower rated product. And the one with the lower w/mk rating outperformed the one with the higher rating, significantly. These days I only trust testing. I do use advertised ratings as a rough guide on which products to test within a product lineup though. I'd also add that you can find PTM7950 for similar price. I've tested some from a few different sources and they all seem to perform fairly well.

Open stand with vertical GPU. Will likely be going to a closed case for Testbench 3 with 6800 (0.1 degree increments on Mem temps as opposed to 2 degree increments on Nvidia cards). Possibly a Fractal Design Torrent. Then I'll put the ambient temperature sensors by the front case intake fans instead of in front of the GPU fans. Will have to monitor the case fan speeds as well and have their RPM show up in the calculator sheet.

@@snarksdomain thanks you for informative feedback

I'll see if I can borrow the card and test it out again to check on temperatures. I might even open it back up and swap the MX-4 for Phase Change Material and convert it over to a higher end putty everywhere on the card. Will salvage the TG-PP10 if I do that.

@@snarksdomain sounds like a great idea. If it works out it should make a great follow up.

Just seeing your message now, sorry for the delayed response. Depends which pads you go with. I usually try to buy them directly from the manufacturer when possible. That being said, I don't recommend pads. I just use thermal putty these days when doing maintenance on my own cards. Lots of good putties to choose from. Comparison charts are located at the Google Drive link in video description (Test Chart Repository>VRAM TIM>ETCHASH). The current top 2 putties are Upsiren UTP-8 and Fehonda LTP81, both of which can be bought on Aliexpress. For the GPU cores I recommend using a Phase Change Material of your choice (Honeywell PTM7950, Upsiren PCM-1, Thermalroght Heilos, etc.)

@ernestphilyaw3959 This test apparatus was my early attempt at testing out different Thermal putties and pads. For pad and putty testing I used a fixed gap of 1.88mm using 3 little ball bearing in between the copper blocks. Now I do testing on an actual GPU, specifically on the VRAM (GDDR6X). You can watch a stream of my application for each individual product. The following stream after each product you can see me taking it apart and looking at the application. I perform 3 tests now. 2 hours of mining ETCHASH, 2 hours Furmark, and 2 hours of mining KAWPOW. The ETCHASH and KAWPOW tests are the most intensive on the VRAM. I log all the data, including ambient temperature, within HWINFO64. The 1st hour of each test is treated as a 'warm up' period. The 2nd hour of data is used for the charts where I compare the performance of the various putties and pads I've tested. You can find these charts on the Google Drive (Test Chart Repository).

Sometimes you don’t want to cool the vram chips as much as possible.

@@Seandude05 My application is laptops, which can run hot compared to desktop. Also, with laptops, if things are hot you can get the fans running loud or running high speed for longer than if things were cooler.

@wilkuuu4016 You can skip to near the end of the next stream for the results. About a 6.7 degree improvement over the MG860 putty. It won't be going on the official charts though, but I did log the data and review it.

@user-mi5hi1uq4z To clean it I first scoop up as much as I can wirh a little plastic paste spreader then I use 99% Isopropyl alcohol, q-tips, toothbrush or similar small brush, and those round cotton makeup wipes. You'll want to make sure the it's fairly clean before applying new putty. Some putties are easier to clean than others. Most of the top end putties are not too bad. Some of the lower performing liquidy/srickt putties take more time to clean. To get an idea you can watch some of my streams where I open up the test card to change which putty/pad is on it. To look at a specific product cleanup, you can find the stream with that product name and just watch the one after it.

@@snarksdomain thank you

@@Sald8Trin I used 44 grams on this 3070. You will likely need 50g at the very least. 100g is a decent purchase for around $35 CAD.

@@snarksdomain And this is u6 pro that you used here right?

@Sald8Trin This is Upsiren UTP-8. This is the latest putty from Upsiren. It is basically tied with Fehonda LTP81 if you look at the charts on the Google Drive.

Thank you for all your answers 🙏

@SilvaTeixeira009 I'd ac4ually recommend looking into using a Phase Change Material (PCM) for both the CPU and GPU core. Could either go with Honeywell PTM7950 sheet, Upsiren PCM-1, Thermalright Heilos or similar PCM. Kryosheet can only compress down to 0.1mm, and that's with more force than the majority of coolers can do. This was my findings when crushing a piece of it with digi5al calipers and squeezing as hard as I could (maybe ~50 lbs of force and a fairly small surface area (3/16" x maybe 1" long). There was a member of the Discord that recently switched from using PTM7950 on his 7900XTX to using Kryosheet. He had higher hotspot temps with the Kryosheet. It's possible you could get good results if the GapU die and coldplate were both lapped to being perfectly flat, but it still will not compress past that 0.1mm, where as a PCM can melt down to a minimum bondline thickness of 0.038mm (almost 1/3rd thinner than Kryosheet). If you do go woth Kryosheet then be sure to protect the ci texts around the core with either Kapton Tape or Conformal Coaring such TG Shield or MG Chemicals 422C, or similar.

@@snarksdomain I remember when I bought my gpu (second hand), I disassembled it and unfortunately I removed and discarded the original thermal pad, I don't know if it was carbon or graphite, and before I opened it it had a good temperature even with zero fan activated , it reached a maximum of 75c on the furmark, but I had to open it because the Vram thermal was dry, I put 1mm gelid extreme, since I opened it the temperature I don't like with thermal paste, I already put kryonaut and after a while the temperature increases, I already put GD-2, DeepCool Z10, and now I recently bought PTM7950 honeywell 40X80 on aliexpress but I'm waiting for it to arrive, I opened my gpu several times and ended up damaging the head of the XClip screws, I had to buy M1.2 5mm and M1. 4 5mm screw, I will have it tomorrow and I hope it will be the same size. sorry I wrote a lot

I remember when I bought my gpu (second hand), I disassembled it and unfortunately I removed and discarded the original thermal pad, I don't know if it was carbon or graphite, and before I opened it it had a good temperature even with zero fan activated , it reached a maximum of 75c on the furmark, but I had to open it because the Vram thermal was dry, I put 1mm gelid extreme, since I opened it the temperature I don't like with thermal paste, I already put kryonaut and after a while the temperature increases, I already put GD-2, DeepCool Z10, and now I recently bought PTM7950 honeywell 40X80 on aliexpress but I'm waiting for it to arrive, I opened my gpu several times and ended up damaging the head of the XClip screws, I had to buy M1.2 5mm and M1. 4 5mm screw, I will have it tomorrow and I hope it will be the same size.

I know that AMD's gpu has problems with hotspot, mine reaches up to 16 degrees of hotspot, I bought thermal pad from arctic, my idea is to close the gpu and not open it anymore, but if you recommend using ptm7950 then I will do it, but I researched about PTM7950, when it completes 1 year of use, the capacity drops.

@@SilvaTeixeira009 I'll be running some tests and then cracking open my laptop soon as it's had PTM7950 for a little over a year now. Will make a video.

@jimpaolo123 the is Kapton tape. It's a special high heat resistant tape that is also good to protect against electrical shorting, if the Kryosheet were to touch any bare metal contacts next the the GPU Die. Another good option for protection is conformal coating, like TG Shield or MG Chemicals 422C (glows under UV light)

@@snarksdomain Good thing you added kapton tape, some guys added kryosheet the whole area without a tape in his Rog Ally 😅

@jimpaolo123 oh no. Sad to hear when people break their gear over something so simple. Graphite film is very good at conducting electricity.

@@snarksdomain Thank you for the knowledge, I will be adding mine soon. 👌🫡