I watercooled an NVME SSD... These results were unexpected!

water cooled ssd is it even worth it viewsonic is proud to announce their all-new elite xg32ou gaming monitor the xg32ou features a 4k inch display with a one millisecond response time and 150 hertz refresh rate and builds upon viewsonic's revered elite panel lineup the viewsonic elite series monitors feature clean aesthetic design that blend well with any setup whether it be professional or personal due to their sleek design and tasteful lighting to learn more about viewsonic's gaming monitors or to see their full lineup of monitors click the link in the description below so nvmes we we know that if the controller gets too hot and the nand gets too hot it slows down just like a cpu it'll throttle so motherboards have started implementing different kinds of cooler options in there so if we look at my msi board right here with my amd uh pcie gen4 it's up to like 7 000 megabytes per second read write on a pcie gen 4 ssd you have these cool these covers right here which are designed with a heat pad to touch the ssd to give some sort of a heat spreader to try and keep it as cool as possible other brands have gone as far as including a heat spreader like an actual heat sink with a synced design to try and improve cooling even more so today i want to try a couple of different things oh and then of course corsair and ek and other brands have even come out with blocks to put on your ssd to keep it as cool as possible uh first and foremost this is not a sponsored video by a corsair even though this is a quarterback block that we're trying i'm just genuinely curious they sent us some of these blocks just with some of the hydrax stuff and i was like wondering if this is even necessary or not so first we'll talk about the block and the way it attaches real quick i did put together a loop which is sitting back here which already has a block on it that i'll add to our ssd so we can do some a b comparison testing in terms of speed because what we're looking for here is as temperature goes up is speed and throughput on the drive going to go down and then i kind of want to see like with our base config with this cover on here is that adequate i want to take a fan and blow it out the cover right this is not a heat sink it's just a piece of metal they attach with a thermal pad and then i want to put the actual corsair thing heatsink on the same drive i'm not gonna change drives because that changes behavior obviously uh each nand may perform a little different but i want to put the heatsink on there and then ultimately we're gonna put the water block on there because what this adds it's just an extra level of complication to a water cooling loop it potentially adds another point of failure as well so but these these little blocks right here are kind of neat like you could stick these on vrms you could stick them on memory like for on the gpu or something and do some custom stuff if you wanted the one thing i'm also curious about is how well they're gonna fit because now that we have much higher capacity drives two terabyte four terabyte you're going to have drives that have chips only on one side and you have drives that have chips on both sides like this a data xpg drive here so that's something i'm curious about so i mean if we take a look at the block already here if i can just put my fingernails off it just clamps on with this guy right here so you can see it's got a thermal pad to a copper plate you can see the copper right there so thermal pads already are not as efficient in transferring temperature as say like thermal paste is but i'm just going to use it the way it's designed to be now you have a thermal pad on the back side right here but this does not make any real contact with this side so if you had a double-sided ssd like this i would be concerned about a cold side and a hot side so the most important thing to cool honestly is the controller not the nand itself but the controller so you can see the controller is on one side which is right here so that touching the block obviously is the most important piece this will just allow some sort of transfer of temperature away from it into this piece but as long as you don't if you're on the airflow then this will just heat soak as well and then this attaches in the exact same manner as you can see they're probably even the same yeah they're even the same bottom piece right there this just gives you an actual heatsink design which is more efficient at radiating the heat away rather than a flat piece of metal like that so our testing methodology here we're learning that it's there's not a lot of software out there yet that's designed to long-term stress tests in ssd i mean we've got crystal mark crystal dismark you know which will allow us to go up to 65 64 gigabyte size file with us just straight sequential read write but the problem is how long does it take 64 gigabytes to copy at over 6 000 megabytes per second right 10 seconds so we have the test running nine times and what i end up having to do here is i'm and i'm kind of skipping the the small sequential read right we're just going with the large files because that's going to actually put the worst uh that's going to be the worst case scenario in terms of temperature if it's just having a copy of one big file and just move those bytes that's faster than it doing kind of a random rewrite which is kind of searching the drive this is a harder faster test so as you can see right here i'm setting this to 64 gigabytes per second we're gonna be running this test nine times now i've gone in here and tested done the ssd mode and the standard default mode i actually find the default mode puts a bigger stress on the drive and makes the temps go up more than the ssd mode itself if we take a look right here you can see on cpu hardware monitor ssd 980 pro one terabyte so this is a one terabyte nvme 980 pro and this is the current temperature the nice thing about the smart stuff on drives you can see the actual temperature it does it does read out so we're just idling right now at 30 32 c so with no fan on here just there's a little bit of ambient airflow because of the vents up there kind of blowing down not enough to really be considered active cooling in any way so crystal disk mark 64 gigabyte file i'm only going to do the sequential read which is just the entire thing at once um nine times that's the max on the test that you can do i really wish we could set this to like a hundred times that way it's essentially just a looping non-stop test um but we're gonna have it uh we're gonna hit start here on just sequential read write we've already gone up to 39c and what we're kind of looking for here is where the temperature stops climbing it's already five tests in on the nine that's the thing that sucks about this is how long it takes uh or how short it does the duration versus how long it takes to create the file right is where we're going to actually see like the biggest temperature increase because reading is one thing writing to the drive that will put it under some serious stress 4 901 megabytes megabyte per second write speed now you can see why pcie gen 4 is so nice so rather than bore you guys with this and please for the love of god comment in the comments down below if you know of a better test for this something that is like cinebench but for your drive see all these tests that we found are like how fast is your drive not how hot does your drive get because i don't think that's something a lot of people think about is how hot does my drive get the only time we actually thought about it was when the fury x led like rgb drive that we had was completely locking up the system because we found that the lighting itself was heating the drive up to the point where it was throttling down to nothing that was the first time we even thought about the fact that ssds could overheat and that could be a source of locking up in your system we were getting blue screens and complete lock-ups i mean just trying to get into the os after the system was on for a while and we did a whole video about that actually and that's the that's the scary thing is that you know these aren't things you really think about and it could be potential points of you know issue on your system maybe what i should do is actually just remove the cooler i think we should just remove the heatsink thing from the board because that'll be the first thing to tell us if we're actually getting any benefit of that and i can do it with the system running because this drive is held down with its own tab the question is taking that off now oh yeah i just hit what from 46 to 50 just like that yeah wow okay if you've ever wondered if these things actually work i mean i know that there's been a lot of video discussions about whether or not these things actually work here's the thing it's hot if it's hot it's working it's taking on the heat right and clearly spreading the heat across this look at that 59c 61c but look we didn't lose any we didn't lose any write speed 63c or read speed sorry yeah it reads me yeah we lost a lot of write whoa it came that's weird at 66c it's that technically still has not slowed down just that one test could have been something anomalous with that one but if you're if you're seeing higher temperatures but it's not slowing down um that's not indicative of cooling not being necessary because it then comes down to a longevity lifespan kind of a thing it'll be a fun test i'm going to reattach this in the middle of it going i mean so far this this speed fluctuation is kind of all within margin of error in terms of just the test look at how far that just dropped down to 48 65 to 48 just like that okay so i know that msi i think it was msi that uh gamers nexus was given a few years back um pretty hard time about these covers not doing a damn thing and then they said okay we redesigned it now they will clearly that worked nice uh example right there of public pressure before we take that off i want to push some air at that and see what happens and i'm gonna have the speed of the fan running very fast i'm just gonna ah nah it's fine it's fine temperatures are still coming down we're pulling the temp down under load by just throwing a fan at that cover the hardest way to test a cooler is to let something get fully saturated temperature wise and then increase the cooling to see if it can pull the temps down it's a lot easier to keep the the peak or the ceiling of the temperatures well there it goes again i think this is something to do with the test more so than it being the actual temperature related drive or performance there as i was starting to say it's a lot easier to keep the temps from climbing really high if it has an adequate cooling from the start of the test versus letting it get above where the cooler wants it to be and then try and pull it down that's a lot less efficient this is not our os drive too that's also important to point out this is a separate drive so it should be getting us you know as fast as we can um like if it was our os drive it probably a little bit slower on the read speed just because it's also doing operating system stuff yeah it looks like 42c is kind of our there's 40. yeah that is i don't understand what's causing that seeing some major fluctuations between write speed from test to test see look at that it just jumped up to 4 000. now it's at 5000 what so clearly the drive being cooler did help performance stay up let's do this now i'm gonna shut down the system and i'm going to install the actual heat sink device versus the plate cooler here because i want to give you guys a real recommendation here like should you take off these little covers and then run a heat sink like this it may not be as pretty but at the end of the day if you're all about straight up performance then obviously you should do what is going to be best for the system in terms of speed so installation should be pretty straightforward so here's the drive yeah the nand's right here the controller's right there the cache is right here nothing on the back at all completely smooth nothing back there with a sticker so installation is pretty straightforward let's just make sure i'm not covering up my pins so now that i've got the corsair heatsink on here our idle temperature is 2c lower already at 30c and we're going to start without the fan the same test and it'll probably take a solid uh two or three times for it to reach its max temperature the fan should be more effective on a heat sink than a flat piece of metal i mean a heat sink without fan or without air moving through it heats soak at that point it needs to transfer the air or the heat to something which is the air which means it needs airflow typically in your case you would have some sort of airflow going here so our open air test bench without a fan blowing on it's kind of the worst case scenario here and it's still not like overheating our open air test bench without a fan blowing on it's kind of the worst case scenario here and it's still not like overheating back up to almost 5000 on the write speed so the test is clearly just kind of weird when it comes to write i'm going to do this test two more times and then we'll come back with the results after it's gotten proper uh heat saturation in there it's not 45c it's not looking too different from the standard heat you know the little piece of metal here i feel like without airflow going over that it's almost worse than this because this is a bigger spread a more thermal mass versus that but now that it's done let me put the fan on let me let it cool it down and then let's run the test again 37c is where it's maxed out there somebody comment below why this is happening why from test to test it like alternates between the 2000s and then the 5 or 4 thousand 37c so it's clearly not temperature related this is just test related there this may not have been the best piece of software for this type of comparison but we are still getting some results at least you know this isn't some uh deep dive as curious like what's the what's the layman's like reaction here to hey thing with fan nothing fan what go now it's time to do the water block this might be the most ridiculous thing i've done in a while that's why you always leave the lids on the caps on here i might have should have given myself a little more slack i also shouldn't put the drive in there backwards appreciate it thank you so we've got a 240 millimeter radiator only cooling the thermal capacity i mean it's idling at 23c okay so it's doing something i feel like in terms of maintenance this is just like the worst idea because i mean you could go like you know gpu to nvme to somewhere but i change drives kind of a lot so i don't and i and i have been using nvmes for years now with just the you know the cooler covers that come with the motherboards and i've never had an issue that stemmed from drive performance so look 29c under that test still right exactly where we were landing on speed if we were going to do this mythbuster style that means we have to find where the slowdown starts it means we will find the point at which it slows down with the help of thermal power i honestly expected maybe to see this number go up this is watch the next test we'll do it again real quick that's going to drop again to the 2000s probably because that's oops that's what the test seems to do for some reason like alternates but our max temp was 30c our max temp was lower than our idle temp of our previous setup so in terms of the water cooling it's like it's nothing more than just a flex which is fine i mean this is flex tubing so it's appropriate but if you were just like i want to fill my system up with all the tubes and stuff that you can well then you might as well get yourself water cooled power supply yes they do exist water cooled ssds water cooled graphics card water cooled cpu and then just like all the distro plates and stuff that you can imagine and all the rads you can think of just to fill your case with water cooling until the first time you go to take a part out and then you'll realize just how much that sucks like i said we need to figure out at what temperature does the test start to uh slow down and we're gonna put the drive back without a cover and i'm just gonna start like slowly hitting it with a heat gun without melting the motherboard hopefully to artificially increase the temp and find where it starts to slow down right back to where we were 67 degrees celsius after the first test with no sort of heatsink or cooler on there this is actually test two i want to get saturated so we can make this faster but the ironic thing about 68c is it still didn't slow down we think right around 70 to 72c is where it might start to throttle based on what we saw with the fury x remember that was a two and a half inch ssd however it's very similar internals it still has a controller it still has nand and all that look at that 70 71 still absolutely no slowdown right here this is the weirdest thing i didn't expect to be doing this today okay i just also made the oh okay i made it the hottest temp that it can be abc oh yeah phil said i'm going to burn a samsung logo into my knuckle if i keep this up there's 98c okay i'll stop right there if that doesn't throttle it's still going 99.7 percent i mean it's still creating the test file which is taking longer than it has before yeah it's clearly throttled clearly 92. we might have actually like overcooked it we basically created like the turkey from national lampoon's christmas vacation where they poke it and just yeah we overcook 1410. dude it could still do 1400 megabytes per second oh dude at 79 it went back at the full speed no you don't oh no 50 okay come on maybe four oh that right speed 138 157 megabytes a second look just by putting the fan on it like how fast it's starting to bring the speed back up yeah look at 82c it's at 5300 so it's clearly throttled about a thousand megabytes per second oh look at that 78 and it's up to full speed so now having found where it starts to throt at least on read we've got to see what right does so it's at 81 now on read and it's probably probably has gone slower than that now this probably is still showing max speed which i would like to see just say what was every test but whatever i need a better test for this 4 54 at 79 c the four scenarios at which we tested here was with the motherboard cover just the cover landed us in the low 40s low to mid-40s which is pretty impressive considering we see where it goes now without a cover 67 to 72 depending on what the ambience is going to be like we then tested it with the heatsink on here which without any airflow performed worse than the motherboard cover but with airflow performed better than the motherboard cover and the reason for that is just the thermal mass this is a bigger chunk of metal it has more thermal capacity than the heatsink but the heatsink is about efficiency it's about taking the heat and having more surface area through all those fins to transfer that heat to the air when air is flowing over it so you have to have air flowing over it the water block just the pure flex i mean yeah i'd obviously it obviously kept it the coldest but as you could see it didn't make a difference to performance whatsoever so is it necessary absolutely positively not under any circumstance technically neither is this or this if you have any sort of airflow going over your drive it's dropped down to 44 right now with the fan just pointing at it obviously it's not doing a test right now so our recommendation here honestly is just use the motherboard cover if it actually has a heat sink uh or actually has a thermal pad touching this and is making solid contact with your drive and you'll find that you're not going to throttle now every controller is different you have you have faison controllers i'm not going to go through all the brands we have a bunch of different brands of controllers and they all are going to have their tolerances in terms of heat but what we've seen right here was that the nvme gen 4 specifically found in the samsung drive it was like 78 to 80 c before we started to see any significant slowdown and here's the thing 6600 to 5700 or so i thought about a thousand megabytes per second you wouldn't necessarily notice that in your day-to-day usage and this is also the worst-case scenario where we are moving a 64 gigabyte file constantly where you have a lot of sporadic things happening with your system it's not going to be under that much load but i think at the end of the day i'm just actually kind of shocked at how well the stock cover did on this particular drive now one thing to keep in mind is there are different size coolers so for instance i don't know how well you can see it on here this cooler is significantly bigger than the one above it so that could absolutely have an effect on the temperature but i think if you're using the stock covers that come on just about every modern motherboard these days and you have airflow moving over it some sort of airflow whether it be a fan case flow or whatever maybe even make a cover like a fan that blows directly on it you'll have longevity in your drive the temps will be perfectly fine in terms of throttling if it's not throttling it doesn't think it's too hot so you know you're within spec and parameter of that i honestly expected it to throttle sooner and i was hoping that by keeping it as cold as possible with the water cooler that that meant that we would get near that 7000 megabytes per second which is advertised that just goes to show until the initial throttling starts there's no curve there's no speed curve like like a so cpus and graphics cards they'll fluctuate that speed that megahertz based on the temperature curve this doesn't have any curve at all it goes full speed until it sees throttling and then it adjusts itself based on temperature so anyway this might be a dumb moment for many of you but this is something i've never actually tested and i wanted to to see how it goes have you guys noticed any issues with your ssds temperature wise and stuff if you didn't know hardware monitor can actually show you the temperatures of both your hard drives and your ssds i recommend looking at that so you guys can have an idea are your drives overheating you may not know some people aren't using the thermal pads on those covers so then what you're doing is you're insulating the drive away from any heat or airflow and then you could be creating a problem for your drive and you may not even know it so download some sort of monitor tool at least look at your temperatures and make sure everything is working as expected i expect you guys to also give this video a like if you appreciated it and subscribe if you're not it actually greatly helps the channel and we'd love to see all your beautiful smiling comments down below thanks for watching we'll see in the next one oh by the way i don't use telegram don't be an idiot

Channel: JayzTwoCents

Views: 460,315

Keywords: ssd, ssd watercooling, watercooling an ssd, nvme ssd, nvme, nvme temps, how hot can an NVME get, when will an nvme throttle, throttling nvme, corsair, samsung, samsung 980

Id: dDTH93dgulE

Length: 24min 0sec (1440 seconds)

Published: Sat May 07 2022