Is repasting your GPU still worth it?

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
well here we are in 2023 and I'm continuing to get asked whether or not it's worth it to repaste a graphics card so today we're going to kind of take a mid-range upper mid-range older card and we're going to see whether or not we gained anything from repasting it we'll also talk about when to repaste what kind of pace to use what kind of pace to avoid and then just kind of take you to the process of getting it cleaned up and new pace of [Music] Life the new mv7 full tower PC case from fantex creat the perfect picture frame to show off your system insane airflow and direct GPU cooling creates the best possible environment for modern high- power gpus while the internal and external Cable Management Systems make building and maintaining your system extremely user friendly the integrated drgb lighting is controlled by an included dual Channel drgb controller with all new effects allowing you to make your system Uniquely Yours to see the full spec list and options follow the link in the description below so what this is right here is my EVGA RTX 370 xc3 ah EVGA in the arms of the Angels something something sad lyrics this card though honestly um it's going to so I was conflicted on how to do this video if I wanted to be repasting an older card cuz we know thermal paste does degrade as time goes on so an older card could definitely benefit from repasting but the questions I'm getting lately actually are whether or not it's worth repasting a new card so kind of hits both of those questions because it's 3 years old the RTX 30 series launched in September of 2020 seems like an eternity ago so this is a minimum well it's maximum about 3 years old at this point now there's different types of thermal paste different types of thermal compound they're not all created equal so like this is Arctic Silver um this was very very popular back in the day because it actually had silver particles in it which were designed to uh add more kind of like a metallic surface area to it to carry heat the problem is it was also conductive so getting this on components that carry electricity that are not conformal coated or have a clear cating on there to protect it from contact this could kill your stuff so that's why Arctic Silver sort of kind of went the way of the the going away then you have stuff like mx4 from Arctic which became very popular because it was in inexpensive it was cost effective um but the thing is like thermal paste has actually improved even Beyond this particular type of formula um KP extreme or Kingpin extreme is what I tend to use the most of because I get the best results out of it it stays liquid for longer the problem with thermal paste is because it's actually a liquid based product like any other liquid it can evaporate over time or lose some of its active materials in there that are liquid and then start to have less um transmissive properties to it so it'll get harder it'll become more like clay rather than being a liquid um which will obviously hurt your thermal transfer um Kingpin extreme the kpx the blue stuff here pretty much never turns into that hard state it it's kind of amazing and in fact if even if it does start to get warm or like cold in the tube and start to thicken up just hitting it with a little bit of warmth it'll start to run really nicely again so every time you turn on your graphics card you're basically kind of like reifying it if it does start to do that and then you have like the thermal Grizzly stuff which is um der's brand he's got thermal Grizzly he's got conductonaut kryonaut it's got all kinds of different knots if you will liquid metal and all that sort of stuff this is what you'll typically find included with a lot of air coolers water blocks a lot of stuff just throws thermal grizzly in there cuz thermal Grizzly kind of bridges the gap between something like mx4 and a and the more Boutique higher-end thermal Pace like kin extreme or conduct a knut or cry knut um anyway first we have to get our Baseline because this card has only been used for a handful of reviews and never ran on a daily basis the thermal um material in there or the thermal paste has not been cycled a whole bunch every time it gets cycled like I said it kind of goes from a little bit more of a solid type of State even though it's still liquid it's a thicker more coagulated and then as it gets warm and is used under temperature it sort of Thins and and over time that pressure can start to oooze out or E like squeeze out of the the gaps that are created because remember all thermal past is designed to do is to fill in the microscopic Peaks and valleys that exist between both the cooler and the actual diet itself to give you much better contact obviously the thinner we can get that contact the closer together those two metals can be um the better it will be so that's why the thermal properties of thermal paste they call it a thermal paste it's really a thermal interface material or a Tim you hear that it's just a material that goes in between two things that are designed to kind of bond them thermally the thing is they're not all created equal so we need to know where we stand right now in terms of the outof the-box temperature so because it hasn't been used a whole lot um it hasn't had a lot of those Cycles on it which means it should still be acting as if it were day one at this point minus whatever may have potentially evaporated um over the the last 3 years and it's as it sits in our warehouse it can go anywhere from Fairly humid to really really dry given the area that we live in so it's had quite a few seasons now of humid dry humid dry humid dry which can obviously have a huge effect uh on the overall performance of the thermal paste I'm going to be using 3D Mark Speedway because it's an RTX graphics card to get full load temperatures I've got to hit the RT cores if I don't hit the RT cores it draws a lot less power it creates a lot less heat and then we may not get a a huge um we might not get an accurate result the other thing too is like I kind of want to do this with like a 3090 TI or a 4090 just because of the massive amount of power but realistically I'm trying to see what the results look like in a more average use case rather than cuz here's the thing if we found oh yeah we dropped three or four C on a 4 90 at 500 watts a card like this which is like 220 or some 200 really low wattage power probably would see no difference whatsoever so I figure we'll just kind of land in the middle here with a 3070 um to see what the actual temperatures look like so what I'm going to do now is i' I've got my core voltage maxed I've got my power limit maxed my temp limit maxed but no added core clock no added memory clock and I've locked the fan speed to 75% if I don't lock the fan speed we don't get an accurate comparison we're also in an HVAC controlled room the temperature in here is going to be at a locked steady 71° fhe that will not change this room is very very the AC in this room is designed to keep it at a very constant temperature so with that said we're going to let this Loop until um the clock speed or the temperature completely stops to R stops Rising goes completely flat and then we'll go through the repase process and we'll compare okay so we have pretty much completely uh maxed out our temperatures 64 C is where we're at it hit 65 for like a split second and then came back down to 64 and it hasn't moved our Hotpot maxed out at 77.9 we're really sitting more like 76.1 to 76.4 depending on what part of the test it's in um so these are the temps right here that we want to at least not exceed if we exceed that after we've repaced then either we didn't get full coverage on the DOT which is a problem um or we didn't put our cooler back together right or we have a mount issue there's a lot of reasons why it could go up we should never see it go up unless we just put complete crap paste in there um so anyway core clock that's the other aspect of this that matters so we started the test here at 2010 and then it pretty much immediately starts to drop with the uh temperature as a temp goes up the frequency drops a little bit so 1980 so it was pretty solid right here see this like the line has got little dips that's every time the test restarts but now look now it's like like it's gotten a little bit more Jagged cuz what's started to happen here now is we're getting the vrms are full temp temperature soaked the memory is temperature soaked the PCB has gotten hot so as the core is cool uh that core temperature may hit 65 or 63 whatever it is right there but then as that temperature radiates out from the core and warms up the rest of the board then you start to get instead of this nice flat 1980 1980 1980 as we get over here you can see it starts to really fluctuate between 1980 and 1965 that's just one boost bin right there 15 MHz you can see now it's up and down up and down up and down because this is how long it took for the board to fully saturate so this is why you run these test for a while because it's it's about more than just the core temp it's everything else on the board as it gets hot so these are the numbers that we are looking for improvement on if we see our core clock improve or at least stay up in the 1980 to 1995 then we might see the temperatures stay roughly the same but if the heat is able to more efficiently escape the core and not spread out as far across the PCB if you will then um that could lead to another boost bin or two higher which would be a slight Improvement in performance but realistically all we're looking for here is just longevity of our card we're trying to keep it alive as long as possible we're trying not to lose performance cuz as thermal Pace degrades and the temperature goes up as I already showed you the core clock will go down with it so let's go ahead now cool the card off I'm going to stop the test I never just turn off the system with it being full hot I always let the card run and then what I'll do is I'll put the fans to 100% when it's not under a load because again even though the Core is coming down it's at 44 43 42 remember the rest of the PCB has also Heat soak so I want to give it time to cool down then I'll shut it off uh and clean off my thermal paste so obviously the way each graphics card comes apart is going to be a little bit different aftermarket cards uh tend to be a lot easier than Nvidia cards by the way um few things you want to be mindful of make sure you know where all your connectors are for your fans this is going to be a bad day if you rip a fan connector so each one of these is for a fan so we got black tan and blue this is just front middle rear and the fans um this one's just using standard phillips heads stuff so we're not going to have to worry about using any sort of like pentalobe or crazy stuff like if we were taking a part an Apple product um but anyway keep in mind too this one doesn't appear to have any warranty void stickers even though in the United States you can't void a warranty by putting a sticker over something brands are still going to try and uh scare you away from working on your own stuff but we believe in write to repair around here um on the other side here too you'll see there's another plug that's because this also has an RGB uh this evg here lights up of RGB so make sure you locate all of your plugs and unplug those gently so to unplug these it'd be useful to have a little like plastic spudger tool or something but I'm just going to use a part of the Phillips head so I'm going to be unplugging these after I get the cooler a little bit loose I need some slack to be able to pull the the cable down I mean you can kind of do it with it together still but I don't want to damage anything if I were to damage anything I'd rather damage the RGB than the fan headers for obvious reasons but I can't really get my fingers down in there so I'll just have more room when the cable or the cooler is loose and get it apart so I'm going to start by taking apart the most perimeter screws I also like to use the lid of the iix it kit as almost like a grid so I can remember where things went like I said obviously your card may be different it depends on the brand every brand is a little different how they do this it's acting like there's a screw that I missed nope so it was just the thermal pad so you saw the way it didn't seem like it wanted to come up that's because the thermal pads that are on the back were making pretty good adhesion this is also adding you know letting Heat come from the back of the vrm to the back plate so be mindful of that make sure you get all your screws off before you start just yanking it up so set that aside the only four screws left now we're going to be the four that are around the GPU perimeter here so these are spring-loaded screws that's because they're designed to bottom out uh and then let the spring hold the tension evenly across the back of the die so these are pretty easy when you put them back together you just tighten them down as far as they'll go go snug you don't have to go any farther than the snug cuz literally the threads are stopping for a reason and the Springs are doing the rest of the work so right now the cooler and the GPU PCB are pretty stuck together and I want to also make sure that there's no screws on the front right here sometimes there'll be screws on this IO plate screwing directly into the PCB this one does not have or directly into the cooler this one does not have that so I like to start on the back of the card and then just sort of twist and it's kind of scary because this one you can tell has been together for a while because this thermal paste is definitely acting like paste and not wanting to come apart but that is believe it or not completely freaking Norm noral he really glued together there you know there we go so you saw there was a fair amount of tension needed this is also fairly typical you can see the thermal pads for the ram kind of just ripped on me right there that's going to probably happen that's probably going to be inevitable believe it or not that is not a problem as long as the thermal pads go right back in their spots it's not going to be be an issue so I'm going to start by cleaning and I'm going to leave this residue for the thermal pad and everything on here I'm not going to clean that but I'm going to take blue shop towel I'm going to take my isopropyl alcohol and I'm going to spray it on the towel right now and not the card fold it over so that the wet part is just like under my [Music] finger it's actually pretty hard by the way this is this this I can demonstrate this this thermal paste see how hard and like kind of flaky this is I it didn't really seem to affect the temperatures I mean 64c is kind of exactly where I would expect this card to land but I just want to point out that it definitely is not what I would classify as liquidy anymore like I'm really having to scrub this off of the copper but it's important to get a really good clean surface we don't want to cross-contaminate the new fluid with any of the residue from the old fluid for chemical reasons I mean they could be two completely different types of fluid i' I've never actually tested what happens if you don't clean one you just take the leftover remnants of one and put a different past on it put it together and see what happens if you guys want me to do that video then make sure you guys comment and like down below so I know you want me to do that I've never done that that might be interesting so I'm going to be using this spreader that comes with all the king pin extreme knowing that it's pre-s spread I don't have to rely on mounting pressure to spread it um it'll squish off the sides it's non-conductive this is all conformal coated nothing's going to hurt it but as long as you kind of just Smash and move in One Direction you're not going to fold air into it it's not going to be a problem it will squish out if any does get in there I've literally never had this be a problem I've been doing this for every CPU and GPU I've been doing in the last like 5 years so now we've got our frosted GPU die we're going to put our cooler back together in the exact opposite order of what you just saw saw we're going to run the test again Corsair brings gaming to the next level with the Zenon 45in flexible OLED Zenon Flex display with up to 240 HZ refresh rate 0.03 millisecond grade to grade response time motion blur cancelling anti-reflective coating burning protection and customizable Bend based on users preference the Zenon Flex from cor air allows Gamers to truly tailor their display to their liking click the link below for more details well said 66 that's hotter by two but I want to point something out before we look at the chart I want you to notice the the more consistency of the clock speed right now so this particular part of the test it it tends to go up down up down that 1950 was unfortunate but watch 1980 so I'm noticing it stays at 1980 longer now what I'm curious about is what our um Hotpot is well our hot spot's at 80.2 81.6 so in this particular instance you can see we came up about 2 C across the board now there's a couple reasons why this could be one it takes time for thermal paste to kind of get melded in if you will um usually the worst temps you'll see is right after it's applied so right now I'm giving it this nice hot um just kind of a looping run to get it nice and hot what I'm going to do is I'm going to stop the run and I'm going to cool it down and after I cool it down then we get get it hot again often times you'll notice over time the temp will start to drop so I wanted to point this out because of the fact that there's a lot of times people will repaste their stuff and they'll notice an maybe an immediate increase in temps but that's really common because the thermal paste like I said does actually take time to liquefy and really start to work its way into all those microscopic Peaks and valleys and then create a bond there so I'm going to go ahead and stop the test right now and I want to see how fast the temps come down I mean look at this 52 51 50 49 and we're still at at 75 uh% on the fan speed so I'm going to go full speed right now I'm going to let it sit for about 15 minutes at full speed to get it as cool as possible and then I'm going to go ahead and uh I'll run the test again okay so I wanted to investigate the temps cuz it just didn't seem right to me and you can I can tell by the mount by looking at this see how it's like real thin on half of the die and real thick on the other half of the die you notice the thick part that means it didn't compress down fully so if it's if it's real thin that means we've got a nice tight and then the past squished out and the only thing that was left is what was needed to fill the voids if it's thicker and it's got that like kind of like a melted craft cheesy look to it that's where there was more Gap and so that's on the side where we had the torn pads I have a feeling that when I put the card back together if you take a look in this pad right here both of them you can see that there's some raised areas because the pad sort of torn in half I have a feeling one of the the torn pieces still on the cooler sat on top of one of these other torn pieces and not within the piece that it tore out of which created a thicker pad because even though it compressed it was creating more of a a thicker area there that couldn't compress down all the way so what I'm going to do now is I'm going to have to try and very carefully transfer this torn pad over to where it goes now it won't be too bad because I'm using the same spreader to just sort of cut it or scrape it off rather than trying to peel it off if I peel it off it'll stretch and it'll do all kinds of things that just won't be beneficial to our end goal here so now just mush those back down I should have just done this the first time but it's important that we show these kinds of things that way you can keep this in mind with yours see now I've got it all kind of like one solid pad again and you can even take this and just be like smoosh SM Smo smoo Smo smush so yeah it may have only been a couple of degrees but it was enough for me to to go you know what I've had enough experience with this thermal paste had enough experience with these gpus to where um something just wasn't right all right so after doing what I did to the thermal pad check this out the 1980 is just stuck 64c which is fine we're power limited that makes sense 1980 it's not fluctuating and the voltage is actually staying down at the 1.03 7 and .025 even more than than it was in the past so there's a 1965 for a second there that's way more solid than it was our hot spot is currently at 77.4 the max is 79.5 but it's it's not about what the max Peak is for a moment it's about what the sustained um average is so we're we're looking pretty good not much change we came down actually almost 2 C on the average though for the hot spot but if we take a look now at MSI After Burner look at the frequency now do you see how solid this is remember how the first half of the test was solid like this and then as the board gets hot it started getting super Jagged this is this this shows right now we actually gained something from this now are you going to notice this in your games no you're not are you going to notice if your thermal pce starts to degrade and starts to get really bad where your temperatures are getting high and then your fluctuation and boost bins are dropping consistently as you game yes that you'll notice so pasting a brand new thermal graphics card really is only going to matter if the thermal paste that they use is crap now most of the stuff they use at these factories is like huge drums and barrels of this uh industrial grade stuff which is I mean realistically here the best thermal Pace from like the most generic thermal paste usually is a couple of C maybe three C at the most on like CPUs and gpus and stuff but what you're gaining here with a quality thermal paste if you're the kind of person that just wants to have peace of mind is knowing that this is depending on if you use the right kind of thermal paste it's not going to dry out it's going to carry um You Know Better Properties to last longer because once you start going five six years of continual use then you'll notice the breakdowns start to happen and if you do this right away the breakdown won't happen uh nearly as quickly or even be as apparent as it's happening the downside is pretty much every single graphics card manufacturer is going to try and avoid you warranty if you take the cooler part again legally they can't do that but they're certainly going to try to deny any RMA issues and or any rmas that may pop up if that sticker is poked and that's a whole different conversation for another day so at the end of the day would I recommend repasting probably not just from the the fact that if we're talking about a brand new graphics card uh the the risk of damaging something outweighs the benefit however if you've body used graphics card or something that has been mined on and has just run weeks days months years at a time even though those are not GPU intensive type of tasks they just being on and having current going through the die that long creates heat or if it's someone that's just naid gamer and they spent 5 six years gaming the crap out of their graphics card you'll probably gain more from taking it apart and deep cleaning it than just the thermal paste so there' be other factors in play there but there you go there's my results you can take that and kind of formulate an opinion on whether or not it's worth it for you thanks for watching guys as always we'll see you in the next one
Info
Channel: JayzTwoCents
Views: 503,710
Rating: undefined out of 5
Keywords: gpu thermal paste, how to change gpu thermal paste, how to repaste gpu, is gpu repasting worth it, should I repaste my gpu, how to repaste, how to reapply thermal paste, how to reapply thermal paste to gpu, how to repaste 3070, how to repaste 3080, how to repaste 4080, how to replaste 4090, how to replate nvidia
Id: 1fEYOIsfnBs
Channel Id: undefined
Length: 23min 32sec (1412 seconds)
Published: Fri Oct 13 2023
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.