Gadgetory

what's up guys geez two cents here and it should come as no surprise to my long-term viewers that I am a huge advocate for water cooling it's it's not necessary but it's fun and it definitely allows you to get more out of your components by overclocking them and keeping them cooler keeping your components cool extends the lifespan but one of the most common comments I get on my videos especially from people who've not really done water cooling before is Jay you're you're doing it wrong you're putting things in the wrong order you should have a radiator in between your CPU and your GPU or you know your GPU going into your CPU is gonna heat it up and you're just putting hot water into it it's unnecessary and hopefully by the end of this video I can prove that what I've been saying all along that loop order for components doesn't matter with the exception of the reservoir has to feed the pump but other than that it doesn't matter if you go CPU to rad or CPU to GPU to rad or CPU to rad the GPU whatever it doesn't matter it doesn't matter let's why I might try and prove by the end of this video and hopefully I can stand by my my claims for all these years we know it does matter protecting your internet experience that's why Nord VPN has sponsored today's video and right now they are offering my viewers seventy seven percent off a three year membership and they are adding more servers across the world every single week in fact right now they are up to four thousand forty eight servers in 62 countries including the Middle East Asia Europe and the Americas now VPN stands for virtual private network and that is important to all of us because it keeps people from being able to spy on our information and Nord VPN doesn't keep any logs so there's no log whatsoever of what you are doing on the Internet when connected to Nord VPN so start taking your internet privacy serious with Nord VPN and save seventy seven percent off a three year membership by using offer code J's two cents at checkout see the link in the description for more details so what we're gonna be doing right now is a couple of different tests we are gonna see what our aisle temps are and this has been running now for a couple of hours actually and for about an hour and a half we've already done some load tests and we've let it come back down to idle to see what our temperatures are and things are pretty damn impressive so in terms of our baseline just to see where these components sit right now under idle conditions it doesn't matter what or order they are obviously under idle nothing's pumping heat into the loop we are sitting at 20 see idle on the GPU I don't if you guys can see this or not 20 see obviously we've got a some chill in the room right here it's only at about 70 degrees Fahrenheit in this room sorry I don't know if the Celsius is I'm so American I can't help that our CPU is idling at 27 C so as you can see things are running extremely cool this is just distilled water also there's no coolant there's no air no additives no anti corrosives we're not leaving this loop together it's just got a little bit pink dye in there so that you guys can see the fluid obviously a little bit better and it looks better on camera so that is our baseline now the order that we're going right here is we are going hump to CPU CPU to GPU and GPU back wait hold on how did I have this going in terms of loop order we are going out from the pump into the CPU out of the CPU into the GPU out of the GPU into the radiator out of the radiator back into the reservoir which is I think a very typical loop someone would run CPU gets cooled first GPU second than the red sometimes CPU then the rad than the GPU but this is our condition and these were our idle temps so what I'm gonna do first is I'm gonna load the CPU up we're gonna let it run for 10 minutes under OC CT if you guys have ever run OC CT you know that this can absolutely hammer a system and I'm running the medium data set small data sets just overheats the CPU just instantly and that's part of the nature of that test that'll be good for like ln2 testing but we're gonna be doing medium data sets what we're gonna be looking at here is to see under load what happens if we run the CPU into the GPU while the CPU is under load is a GPU temp going to come up well it probably will but not because of the fact that the CP was heating up the GPU but because of the fact that the CPU is heating up the loop as a whole remember my claims have always been and this is the way that thermodynamics actually works this isn't just a claim the total temperature of the loop is gonna be controlled by the thermal capacity of the radiator regardless of where the radiator is in the loop the rat is responsible for the temperature of the loop as a whole this is not like a car we have a thermostat closing off the flow to let it sit in the radiator longer it's just constantly cycling what I would have loved to have done has had temperature probes on each one of these you know outlets to be able to see what the coolant temperature is but unfortunately I don't have any of those sensors so we just have to go off component temp at the end of the day that's what actually matters now with the Vardar fans running at a hundred percent in this to 40 by 40 mil read I don't think the CPU is going to be enough to heat up this loop as a whole and to you know have there be drastic changes on the total temperature of the system but you can see our CPU cores maybe you can't see it's really small where our CPU cores went into the 50s and our GPU did jump up one see but I guess we're gonna see what happens at the end of the ten minute tests so we'll let it run for 10 minutes I think that's gonna be enough to equalize this loop and we'll see what the temperatures do okay so we've been running for over 12 minutes 12 minutes and 20 seconds now in our GPUs temperature remember it's under none under load at all when from 20 into 24 now on the surface that's when someone would say see see it heated up the GPU but we obviously have to reverse this test to see if the same if my claim is true that the temperature as a whole went up which means if we then go from GPU to CPU and then to radiator to reservoir and the temps do the same thing then you can see that well hey if the CPU was after the GPU how is it heating up the GPU but we noticed something here when we started to take a look at our temperatures look at this this is our GPU line right here every few seconds something is triggering the GPU to go under a little bit of load up to about 20% load so right about here we are sitting at 11 7 10 so it's like 15% 11% you'll notice as I move the mouse around to look look at the GPU usage the GPU has actually got a part in rendering the mouse so as we move the mouse at or more load but with it just sitting here we noticed these intervals of the GPU doing something I'm not sure if OCC T is pinging the GPU for information or if it's putting some sort of a slight load on the GPU it's possible I mean is GPU 3d tab right here but yeah it's it's not doing anything with the GPU whatsoever in this test so this could very well be what caused the GPU temp to go up and not the CPU at all because the CPU has been sitting steadily at 6261 C so if you look at our CPU temperature line right here you can see the curve where it came up and then this is the test kind of doing its thing and it actually came back down slightly so our temperatures have been extremely consistent on the CPU which means that our loop temperature is equalized and it's not getting any hotter that means we have not exceeded the cooling capability of our radiator but that's why how well these quick disconnects on here because now I can rearrange the loop how I want it after I go ahead and stop this test and let the temperatures come back down to normal before I do that though I want to point out when I stop the test you can see this interval pinging on the GPU stopped it's now pretty consistent right here at 0 or 1 so I again I don't know what OCC T is doing I didn't want to use prime95 and I didn't want to use a 264 because I don't have a key for it right now so OCC T I just want to use it to get heat in there but obviously it's doing something with the GPU but our temp though did come back down to 21 C already in the last like minute or so so this is why we're using quick disconnects obviously because we can just push the button and then check that out we got what one little drop of coolant right there I like to use OCC T because as you can see it literally hammers the CPU we've got the order now switched around so just like before we are going to start our test we're gonna run it for at least 10 minutes to let the temperatures kind of go up my prediction on this is that we're still gonna see 24 C on the GPU which then would be good enough to just say hey you know we are we're done but then adds a new GPU tests everyone knows GPUs put way more heat into a loop than a CPU so as expected we've been running for 12 minutes and 30 seconds the same as before look at that 24 C on the GPU and our CPU is sitting at 62 C sixty-five of course it's fluctuating but you can see the curve is definitely like flattened out here are those were strange intervals again but Phil and I were talking we don't think this is enough to actually make the GPU temperature go up because the core speed doesn't increase is just a random like 2 or 3% up to 10% worth of load for like a split second so what we're gonna do now is we are going to go ahead and let's stop we're gonna stop this test and we are going to let the temperature equalize now we're gonna let it come back down to its idle temperatures and then we're gonna run a GPU load to see what happens to the CPU how much does a CPU come up and then we're gonna switch the order back to the way it was where the GPU is not directly feeding the CPU like it is now and see if the temperatures have changed I think based on the test we've just seen though you guys know what to expect we'll probably see this the CPU temperature come up because of just the increased temperature of the loop as a whole Wow that's stupid wire and then we're gonna run a GPU test we're letting the GPU loop heav'n benchmark and get nice and hot fine it's max temperature record the CPU temp versus idle and then we're gonna switch it back to where the CPU goes into the GPU and do the same you know test again based on what we've already seen I think you guys can can kind of already form a hypothesis and what's gonna happen we're gonna see the CPU temperature come up slightly but I think we're gonna see it come up slightly when we reverse it as well because again the entire temperature of the system as a whole is coming up so I guess a good testament to the fact that we have a good controlled environment is once again we let everything sit our CPU hit 28c and our GPU is back down to 20 see idle yeah those are pretty impressive temperatures I think given the fact we only have a single 240 read something else I should have probably mentioned before all the CPU tests it is overclocked to 4.3 gigahertz on all eight cores at one point three five volts so we were pumping more heat into it on purpose so we're gonna to let heav'n run now for however long it takes for the GPU temperatures to stop increasing so this might take longer than 10 minutes so what we're gonna actually be checking here is again our max GPU temperature I think we know that's gonna be the same no matter what we want to see what our max CPU temperature average is under loads remember this is still putting a slight load on the CPU not a huge load but a slight load we want to see if this temperature comes down because member we currently have it feeding GPU to CPU back to radiator so then if we switch that order we go CPU GPU to radiator does the overall CPU temperature come down and then I guess vice versa does the GPU temperature come up with the CPU feeding it based on our previous test which was on a smaller scale I think you guys already know how this is gonna turn out but if I don't actually complete the test then it gives you guys room to kind of challenge the results I guess you guys are gonna probably challenge it anyway but that's okay I've been doing this now for like 15 years and I already know what the results are going to be but I wanted to at least give you guys a video that you can reference when you see people talk about this on forums and read it and stuff to be like hey go and check out this video done by this guy who knows nothing in a very unscientific way that proves a point it's we were running for about 15 minutes maybe a little more you can see our GPU temperature is sitting at 33 C it's been going between 32 and 33 and that's pretty much where it's maxed out but the CPU temperature this is gonna be kind of hard to gauge the maximum templating so much right depending on what the scene is doing how much FPS because right now it's 150 fps so obviously this is gonna fluctuate but I guess what we can look at here is our GPU max temp was 33 so we know that but I guess this isn't even that helpful though because it still shows 72 max which from our previous test I forgot to clear this so let's just go with I mean I guess if it goes any hotter than like 47 48 C I don't know I I don't think anything's gonna change to be honest with you but now we have to flip that order around after we let it cool down so much waiting in doing these tests I mean it's cool and all it takes forever to just like equalize let it cool down for you guys it's like only a few minutes for us it's like hours worth of stuff so I'm running for about another 15 minutes and look at the temperature came up about one and a half see because it'll go between 34 and 35 where before I went from 32 to 33 sitting mostly at 33 so we came up about 1 1 and 1/2 C temperatures are the same GPU temperatures are like I said up about 1 C CPU temperatures are doing the exact same thing as before so doing the GPU test now again the same result with a difference of about 1 C and actually think that that might be the temperature in the room came up ever so slightly is getting pretty warm here in California but as you can see right here the tests don't really lie the order of the components like I've been saying all along make absolutely no difference to the temperature the radiator is going to control the temperature of the loop as a whole so if you were to measure the temperature that's going into the radiator and the temperature of the fluid coming out of the radiator you would find they'd probably be one or two C difference at the hottest and the coolest point of the loop and that is not enough to make one part heat up another what's heating up is the fluid as a whole well that means that you don't have to worry about trying to hit a particular loop order it means that you can actually run components in the order that makes them cleanest route if that's what you're looking for unless you want a really busy loop that looks a lot like this you could then cross over tubes and maybe make that look cool but don't rack your brains over trying to find the right loop order just have your reservoir feeding the pump and everything else just whatever is the cleanest and easiest to route so I'm going to put this to rest I'm not going to revisit this topic anyone that wants to basically make the comment of hey you know your loop borders wrong just share them this video if you guys have seen any one on forums claiming hey you can't do that you're gonna heat up your CPU with your GPU if you go that order share them this video so that once and for all people will shut up and not pass around false information physics art physics they don't lie guys I'm gonna go thanks for watching and putting up with this if you've already known this then no surprise to you maybe you learned something here today if you did once you give this video a like and now I've gotta tear this monstrosity apart thanks for watching guys and as always we'll see you in the next one you really want me to just connect one while it's running but I'm not gonna but now