Hi Jay, physics major here. First of all, thank you for actually conducting the experiment. I don't doubt your results, but I have to correct you on your explanation. You seemed to be suggesting that there's a fundamental law of thermodynamics that implies loop order shouldn't matter. There isn't. In fact, the basic argument in favor of loop order mattering makes sense, though it depends on several parameters whose effects are hard to quantify. It is absolutely true that as the coolant fluid passes through a component which introduces heat into the loop, the fluid will get warmer and be less effective at cooling the next component it comes into contact with (and the opposite is true for fluid that passes through a component which extracts heat). The only question here is the effect size. That is to say, "will the difference actually be noticeable?" As you've shown here, the difference isn't noticeable, but it's not due to the laws of thermodynamics. It's due to the fact that the coolant fluid is passing through components quickly enough that the exchange of heat doesn't cause the temperature of the fluid at any specific location to deviate much from the average coolant temperature (though it does still deviate by some small amount). If the coolant were to be pumped through the tubes slowly enough that the exchange of heat resulted in a sizable change in coolant temperature, the difference in loop order would be noticeable. Bottom line is that the combination of fluid dynamics and thermodynamics required to fully understand cooling solutions is not at all trivial, even for people who understand physics, let alone armchair physicists who try to make proclamations without actually doing experiments. The fact that you actually did the experiment is more helpful toward answering this question than my physics degree.
Former Rocket Scientist (really) here. Excellent response, Jon. We could design cooling systems that were more effective, more efficient and far more expensive, but what is available today is probably already past the cost-benefit points an average person would consider acceptable. Jay's point is, I think, that in common usage today, it does not matter. And on that point, he is correct.
it's fantastic to see youtube users replying to this video with well-reflected arguments, and then continuing onwards to have a well-thought-through debate that serves to enlighten curious minds. i say, this is an amazing corner of the internet!
yes. this. the logic was okay, we tend to think that the liquid stayed on the block to absorb / transfer heat for a while before going somewhere else. it's like when you put water in a kettle, the water will heat up, then you put the warm water into another kettle, the kettle will warm up instead. however, in a pc the water just kept on flowing, it only matters if you check the temp on .001 accuracy or something.
Except Jay did say all that, without going into a bunch of physics. At one point he exactly said the rise in temperature accross components isn't enough to matter. So while you went into this long ramble to correct someone who wasn't wrong in the first place....
That's me currently, it's definitely not sad, currently I don't even know what kind of tubes to use even though my blocks and reservoirs are ready and epic. Every thing should be ready in like 2 weeks from today I hope.
The Firehawk EVERYTHING is ultimately air cooled seeing as eventually part of the system reaches air... however most of your heat is CONVEYED to the surface by liquid. So you are essentially liquid cooled to a liquid to air radiator. You're having a bad day if your heart becomes air cooled...
@RectalDiscourse Nahh man, I have the same PSU, and I can physically SEE the fan. The fan almost never spins up at all, it sits idle unless the system is under very heavy load.
But it’s good to have a powerful PSU! I run a Seasonic Prime platinum 1200 and it is platinum efficiency at 50% load. And I am usually drawing 600-700 watts of power while gaming with my single 2080Ti at 2,130 and 7980XE at 4.8.
I have run multiple loops in different configurations. Bottom line, it does matter, but not the way you think. No matter what order I ran, CPU temp never varied more then +/-4 degrees Celsius. What DOES matter is running the cooled water into your pump first! This is the main source of cooling for your pump. This can GREATLY extend pump life. I have seen pumps fail in less than 6 months because of this.
To be honest, I'd love to have the differences in running parallel vs serial(like when you have more then one graphics card) explained/tested and why you might go with one over the other.
I appreciate this a lot. I'm slowly expanding my loop and I'm wanting to add the GPU block without adding a second radiator temporarily. It will only be a couple of weeks and I don't really game or do any major overclocking anyway. My loop is purely for aesthetics. I just get the bonus of lower temps.
This was a great video. Being a car mechanic myself I've always wondered how a water cooling loop on a computer would be different compare to a car. I know the temperatures on a car are more extreme, but the same concept still applies. Great video here!
I kno guys but alot of people are complainin about him postin car vids .... its his channel and he even said he has 2 hobbies cars and computers .... so let him post whatever he wants
To clarify, I have no problem with Jay's car videos. I think they're great, but some or the videos he's posted over the past 6 months have been useless, like the beer loop. I get it, people requested it, but Jay is so much better than those low brow videos that don't teach us anything.
Still one of my favorite videos, and I share it with friends whenever they ask me questions about how to determine loop order or how I determined my loop order lol.
"Equilibrium: a state in which opposing forces or influences are balanced." "Entropy: a thermodynamic quantity representing the unavailability of a system's thermal energy for conversion into mechanical work, often interpreted as the degree of disorder or randomness in the system." Humm
I've done tests with loop order with a 2600k and a HD 5970. the GPU being first or last didn't make any difference. and an overclocked 5970 is possibly the HOTTEST GPU ever made.
Agreed.Thermal gradient in most liquids is minimal so steady state temps are what matter. This is dictated mostly by the cooling performance of the radiator.There is a prerequisite. You need sufficient rate of flow to push the fluid around the loop for the rad to do it’s job.Then the sequence, (to all intents and purposes), does not matter.
I´m sorry Jay, as much as I like you, as an energy and thermals profesional, I really think you´re partially wrong on this one, but it is all about context and math. In summary, by doing a CPU-GPU in series loop you´re just decreasing the 2nd component cooling efficiency, just because the conduction-forced convection heat transfer rate is directly proportional to the temperature delta between the cooling fluid and the heat exchanger(block). By even adding a fraction of a degree to the fluid from the first component, you´re directly decreasing the fluid-2nd component temperature delta, which decreases the heat transfer rate efficiency by the same proportion. Now, it is imperative to state the coolant has a masive heat capacity itself, it is capable of absorbing masive amounts of energy as heat every second 4.186J/g°C for destiled water for example which is the second best heat capacitor just under Amonia (4.6-7 J/g°C), enough to get the job done by far margin. Lets take the Ryzen 7 5800x as an example, with 105W of TDP, meaning it can deliver up to 105 Joules of heat every second. With such TDP, you´ll at least need around 25g of coolant every second (0.03L/sec) to incrase the coolant temperature more than 1 degree, and since water loops should be no less than 120L/H (0.03L/s,33.3g/sec) flow/mass rate, we gotta say the coolant fluid itself gives a lot of room to play between order and pump head/flow rate configurations. Having said that, again, topic goes back to its usual genesys, use cases and budget; for the vast majority, the coolant heat capacity will get the job done by itself regardless of the configuration, since they wont be squeezing power to its limits, but for those who need max power for long lasting periods of time and longevity, parallel loops (ideally redundant) grant the highest cooling effiency overall since the requirement is to have the biggest temperature deltas between coolant/components, the highest flow rate and least head resistance through the system. And since we´re talking about budget now, even a sub-ambient temperature and dew point asymptote gradient cooling system should be worth the investment for the one who really REALLY needs it. At the end, limits are defined, Ambient temperature, dew point function and TDps; everything else such as system configurations is up to how creative we can get.
If Vue fucked up your loop, that's on YOU. Not him. He said many times that Vue is a work in progress, and you REALLY need to clean your loop with a blitz kit before using it. He never said it was perfect. Blaming someone else because you're a knob isn't his fault. Fuckin' noobs, always needing their hand held.
Then you're the exception. :-) It makes sense, but this is one of those things that noob water coolers always seem to question, before doing the research.
Bartacus this video is straight misinformation. And Jay even dares to mention Thermodynamics... Testing here is extremely flawed as it's pretty obvious that radiator out will be cooler than intake, otherwise if the radiator wasn't doing anything there, might as well take it out altogether. What happens here is that you have extremely overkill parts (pump, rad, fans, blocks) for relatively little heat, that's why you don't see a difference. If you had a smaller rad or an SLI configuration, you would see together with higher component temperature, radiator out would also be much cooler than intake and loop order would be a factor
Adding a radiator will of course improve performance. You're increasing the amount of surface area which the fans are moving air through and thus extracting more heat from the loop. But if you use, say 2 120mm radiators instead of 1 240, the overall amount of heat removed from the system by the radiators remains the same, and temperatures will stay the same. This is because the entire system's heat removal capacity remains the same. You have to add heat removal capacity in the form of more radiator surface area in order to remove more heat from the loop.
@@ashesofempires04 If you were to add multiple radiators where water returning from each component it cools is brought back down to ambient temp, would it not make it so that each component is running cooler since each component receives water which is at ambient temp instead of water which was just moved through a hot component?
The fluids move at such a high rate of speed the entire loop reaches a equilibrium. Loop order doesn't matter. Size of radiator and fanspeed will make the biggest difference. In my loop I have two 560mm radiators. Fans on vs off in my loop makes a 25c difference.
@@corybolton7462 Just an overkill setup with 1080 ti SLI and a delidded 8700k. Once the RTX 3090 releases I'm switching to a small form factor build. Either an ncase m1 or sidearm t1. www.overclockers.com/forums/showthread.php/788563-Thermaltake-Tower-900-Build-Log
Love your content, Jay. What I really like is that it is always helpful & informational (instead of just trying to be funny, showboating, etc.). Over time yours has become my favorite tech/hardware channel. Specifically, your videos were super helpful as I was getting started with water cooling. Thanks for all you do!
at normal pressure: 0°C: Water starts to freeze. 100°C: Water starts to boil. 1°C difference is 1 Kelvin difference. Very easy if you compare it to fahrenheit ;)
What would be more interesting to know is the difference that radiator fan orientations can make in a case. For example if a front intake radiator feeds hot air to an exhaust top radiator, compared to both radiators blowing air in or out. Practically all Jay's builds I've seen have one radiator pushing hot air to another, which I've heard actually makes a noticeable difference in overall water temperature. Would love to see more tests on this.
Thanks for this, my take away lesson is to strategize tube trajectory to avoid bends as much as possible whether it be from CPU to GPU or vice versa.. Removes a little stress!
And to make the overall length of tubing as little as possible. Although in general most liquid cooling systems are rather overkill, limiting bends and tubing increases laminar flow and reduces wall friction. Water is an amazing energy absorber so won't reduce heat to any noticeable degree but takes substantial load of the pump increasing service life and saves a little power.
Don't know much about physics, but that part about the radiator being the cooling part of the loop is what made this whole thing make sense before he went to demonstrate it empirically
This - as he probably forgot, starting with Vista, the UI of Windows is rendered on the GPU, and culling does not take place (AFAIK) for windows overlapping or being totally obscured. Not sure on minimizing, but maybe (I have seen evidence of a GPU temp drop by about 1 degree or so by minimizing my browser, but that might be because most browsers have extra GPU use that, say, Office doesn't have (or maybe at this point it also does lol).
Thanks for this, I’ve never even thought about this but I’m new to the pc building community and it’s nice to pick up the right knowledge and useful tips straight away
You are absolutely correct. As long as the loop is flowing, it WILL cool everything. The fluid will not be there liong enough to really heat up is the radiator is working.
Jay remember this. Room temp usually 72°F ~ 22°C Great vid comparison with data. Thanks! Hope this helps. Cheers* *added tip: °C=(°Fx2)+28 Little trick to do in head and is pretty accurate. I know ppl, correct formula is C=(Fx1.8)+32, but multiplying 1.8 in head is something "don't wanna think about" even though can do, lol.
I totally agree with you on this Jay. I got a Core P5 and my loop runs from my CPU to GPU to Pump to Rad. I have zero issues with heat, period. Loop order does NOT matter. 6600k delided @ 4.8 GHz (1.3975 V) running 60°C max and Strix 1080TI O8G oc'ed further running max temps of 44°C. Thanks for all you do. Your the Real Tech Jesus....
Mike Lovecchio I don't believe theres a voltage that can be considered normal voltage honestly. That's just what my chip needs. Results will vary chip to chip. Anything below 1.4 is safe enough.
Oddly enough, I did have a build that had that from like 9 years ago. Had SLI GTX480s, motherboard block and a i7 930 with three 360mm rads. Damn that tower was heavy. If I wanted to get a bit froggy, I could have added two dual 80mm rads dedicated to cool the memory and hard drives.
What about 2 loops. Pump>gpu>cpu>rad and back to pump. Pump>gpu>rad1>cpu>rad2 and back to pump. Will 2 radiators will make a big difference on hard load?
If gamers nexus had done this test, there would be thermocouples in each leg of the loop and we could see the effect of lowering the pump speed (larger delta T across each component).
I understand loop order in a single rad setup does not matter however, it follows the laws of thermodynamics to come to the conclusion a greater temperature difference will increase the cooling effect. (More heat can be dissipated with a larger temperature differential). This is because there are two heat generators (A B) and one heat dissipation element (C). No matter how you order these ABC, ACB, BAC, BCA, CAB, CBA hot water from one component will flow into the radiator be cooled then be warmed by another component which then flows to the first component again. The following assumptions apply (CPU is 80deg C, GPU is 60deg C, ambient is 20deg C, at all times, heat dissipation causes temperatures to balance in exactly the center of the two temperatures eg 60 deg coolant with 20 deg ambient through rad => 40 deg coolant as a result ). What may be interesting to test is the following: CPU, RAD, GPU, RAD vs CPU, GPU, RAD, RAD. In the latter setup the CPU may heat the coolant up to 50 deg C, that then flows to the GPU where the coolant balances between 50deg (coolant) and 60deg (gpu) at 55 deg. This coolant then passes through two rads to reach back down to 20deg. VS In the former however, the coolant cools the CPU down to 50deg (20deg, vs 80deg) then passes through rad to become 35deg (20deg vs 50deg). The coolant then flows to the GPU and cools the GPU to 47.5 deg (35deg vs 60deg). The coolant finally passes through the final rad and reduces to 33.75deg this then repeats until the coolant settles at ~34.5deg with the same effect. This means the former setup reduces the temperature of the GPU by over 7deg. A considerable amount (~15%) difference
You would be right, but you did never take into account het heat cappacity of the coolant. To keep the GPU cool at 60 deg, water will not need to warm up all that much, certainly not with the fast flow of the water. So the coolant will not have a temperature in between ambient and component temperature, but rather just slightly above ambient, let's say 25c. It wil then maybe be heated up about 1-2 degrees by the GPU or CPU in the short timespan it is in contact with the component. This is not significant to observe in overal heating of the components.
I had thermal sensors in my loop back when I had a loop. Going through two GPUs and the CPU raised water temperature by 3 degrees under heavy load. It was good proof that in a loop with proper flow, the water doesn't heat up significantly at each block.
The best order for your loop is whatever is most convenient for your configuration. The differences between order is so minimal that it doesn't matter, and if you can get by with shorter runs (or more fancy runs) or more direct runs, go for that. I ran pump -> cpu -> rad -> res -> rad -> gpu -> pump for over a year with no problems.
Unfortunately the doubters will still doubt becasue you only used one rad. Should have shown them CPU > Rad1 > GPU > Rad2 vs CPU > GPU > Rad1 > RaD2 !!!!!
Thanks for the vid Jay! Can you do a video on whether running a loop in parallel is no worse than in serial as long as flow is ok? So I have something to send people when the say my loop sucks.
Loop order doesn't matter at all when there is only 1x radiator in the loop. The order only becomes relevant once you have 2 or more radiators in the loop :)
Yes thanks for asking that I'm wondering also if we have 2 rads if we should switching between them : Pump/res -> Rad -> CPU -> Rad -> GPU -> back to pump/res better than : Pump/res -> rad -> CPU -> GPU -> Rad -> back to Pump/res Or still doesn't matter ?
@@adrienguillemin4088 In both of your setups the CPU will show exactly the same temperature. However, in your 1st setup the GPU will run cooler, so that would be the best setup when running 2 rads ;)
I'm making my own out of oak, aluminum and plexiglass. I'm parting out a Koolance ERM and Exos2 for the controller and startup system and some of the trimmings and using al Alphacool XT45 1080 radiator, an EKWB 400mm reservoir and a Koolance PMP-600 pump. 9 Aerocool fans and 4 mcpctech fan grills for blowholes in the plexiglass on each side. I'm taking a lot of pictures of my current ERM system setup so I can repurpose the startup circuitry and controller in my homebuilt system.
Hi Jay! The order would matter only within first few minutes of the loop running when the thermal energy produced by the components and the energy dissipation in the radiator reaches the equilibrium. That means the second component will be first to reach the max idle temperature by a few seconds. Very interesting case for the precise measurements but completely negligible in a real life scenario.
temperature sensor are getting more inaccurate the further the temp is away from the temperature the sensor was calibrated for/with. Or Jay knows how to defy physics. Wikipedia : Silicon bandgap temperature sensor
Jay usually just looks at the thermostat setting in the room. Air temp can vary depending on the actual location of the temperature probe reading that temp as well as location and direction of air vents. As we all know, sitting directly under an ac vent is much cooler for example. A far more accurate way would be to place a temp sensor DIRECTLY in front of the radiator measuring the air temp going INTO the radiator. This could easily explain how something may SEEM cooler than ambient of you haven't ACTUALLY measured the ambient temp AT the radiator. However for this video, it's not really necessary because it was just to show the difference in radiator position and order of the loop.
The Celsius scale is very easy to navigate, Jay: 0 deg. - water freezes, 100 deg. - water is boiling. This is convenient for figurative perception. But on the Fahrenheit scale it is already difficult, so I have never seen either frozen or boiling alcohol in my life. 🤔🧐🙂
True story, Jay! The most impact on a system would NOT be component order, but indeed the routing and bends. Keep it more straight, simple, it will optimize temps.
Dude has great hair man. As man that's losing his hair, I always compliment my fellow man that has a full head of healthy hair. Luck of the draw I tell ya.
I've never done a custom loop, but I suspect that that depends on the flow rate/pressure head curve of the pump. Also, there is a conceptual risk that the parallel split will not evenly distribute coolant to each GPU, if the impedance of each path is different. If I had to guess, I'd definitely go in series.
A better experiment would have used two machines and two loops, each loop with one GPU and one CPU from each system. That way you can put load on one system and leave the other completely idle, and see how much each component in the idle machine increases in temp. If one increases more than the other, your claim is false. If they both increase by about the same amount, then your claim is true.
This was a great explanation Jay! I believed before that Loop order was true but when I replaced my flexible hose with glass pipe and changed the order i noticed that it didnt matter...... Great info for the newcomers on liquid cooling though!
I have been using Corsair 280 AiOs for 10 years and they worked just fine top mounted. Last week, my i7 6950x decided its time to go to digital heaven. I upgraded to i9 12900k and that AiO was no good, so did research on custom loops. This video was helpful as I thought order of things were critical and you have proved that not to be the case. Thank you. I have never cooled my graphics card and never found it necessary on my system. Today I installed all Corsair Hydro components into my 10 year old box which is a full tower, but now used 3 radiators all with push and pull fans for each rad and Corsair Hydro pump reservoir combo XD5. BTW I also used the water block JTZ edition and apologies as I had no clue it was related to your channel. I only got it as it's the latest in the series and read reviews that earlier blocks didn't help 12900k very much. My loop is currently in drip test mode, so have no clue yet as to performance. My loop order is pump, 140 rad, cpu block, 140 rad, 280 rad, then back to pump. I'm looking forward to seeing this work later once no leak system is ensured. Thank you again, your videos have been instrumental in getting my custom loop in place.
Fahrenheit really is the worst unit on earth. I mean imperial length and weight also are very "unscientific", but Fahrenheit is just a complete mess. Anyway, nice video!
For those of you that still believe this myth here is some science why it doesn't make any sense. Lets say your GPU the moment it gets under load goes to 100C. The fluid will then take the thermal energy because it is at a lower temperature. thermal energy goes from High to Low. (this is also why heat death is inevitable). the GPU then cools down to 70C. that fluid is not 100C, it is far less. The fluid has a high specific heat. This means it take more energy to raise the temperature. So the fluid lets say is 50C. it then goes tot he CPU. the CPU is temperature is HIGHER than fluid temp so it actually gives the fluid more energy. but if the CPU is lower than the fluid temp in theory the cpu will heat up but in practice it doesn't. this is because the block is design to facilitate thermal energy transfer in 1 direction. there is an imbalance of surface area between he cpu and the fins in the block. This on top of the fact that the fluid is moving so fast that a section of water doesn't have time to heat up the cpu.
It does matter... when you have small water flow. I tested my loop with 25 litres / hour (from 500l/h) and then it really mattered. Because you have so small body of water to heat up per second, so the water is actually hotter aftter exiting the block. I got 5°C higher at the exit of a loop cooling 300W.
First thing, excellent video and valid information; hopefully people listen. Secondly, to further reinforce Jay's points...order doesn't matter. The key here is transfer of BTU through materials and turbulent flow. As loop temperature rises, the radiator expels this heat to atmosphere and heat transfer continues at the rate of materials used (most common cases this is copper). As long as there is a delta between loop temperature and components, transfer will happen.
Thank you sir. I did see. a lot of general criticism as to how the loop should be set up, however looks like science proves the point. Get a 360-420 radiator and feed the liquid through as you indicated.
Thanks for the info. Doing a AMD 50th Anniversary Build with dual Radiators and was concerned about the order of the loop using the EKWB for the Anniversary Edition 5700XT and a R9 3900. Your the best Jay. BTW, I love 15 mins away from PPCS! They are the best!
I'm pretty sure that none will see this comment. But I wonder if the length / size of the loop could play a part. Or will the loop always find were equilibrium independent on how far the hot vs cold water needs to travel.
Thank God for this video. I had been searching like hell fpr a little insight on which way the loop is supposed to go. The only constant i see is that the radiator must be connected to the pump / resevoir. Other than that the order doesnt matter.
Very valuable info - even 4 years later. I have always said the same thing, order doesn't matter. But I had no stats to back that up (I am an engineer - I knew I was right without stats). This is a very good confirmation that I will share with the haters that occasionally cross my path.
The tempaturediffrence on my loop is 1°C - 2°C idling and 3°C - 5°C under full load, messured betwen Radiator in- and outlet. My rad is a Mora 3 and the flow is 100l/h (1,6l/h)
The main thing you want is that the radiator is before the liquid goes into the reservoir, so the reservoir doesn't get damaged, and as you said, from the reservoir right into the pump. Nice vid Jay.
12:22 If you have no patience, like Jay..watch from here. Error of margin is usually 1-2c, he says room increases as SoCal is HOT AF. It doesnt truly matter your loop order...just res feeding pump. Thank for all your hard work Jay and Nic!
And to think I installed a Corsair H100i V2 for an AMD 2400G. But, It's only for an upgrade to a 2700X in a few months. Great explanation on the cooling differences.
Just to add my own test which I think corroborates your conclusion JayzTwoCents, I ran OCCT medium data set on my machine (Ryzen 2700x, Vega 56, MORA-3 9x120mm radiator) with temp sensors right at the inlet and outlet of the radiator. After about 10 minutes, the delta between the two sensors was just .4 degrees C. I would think with such a big radiator, if I left it go for an hour or so the delta would shrink even further, but I ain't got time for that.
Did my first one today, did it by pump to rad, to cpu block back to pump and repeat... I thought of a car cooling system and went by that and then found I had a instruction manual telling me (pump to cpu, cpu to rad, rad to pump and repeat???) Thought it was odd but meh it still goes... and now I've seen you talk about it lol no worries at all and for my first crack nor bad... set up I got is a c360 thermaltake pacific soft tube kit, goes hard 👌
That setup would make it very easy to setup liquid cooling in an office situation. If you set it up right you could have a number of systems rads all blowing the warm air out a vent and keep the office temp much or stable. Also you could shut off the vent in cold weather and reap benefit for the warm air from your rads.
I would say also. Take care to put the rad before the pump or pump tank combination to avoid hot water to warm up the pump. It needs also to run cool and should be better to send cool water to the pump. So I think to keep the order rad tank pump is somewhat usefull.
![ILLSLICK - KILLSHOT REMIX [FULL VERSION]](http://i.ytimg.com/vi/RIK8RrqIAzE/mqdefault.jpg)
Hi Jay, physics major here. First of all, thank you for actually conducting the experiment. I don't doubt your results, but I have to correct you on your explanation. You seemed to be suggesting that there's a fundamental law of thermodynamics that implies loop order shouldn't matter. There isn't. In fact, the basic argument in favor of loop order mattering makes sense, though it depends on several parameters whose effects are hard to quantify.
It is absolutely true that as the coolant fluid passes through a component which introduces heat into the loop, the fluid will get warmer and be less effective at cooling the next component it comes into contact with (and the opposite is true for fluid that passes through a component which extracts heat). The only question here is the effect size. That is to say, "will the difference actually be noticeable?"
As you've shown here, the difference isn't noticeable, but it's not due to the laws of thermodynamics. It's due to the fact that the coolant fluid is passing through components quickly enough that the exchange of heat doesn't cause the temperature of the fluid at any specific location to deviate much from the average coolant temperature (though it does still deviate by some small amount). If the coolant were to be pumped through the tubes slowly enough that the exchange of heat resulted in a sizable change in coolant temperature, the difference in loop order would be noticeable.
Bottom line is that the combination of fluid dynamics and thermodynamics required to fully understand cooling solutions is not at all trivial, even for people who understand physics, let alone armchair physicists who try to make proclamations without actually doing experiments. The fact that you actually did the experiment is more helpful toward answering this question than my physics degree.
Former Rocket Scientist (really) here. Excellent response, Jon. We could design cooling systems that were more effective, more efficient and far more expensive, but what is available today is probably already past the cost-benefit points an average person would consider acceptable. Jay's point is, I think, that in common usage today, it does not matter. And on that point, he is correct.
Thanks for some science, my thermodynamics understandings were like... no Jay, that's not how it works despite how insistent you are!
it's fantastic to see youtube users replying to this video with well-reflected arguments, and then continuing onwards to have a well-thought-through debate that serves to enlighten curious minds.
i say, this is an amazing corner of the internet!
yes. this. the logic was okay, we tend to think that the liquid stayed on the block to absorb / transfer heat for a while before going somewhere else. it's like when you put water in a kettle, the water will heat up, then you put the warm water into another kettle, the kettle will warm up instead. however, in a pc the water just kept on flowing, it only matters if you check the temp on .001 accuracy or something.
Except Jay did say all that, without going into a bunch of physics. At one point he exactly said the rise in temperature accross components isn't enough to matter. So while you went into this long ramble to correct someone who wasn't wrong in the first place....
When I was building my custom loop PC I used to lie in bed thinking about different loop orders to send myself to sleep. True story. Not sad at all.
This was me last night haha
Dude. Same. About to be in bed planning my loop I'll build tomorrow.
That's me currently, it's definitely not sad, currently I don't even know what kind of tubes to use even though my blocks and reservoirs are ready and epic. Every thing should be ready in like 2 weeks from today I hope.
Hah, I been stuck in this "loop" for a few days.
This is me lmaoo and while I'm working
*our bodies are water cooled and no one moans about the order of that*
You should watch man vs wild before saying that.
Our bodies are air cooled...
sooo we don't sweat anymore...
The Firehawk EVERYTHING is ultimately air cooled seeing as eventually part of the system reaches air... however most of your heat is CONVEYED to the surface by liquid. So you are essentially liquid cooled to a liquid to air radiator.
You're having a bad day if your heart becomes air cooled...
This is great point. We also replenish our water hopefully daily.
Being someone thats about to do his first loop this saved me a massive headache thank you.
2 years later I'm now in the same position! To any other beginners, hard tubing loops are fucking hard!
When your psu is so powerful it doesn't spin under load 😂
I know that feeling i have a corsair Rm1000x and I have never seen it spin😅
My 1600 T2 doesn't turn on until it hits 800w. It only turns on when using SLI which is almost never.
@RectalDiscourse Nahh man, I have the same PSU, and I can physically SEE the fan. The fan almost never spins up at all, it sits idle unless the system is under very heavy load.
But it’s good to have a powerful PSU! I run a Seasonic Prime platinum 1200 and it is platinum efficiency at 50% load. And I am usually drawing 600-700 watts of power while
gaming with my single 2080Ti at 2,130 and 7980XE at 4.8.
Main reason why I run a 1600w power supply.
Jay with the Linus ad transition
My mind still fills in "TUNNELBEAR!" even though they switched to Private Internet Access.
Tuunelbeeeeaar
at first I was like: "No way, Jay please xD"
Yeah, I'm not a fan of it. Please stick with clean cuts, it comes across as more professional.
Except Jay seems a lot less annoying
I have run multiple loops in different configurations. Bottom line, it does matter, but not the way you think. No matter what order I ran, CPU temp never varied more then +/-4 degrees Celsius.
What DOES matter is running the cooled water into your pump first! This is the main source of cooling for your pump. This can GREATLY extend pump life. I have seen pumps fail in less than 6 months because of this.
those quick disconnect things are cool
and useful
and expensive
What is the company that makes them?
EK. www.ekwb.com/shop/fittings/extra/qdc
well there goes any hope of getting a bunch of those... not really worth the price unless you switch stuff often
You guys are badass. I search every question I have about PC building and your videos always pop up with answers presented in an entertaining way.
To be honest, I'd love to have the differences in running parallel vs serial(like when you have more then one graphics card) explained/tested and why you might go with one over the other.
I appreciate this a lot. I'm slowly expanding my loop and I'm wanting to add the GPU block without adding a second radiator temporarily. It will only be a couple of weeks and I don't really game or do any major overclocking anyway. My loop is purely for aesthetics. I just get the bonus of lower temps.
This was a great video. Being a car mechanic myself I've always wondered how a water cooling loop on a computer would be different compare to a car. I know the temperatures on a car are more extreme, but the same concept still applies. Great video here!
Now THIS is a JayzTwoCents video, wooo!
Celsian what does that mean as far as i kno of every video he posts is his videos .... so yes thank for postin that its his video captain obvious
I kno guys but alot of people are complainin about him postin car vids .... its his channel and he even said he has 2 hobbies cars and computers .... so let him post whatever he wants
To clarify, I have no problem with Jay's car videos. I think they're great, but some or the videos he's posted over the past 6 months have been useless, like the beer loop. I get it, people requested it, but Jay is so much better than those low brow videos that don't teach us anything.
Getting your computer temps in Celsius and your ambient in Fahrenheit .. NICE!
Still one of my favorite videos, and I share it with friends whenever they ask me questions about how to determine loop order or how I determined my loop order lol.
It is called Equilibrium.
Jondoe Smith it's entrophy
It's a nice german Band.
Tarun Kumaar Equilibrium is a german Metal Band ;)
Not really the water temperature technically rises in the blocks but it's so little that it doesn't really matters because the flow rate is so high.
"Equilibrium: a state in which opposing forces or influences are balanced."
"Entropy: a thermodynamic quantity representing the unavailability of a system's thermal energy for conversion into mechanical work, often interpreted as the degree of disorder or randomness in the system."
Humm
I've done tests with loop order with a 2600k and a HD 5970.
the GPU being first or last didn't make any difference.
and an overclocked 5970 is possibly the HOTTEST GPU ever made.
3rd Gen Guy so your the reason we have global warming lol
bbmatias22 My farts count
Perhaps after my rx 590 fatboy... god I regret that buy
HD 4870x2 enters chat.
Do parallel loop testing!
Agreed.Thermal gradient in most liquids is minimal so steady state temps are what matter. This is dictated mostly by the cooling performance of the radiator.There is a prerequisite. You need sufficient rate of flow to push the fluid around the loop for the rad to do it’s job.Then the sequence, (to all intents and purposes), does not matter.
I´m sorry Jay, as much as I like you, as an energy and thermals profesional, I really think you´re partially wrong on this one, but it is all about context and math.
In summary, by doing a CPU-GPU in series loop you´re just decreasing the 2nd component cooling efficiency, just because the conduction-forced convection heat transfer rate is directly proportional to the temperature delta between the cooling fluid and the heat exchanger(block). By even adding a fraction of a degree to the fluid from the first component, you´re directly decreasing the fluid-2nd component temperature delta, which decreases the heat transfer rate efficiency by the same proportion.
Now, it is imperative to state the coolant has a masive heat capacity itself, it is capable of absorbing masive amounts of energy as heat every second 4.186J/g°C for destiled water for example which is the second best heat capacitor just under Amonia (4.6-7 J/g°C), enough to get the job done by far margin.
Lets take the Ryzen 7 5800x as an example, with 105W of TDP, meaning it can deliver up to 105 Joules of heat every second. With such TDP, you´ll at least need around 25g of coolant every second (0.03L/sec) to incrase the coolant temperature more than 1 degree, and since water loops should be no less than 120L/H (0.03L/s,33.3g/sec) flow/mass rate, we gotta say the coolant fluid itself gives a lot of room to play between order and pump head/flow rate configurations.
Having said that, again, topic goes back to its usual genesys, use cases and budget; for the vast majority, the coolant heat capacity will get the job done by itself regardless of the configuration, since they wont be squeezing power to its limits, but for those who need max power for long lasting periods of time and longevity, parallel loops (ideally redundant) grant the highest cooling effiency overall since the requirement is to have the biggest temperature deltas between coolant/components, the highest flow rate and least head resistance through the system.
And since we´re talking about budget now, even a sub-ambient temperature and dew point asymptote gradient cooling system should be worth the investment for the one who really REALLY needs it.
At the end, limits are defined, Ambient temperature, dew point function and TDps; everything else such as system configurations is up to how creative we can get.
Thanks for dispelling another myth that people seem to never understand Jay. Nice work! Keep schooling the noobs!
Kinda like when he recommended Primo Chill Vue, and fucked up everyone's loops.
If Vue fucked up your loop, that's on YOU. Not him. He said many times that Vue is a work in progress, and you REALLY need to clean your loop with a blitz kit before using it. He never said it was perfect. Blaming someone else because you're a knob isn't his fault. Fuckin' noobs, always needing their hand held.
Bartacus im a noob. I didn’t think this was untrue. It just makes sense
Then you're the exception. :-) It makes sense, but this is one of those things that noob water coolers always seem to question, before doing the research.
Bartacus this video is straight misinformation. And Jay even dares to mention Thermodynamics... Testing here is extremely flawed as it's pretty obvious that radiator out will be cooler than intake, otherwise if the radiator wasn't doing anything there, might as well take it out altogether.
What happens here is that you have extremely overkill parts (pump, rad, fans, blocks) for relatively little heat, that's why you don't see a difference. If you had a smaller rad or an SLI configuration, you would see together with higher component temperature, radiator out would also be much cooler than intake and loop order would be a factor
I wanna see a loop with a rad before and after cpu,thus having a rad for each component in a single loop. Just for comparison.
Adding a radiator will of course improve performance. You're increasing the amount of surface area which the fans are moving air through and thus extracting more heat from the loop. But if you use, say 2 120mm radiators instead of 1 240, the overall amount of heat removed from the system by the radiators remains the same, and temperatures will stay the same. This is because the entire system's heat removal capacity remains the same. You have to add heat removal capacity in the form of more radiator surface area in order to remove more heat from the loop.
@@ashesofempires04 If you were to add multiple radiators where water returning from each component it cools is brought back down to ambient temp, would it not make it so that each component is running cooler since each component receives water which is at ambient temp instead of water which was just moved through a hot component?
The fluids move at such a high rate of speed the entire loop reaches a equilibrium. Loop order doesn't matter. Size of radiator and fanspeed will make the biggest difference. In my loop I have two 560mm radiators. Fans on vs off in my loop makes a 25c difference.
@@Ghettochild.2600 jesus 2 560 rads are you running nuclear power cores? collapsed suns?
@@corybolton7462 Just an overkill setup with 1080 ti SLI and a delidded 8700k. Once the RTX 3090 releases I'm switching to a small form factor build. Either an ncase m1 or sidearm t1.
www.overclockers.com/forums/showthread.php/788563-Thermaltake-Tower-900-Build-Log
Love your content, Jay. What I really like is that it is always helpful & informational (instead of just trying to be funny, showboating, etc.). Over time yours has become my favorite tech/hardware channel. Specifically, your videos were super helpful as I was getting started with water cooling. Thanks for all you do!
wow those GPU temps are really impressive . I know it's not in a closed case but nonetheless . One day I will maybe do it . Good work Jay
at normal pressure:
0°C: Water starts to freeze.
100°C: Water starts to boil.
1°C difference is 1 Kelvin difference. Very easy if you compare it to fahrenheit ;)
Fahrenheit:
0°F: salt water starts to freeze
100°F: temperature of the human body
0:56 Okay Linus, thanks for the uncanny transition from your topic into a sponsor xd
You learned a lot from Linus, especially the way of introducing a sponsor
Jay yeah, but learning to drop stuff like Linus....BAAAAAAD 😱
LinusDropTips, tips for dropping things yet keeping them safe at the same time :P
I don't think Jay picked up the dropping stuff habit from Linus - pretty sure that's a life long skill for both of them...
Except Jay doesn't suddenly change the sponsor and sticks to nordvpn haha (we will see)
This comment IST brought to you .... By tunnel bear
What would be more interesting to know is the difference that radiator fan orientations can make in a case. For example if a front intake radiator feeds hot air to an exhaust top radiator, compared to both radiators blowing air in or out. Practically all Jay's builds I've seen have one radiator pushing hot air to another, which I've heard actually makes a noticeable difference in overall water temperature. Would love to see more tests on this.
Thanks for this, my take away lesson is to strategize tube trajectory to avoid bends as much as possible whether it be from CPU to GPU or vice versa.. Removes a little stress!
And to make the overall length of tubing as little as possible. Although in general most liquid cooling systems are rather overkill, limiting bends and tubing increases laminar flow and reduces wall friction. Water is an amazing energy absorber so won't reduce heat to any noticeable degree but takes substantial load of the pump increasing service life and saves a little power.
Don't know much about physics, but that part about the radiator being the cooling part of the loop is what made this whole thing make sense before he went to demonstrate it empirically
OCCT is rendering the Graphs @JayzTwoCents
Thats the spiking
This - as he probably forgot, starting with Vista, the UI of Windows is rendered on the GPU, and culling does not take place (AFAIK) for windows overlapping or being totally obscured. Not sure on minimizing, but maybe (I have seen evidence of a GPU temp drop by about 1 degree or so by minimizing my browser, but that might be because most browsers have extra GPU use that, say, Office doesn't have (or maybe at this point it also does lol).
But why would it be *spiking* when the graphs are being continually updated?
0:56 doing it the cool linus way !
=)
Thanks for this, I’ve never even thought about this but I’m new to the pc building community and it’s nice to pick up the right knowledge and useful tips straight away
I was wondering about this. Very helpful to know order does not in fact matter (working toward first water cooled)
You are absolutely correct. As long as the loop is flowing, it WILL cool everything. The fluid will not be there liong enough to really heat up is the radiator is working.
Jay remember this. Room temp usually 72°F ~ 22°C
Great vid comparison with data. Thanks!
Hope this helps. Cheers*
*added tip: °C=(°Fx2)+28 Little trick to do in head and is pretty accurate. I know ppl, correct formula is C=(Fx1.8)+32, but multiplying 1.8 in head is something "don't wanna think about" even though can do, lol.
C = (Fx2) + 28
My room is at 80F
(80 x 2) + 28 = 188 C
oh yeah it's so hot
I totally agree with you on this Jay. I got a Core P5 and my loop runs from my CPU to GPU to Pump to Rad. I have zero issues with heat, period. Loop order does NOT matter. 6600k delided @ 4.8 GHz (1.3975 V) running 60°C max and Strix 1080TI O8G oc'ed further running max temps of 44°C.
Thanks for all you do. Your the Real Tech Jesus....
Is that the normal vcore for a OC'd 6600k?
I love my TT Core P5. I got the Tempered Glass edition.
Mike Lovecchio I don't believe theres a voltage that can be considered normal voltage honestly. That's just what my chip needs. Results will vary chip to chip. Anything below 1.4 is safe enough.
Marcus Colwell Aww... I wanted that one. But I got the Plexiglass version because it had a mail in rebate. I might upgrade one day tho.
Xclio you can purchase the tempered glass Kit. It comes with the 3 glass panels and the mounting hardware.
Loop order
>implying each component in my rig doesnt have a 360mm rad
even my rads feed intro their own dedicated rads
Oddly enough, I did have a build that had that from like 9 years ago. Had SLI GTX480s, motherboard block and a i7 930 with three 360mm rads. Damn that tower was heavy. If I wanted to get a bit froggy, I could have added two dual 80mm rads dedicated to cool the memory and hard drives.
What about 2 loops.
Pump>gpu>cpu>rad and back to pump.
Pump>gpu>rad1>cpu>rad2 and back to pump.
Will 2 radiators will make a big difference on hard load?
If gamers nexus had done this test, there would be thermocouples in each leg of the loop and we could see the effect of lowering the pump speed (larger delta T across each component).
"You know what does matter? Protecting your internet experience!"
Wow he's really turning into Linus
I came here to say this, but I knew in my heart that it had already been said.
I understand loop order in a single rad setup does not matter however, it follows the laws of thermodynamics to come to the conclusion a greater temperature difference will increase the cooling effect. (More heat can be dissipated with a larger temperature differential). This is because there are two heat generators (A B) and one heat dissipation element (C). No matter how you order these ABC, ACB, BAC, BCA, CAB, CBA hot water from one component will flow into the radiator be cooled then be warmed by another component which then flows to the first component again.
The following assumptions apply (CPU is 80deg C, GPU is 60deg C, ambient is 20deg C, at all times, heat dissipation causes temperatures to balance in exactly the center of the two temperatures eg 60 deg coolant with 20 deg ambient through rad => 40 deg coolant as a result ). What may be interesting to test is the following:
CPU, RAD, GPU, RAD vs CPU, GPU, RAD, RAD.
In the latter setup the CPU may heat the coolant up to 50 deg C, that then flows to the GPU where the coolant balances between 50deg (coolant) and 60deg (gpu) at 55 deg. This coolant then passes through two rads to reach back down to 20deg.
VS
In the former however, the coolant cools the CPU down to 50deg (20deg, vs 80deg) then passes through rad to become 35deg (20deg vs 50deg). The coolant then flows to the GPU and cools the GPU to 47.5 deg (35deg vs 60deg). The coolant finally passes through the final rad and reduces to 33.75deg this then repeats until the coolant settles at ~34.5deg with the same effect. This means the former setup reduces the temperature of the GPU by over 7deg. A considerable amount (~15%) difference
You would be right, but you did never take into account het heat cappacity of the coolant. To keep the GPU cool at 60 deg, water will not need to warm up all that much, certainly not with the fast flow of the water. So the coolant will not have a temperature in between ambient and component temperature, but rather just slightly above ambient, let's say 25c. It wil then maybe be heated up about 1-2 degrees by the GPU or CPU in the short timespan it is in contact with the component. This is not significant to observe in overal heating of the components.
I had thermal sensors in my loop back when I had a loop. Going through two GPUs and the CPU raised water temperature by 3 degrees under heavy load. It was good proof that in a loop with proper flow, the water doesn't heat up significantly at each block.
The best order for your loop is whatever is most convenient for your configuration. The differences between order is so minimal that it doesn't matter, and if you can get by with shorter runs (or more fancy runs) or more direct runs, go for that. I ran pump -> cpu -> rad -> res -> rad -> gpu -> pump for over a year with no problems.
It does matter.... cpu>>>gpu>>>res>>>pump will definately put pressure on your pump and cause wear and tear faster...
@@willis23jm1nope
Unfortunately the doubters will still doubt becasue you only used one rad. Should have shown them CPU > Rad1 > GPU > Rad2 vs CPU > GPU > Rad1 > RaD2 !!!!!
I'm here from the future, Jay... and NordVPN is a no go!
I don't think Jay recognizes the magnitude of the pun he made when he said "it's *cool* and all".
Thanks for the vid Jay! Can you do a video on whether running a loop in parallel is no worse than in serial as long as flow is ok? So I have something to send people when the say my loop sucks.
Loop order doesn't matter at all when there is only 1x radiator in the loop. The order only becomes relevant once you have 2 or more radiators in the loop :)
Yes thanks for asking that I'm wondering also if we have 2 rads if we should switching between them :
Pump/res -> Rad -> CPU -> Rad -> GPU -> back to pump/res
better than :
Pump/res -> rad -> CPU -> GPU -> Rad -> back to Pump/res
Or still doesn't matter ?
@@adrienguillemin4088 In both of your setups the CPU will show exactly the same temperature. However, in your 1st setup the GPU will run cooler, so that would be the best setup when running 2 rads ;)
This maybe the best jayz vid I've seen. Learned a lot here. Thanks.
Private internet access vs Nord VPN ? I can't decide .
Nord. By far the best. And not based in the US so no NSA compromising it.
What about multi rad setups? gpu -> Rad ->cpu ->rad -> pump/resivior ->gpu
insta-liked for the smugface in the thumbnail!
I'm making my own out of oak, aluminum and plexiglass. I'm parting out a Koolance ERM and Exos2 for the controller and startup system and some of the trimmings and using al Alphacool XT45 1080 radiator, an EKWB 400mm reservoir and a Koolance PMP-600 pump. 9 Aerocool fans and 4 mcpctech fan grills for blowholes in the plexiglass on each side. I'm taking a lot of pictures of my current ERM system setup so I can repurpose the startup circuitry and controller in my homebuilt system.
Hi Jay! The order would matter only within first few minutes of the loop running when the thermal energy produced by the components and the energy dissipation in the radiator reaches the equilibrium. That means the second component will be first to reach the max idle temperature by a few seconds. Very interesting case for the precise measurements but completely negligible in a real life scenario.
Yoooo that nordVPN transition !!! He just did it linus style !!!!!!
This is even more confusing than the time I locked myself in my car.
th-cam.com/video/grHc6MbpUJs/w-d-xo.html
Jay 70 F is about 20 C.
21, to be exact. So the GPU was colder than ambient. Que?
temperature sensor are getting more inaccurate the further the temp is away from the temperature the sensor was calibrated for/with. Or Jay knows how to defy physics. Wikipedia : Silicon bandgap temperature sensor
Or maybe it was closer to 68 F and he rounded up.
Yeah, I'm thinking it was 68. That's an average setting.
Jay usually just looks at the thermostat setting in the room. Air temp can vary depending on the actual location of the temperature probe reading that temp as well as location and direction of air vents.
As we all know, sitting directly under an ac vent is much cooler for example.
A far more accurate way would be to place a temp sensor DIRECTLY in front of the radiator measuring the air temp going INTO the radiator. This could easily explain how something may SEEM cooler than ambient of you haven't ACTUALLY measured the ambient temp AT the radiator.
However for this video, it's not really necessary because it was just to show the difference in radiator position and order of the loop.
The Celsius scale is very easy to navigate, Jay: 0 deg. - water freezes, 100 deg. - water is boiling. This is convenient for figurative perception. But on the Fahrenheit scale it is already difficult, so I have never seen either frozen or boiling alcohol in my life. 🤔🧐🙂
True story, Jay! The most impact on a system would NOT be component order, but indeed the routing and bends. Keep it more straight, simple, it will optimize temps.
Can't stop staring at his amazing hair.
Dude has great hair man. As man that's losing his hair, I always compliment my fellow man that has a full head of healthy hair. Luck of the draw I tell ya.
Hey Jay, is there a cooling difference when water cooling multiple gpus in series vs parallel?
I've never done a custom loop, but I suspect that that depends on the flow rate/pressure head curve of the pump. Also, there is a conceptual risk that the parallel split will not evenly distribute coolant to each GPU, if the impedance of each path is different. If I had to guess, I'd definitely go in series.
A better experiment would have used two machines and two loops, each loop with one GPU and one CPU from each system. That way you can put load on one system and leave the other completely idle, and see how much each component in the idle machine increases in temp. If one increases more than the other, your claim is false. If they both increase by about the same amount, then your claim is true.
This was a great explanation Jay! I believed before that Loop order was true but when I replaced my flexible hose with glass pipe and changed the order i noticed that it didnt matter...... Great info for the newcomers on liquid cooling though!
I have been using Corsair 280 AiOs for 10 years and they worked just fine top mounted. Last week, my i7 6950x decided its time to go to digital heaven. I upgraded to i9 12900k and that AiO was no good, so did research on custom loops. This video was helpful as I thought order of things were critical and you have proved that not to be the case. Thank you. I have never cooled my graphics card and never found it necessary on my system. Today I installed all Corsair Hydro components into my 10 year old box which is a full tower, but now used 3 radiators all with push and pull fans for each rad and Corsair Hydro pump reservoir combo XD5. BTW I also used the water block JTZ edition and apologies as I had no clue it was related to your channel. I only got it as it's the latest in the series and read reviews that earlier blocks didn't help 12900k very much. My loop is currently in drip test mode, so have no clue yet as to performance. My loop order is pump, 140 rad, cpu block, 140 rad, 280 rad, then back to pump. I'm looking forward to seeing this work later once no leak system is ensured. Thank you again, your videos have been instrumental in getting my custom loop in place.
0:57 Man, that was smooth, but you've still got a long way to go to catch up to Linus on the $hilling department, lol.
Fahrenheit really is the worst unit on earth. I mean imperial length and weight also are very "unscientific", but Fahrenheit is just a complete mess. Anyway, nice video!
Haha.....he said “huge load” haha haha 🤪🤣
With only 1 small radiator u are correct. With 2 big and a high end setup I bet the sound would come down if mounted correctly. And vice versa.
Good video. I have wondering if mattered if you have both on the same loop. Well done.
For those of you that still believe this myth here is some science why it doesn't make any sense. Lets say your GPU the moment it gets under load goes to 100C. The fluid will then take the thermal energy because it is at a lower temperature. thermal energy goes from High to Low. (this is also why heat death is inevitable). the GPU then cools down to 70C. that fluid is not 100C, it is far less. The fluid has a high specific heat. This means it take more energy to raise the temperature. So the fluid lets say is 50C. it then goes tot he CPU. the CPU is temperature is HIGHER than fluid temp so it actually gives the fluid more energy. but if the CPU is lower than the fluid temp in theory the cpu will heat up but in practice it doesn't. this is because the block is design to facilitate thermal energy transfer in 1 direction. there is an imbalance of surface area between he cpu and the fins in the block. This on top of the fact that the fluid is moving so fast that a section of water doesn't have time to heat up the cpu.
Water/liquid coolers always seemed like witchcraft to me. Man was never meant to tamper in that part of thermodynamics!
It does matter... when you have small water flow. I tested my loop with 25 litres / hour (from 500l/h) and then it really mattered. Because you have so small body of water to heat up per second, so the water is actually hotter aftter exiting the block. I got 5°C higher at the exit of a loop cooling 300W.
You moved the Rad, tests are invalid.
wobbly sauce To show loop order doesn't matter?
This was really helpful for me. After 10+ years I want to build a water loop and these type of video helps take care of some of my questions.
Infrared imaging camera can provide a pretty nifty visual representation of where the hot spots are in various setups.
The real question is, why do people state their opinions as factual when that's not how it works?
You're referring to a claim, not an opinion.
Because in a world where people are told 'opinions can't be wrong' the stupid reign supreme.
because people don't understand that their entitled to their own opinions but not their own facts.
Welcome to the internet, where everyone is an expert :)
That's how politics work now, actually.
First thing, excellent video and valid information; hopefully people listen. Secondly, to further reinforce Jay's points...order doesn't matter. The key here is transfer of BTU through materials and turbulent flow. As loop temperature rises, the radiator expels this heat to atmosphere and heat transfer continues at the rate of materials used (most common cases this is copper). As long as there is a delta between loop temperature and components, transfer will happen.
Thank you sir. I did see. a lot of general criticism as to how the loop should be set up, however looks like science proves the point. Get a 360-420 radiator and feed the liquid through as you indicated.
Great video, i was racking my brain about this. This will simplify my life during my current build for sure.
Thanks for the info. Doing a AMD 50th Anniversary Build with dual Radiators and was concerned about the order of the loop using the EKWB for the Anniversary Edition 5700XT and a R9 3900.
Your the best Jay. BTW, I love 15 mins away from PPCS! They are the best!
I’ve found putting your radiator after your pump does increase temperatures quite a bit... (but you have to run the stress test for a long time)
Finally someone covers this properly!
I'm pretty sure that none will see this comment. But I wonder if the length / size of the loop could play a part. Or will the loop always find were equilibrium independent on how far the hot vs cold water needs to travel.
Thank God for this video. I had been searching like hell fpr a little insight on which way the loop is supposed to go. The only constant i see is that the radiator must be connected to the pump / resevoir. Other than that the order doesnt matter.
Very valuable info - even 4 years later. I have always said the same thing, order doesn't matter. But I had no stats to back that up (I am an engineer - I knew I was right without stats). This is a very good confirmation that I will share with the haters that occasionally cross my path.
Good to hear you use Fahrenheit not C I'm English and still use Fahrenheit.
Great videos I get a lot of information thanks
The tempaturediffrence on my loop is 1°C - 2°C idling and 3°C - 5°C under full load, messured betwen Radiator in- and outlet. My rad is a Mora 3 and the flow is 100l/h (1,6l/h)
Good to know. Just received my CL SMA8-A. ONE DAY..... I too will build a badass custom loop.
The main thing you want is that the radiator is before the liquid goes into the reservoir, so the reservoir doesn't get damaged, and as you said, from the reservoir right into the pump.
Nice vid Jay.
NordVPN ad hasn't aged well...
Drinking game of death : take a shot every time Jay says CPU or GPU
Tlaloc = Blackout within 14 minutes
12:22 If you have no patience, like Jay..watch from here. Error of margin is usually 1-2c, he says room increases as SoCal is HOT AF. It doesnt truly matter your loop order...just res feeding pump. Thank for all your hard work Jay and Nic!
And to think I installed a Corsair H100i V2 for an AMD 2400G. But, It's only for an upgrade to a 2700X in a few months. Great explanation on the cooling differences.
Apart from the radiator size and fan speed, the water flow rate also affects the heat transfer.
Thanks Jay. Your series of water cooling loop videos has been very helpful.
Just to add my own test which I think corroborates your conclusion JayzTwoCents, I ran OCCT medium data set on my machine (Ryzen 2700x, Vega 56, MORA-3 9x120mm radiator) with temp sensors right at the inlet and outlet of the radiator. After about 10 minutes, the delta between the two sensors was just .4 degrees C. I would think with such a big radiator, if I left it go for an hour or so the delta would shrink even further, but I ain't got time for that.
I built a loop with a coolant temp sensor on the inlet and outlet of each component in the loop. 1 degree difference at most between any given sensor.
Did my first one today, did it by pump to rad, to cpu block back to pump and repeat... I thought of a car cooling system and went by that and then found I had a instruction manual telling me (pump to cpu, cpu to rad, rad to pump and repeat???) Thought it was odd but meh it still goes... and now I've seen you talk about it lol no worries at all and for my first crack nor bad... set up I got is a c360 thermaltake pacific soft tube kit, goes hard 👌
That setup would make it very easy to setup liquid cooling in an office situation. If you set it up right you could have a number of systems rads all blowing the warm air out a vent and keep the office temp much or stable. Also you could shut off the vent in cold weather and reap benefit for the warm air from your rads.
scarmenl Linus tech tips tried this several years ago watch their videos on it for a good laugh
I would say also. Take care to put the rad before the pump or pump tank combination to avoid hot water to warm up the pump. It needs also to run cool and should be better to send cool water to the pump. So I think to keep the order rad tank pump is somewhat usefull.