Is This Water Block Really Worth $200?
Vložit
- čas přidán 9. 06. 2024
- All information about Hetzner:
derbauer.hetzner.com/en/0823/
---------------------------------------------------------
Support me on Patreon:
/ der8auer
---------------------------------------------------------
Music / Credits:
Outro:
Dylan Sitts feat. HDBeenDope - For The Record (Dylan Sitts Remix)
---------------------------------------------------------
Paid content in this video:
- Hetzner Spot
Samples used in this video:
- Corsair XC7 RGB ELITE LCD
Timestamps:
0:00 Intro
2:12 Hetzner Ad
2:54 Flow Sensor
4:41 The new XC7 LCD
6:43 Flow-Test
7:27 Setup
8:30 iCUE Features
9:10 Test Results
10:35 Teardown
12:29 Built-In Temperature Sensor?
13:43 Structure in Detail
15:21 Conclusion - Věda a technologie
Since thermistor is somewhere on the periphery on the cold-plate they can market it as built-in because it's integrated into the block
i was under the assumption that the coldplate itself is also acting as a thermometer (changes conductivity across it as the temperature changes) but that would conduct to the IHS and the ground plane so i guess this is the closest we can get to 'cold plate' temp measuring
Correct. It also seems rather inefficient to place the thermistor any closer as it would interfere with either conduction or cooling flow or both. In the end of a X vs X comparison, the ultimate comparison is the final temps of either the CPU or GPU depending on which you are testing.
Yeah, from a semantic English standpoint "Built-in coldplate temperate sensor" isn't inaccurate; it's built-in and measures the coldplate.
Nice blocks and performance but the most chill on the table is the cat.
I appreciate the inclusion of a flow meter. Thank you very much.
I like this channel a lot. As to the QD's (I use the latest Koolance QD3 version). I find it convenient to orient QD's such that QD junctions are arranged to be uniformly flow oriented. I find this helpful for maintenance and to ensure that I reconnect the loop correctly. One possible convention is the QD;s are arrange such that coolant always exits the male connector and always enters the female connection. I have a many such junctions and found this helpful. I construct a loop from scratch, I tend to create a diagram indicating the orientation of the various connectors. This helps assess what connectors may need to be ordered.
Thanks for the in depth testing. Need more of this.
Yeah, the alalysis of the design of the flow plate, seing on microscope the level of details in machining and such gives such a good idea of the quality of the product. Most useful review
Given that you have a flow sensor, and you know the flow of a given test loop, you could create a flow drop percentage chart (flow impedance chart) for any individual component that can be added to a loop, and then that could be a usable tool for users who want to buy components, or could at least explain why certain parts perform differently to others
block looks pretty slick but i wouldnt use any part with acrylic threading, personally
Agree to that, even tho Acryl is all good in the hood, used it a lot in aquariums but there is a but when it comes to computer :)
I feel like acrylic is just waiting to break on you
Quality acrylic should not crack under hand tightening. People who crack acrylic when tightening a fitting are usually using a hand tool and applying way too much torque.
agree@@tomppeli.
agree to that too, and thats facts! Acrylic should not be tightened with too much torque, same with the lid for the oilfilter on a motorbike, you cant just torque the fuck out of it because the threadings are gonne break on you and then you get a big problem with the preassure and thats not even acrylic but aluminum threads
@@paskowitz
Flow rate also depend on how pump is regulated, if pwm controls pump power, then adding restriction should lower flow rate measurably, if pwm controls impeller speed, then when adding restriction the pump will pull more power and flow rate shouldnt change signicantly up to a point
I was also thinking this, for an accurate measurement you'd probably also want to include pump RPM and power draw.
@@TheAnoniemo and a lower powered pump. i think he's got two of those big pumps running. if you're only using one of the smaller mcp355 style pumps it may have a bigger impact?
I think this is assuming speed is fixed flow which is only true for displacement pumps, not centrifugal ones like the DDC and D5?
pump is regulated by temperature and in his flow tests there is no temp load so the pump is at the same speed
Wow are you really saying that the flow rate depends on the flow rate? I learn so much from youtube comments.
Quick side note in regards to mounting this water block in any direction. Yes, you can, however, you won't be able to rotate your screen. iCUE does not support screen rotation so you're stuck in 1 position, which is the one with the tube mounts on the right side (between the block and RAM). Corsair said they're working on on a potential solution, but it's been a few years now and still we've heard nothing from them since then.
I'm in the process of shopping for CPU blocks, so the overall comparison you did + adding to it as new blocks come through has been super helpful. I'm still waiting to see what direct die blocks come available for Intel 13th Gen (der8auer micro?), but right now running a GPU only loop looks a little funny haha.
Do you know if this can be used for direct die 13th gen cooling?
I got the non-LCD version of this block, replaced my corsair xc8 JTC edition, same setup and thermal paste and dropped 5 degrees C, I think it has more to do with the mounting solution in my case as I am on AM5 and my old block was designed before AM5, either way I'm loving the performance of the block, and it looks great in my rig!
If the block is oriented with both G1/4 fittings on the bottom, is the left or right one the inlet?
@@Personalinfo404 I have mine oriented with the fittings on the right, and my inlet is on the bottom then. So it should be the left side if you have your fittings both on the bottom
Would be interesting to see those angled connectors impact on flow rate, I remember since the early days when I started with watercooling in about 02, 03, many many ppl on forums (sound old right) were telling that those are really restrictive... Maybe that had an impact back in the days when not all blocks were that low flow restrictive and I can still remember a guy telling that running D5 from koolance on 12V at 600l/h is a waste and you need to pull 24V from ATX connector to get 1200l/h :DD
Excellent Explanation as always.
Hey Derbauer, what Corsair means is that the temperature sensor is built-in, measuring the the cold plate temperature. Not that the temperature sensor is built-in to the cold plate.
Hi Roman, thanks for the video. I'm not quite sure how given your testing process and where you put the temperature probe in the contact frame, however do you think it would be possible to do a direct die test comparison here using the same methods? I'm curious to see how the velocity2 with direct-die option that you already tested in a previous video fares in this setup / comparison.
"built-in cold plate thermal sensor" tells me that it's supposed to measure the temperature of the cold plate. You know, like a useful measurement point in between the CPU temperature and the water temperature. But the way it's set up, it's not going to be an accurate measurement at all. It's not a water temp sensor, because it will conduct some heat from the cpu. It's not a cold plate sensor because it's like 3cm away from the heat source. That means it's useless marketing garbage, just like everything else Corsair makes
Came for the cat. Stayed for the tech review. Danke.
Well I hope this sets a trend, losing a display like this is one thing I miss from moving from my old AIO (the ROG RYUO) to full custom cooling. Love if EK started putting these on monoblocks.
thumbs up for the kitty!
15:25 i love this Bond Villain style summary...
Good review/breakdown 🤙🏻
Random question tho, why have the male part of the QDC on the block? All the pressure from the disconnecting is put on the male end, the female end pushes it away, I always thought that’s a way to eventually crack something cuz of the pressure on the blocks/plexi/etc. just a thought.
Built-In cold plate temperature sensor is correctly advertised. It does not advertise that it is "built into" just "built-in". Putting the sensor directly into the plate would create an inconsistency in thermal spread due to a void created by the pocket drilled out for the sensor. Sorry if I'm coming off as a grammar goblin, its an OCD thing.
Hmmm on the notion of measuring the cold plate, could it be extrapolated through the use of:
* The temperature probe that's as near as the cold plate and effectively measures the water temperature
* Use the temperature of the CPU cores and assume that the IHS is mounted with the prescribed mounting pressure
Then use a parametric temperature model of the cold plate, that is gathered through the use of thermal simulations/material properties and real-world simulation. Thus be-able to proactively estimate the temperature of the cold-plate real time. I think the use of such data can be used to more effectively tune the necessary flow rate and fan speed.
On the other hand, it may be overengineering as the thermal conduction of heat ,for such a concentrated component such as a CPU is constrained by physics.
Hey Derbauer, it would be really neat if you did a "here's some great starting items that are good bang for your buck items for beginner watercooling"! I've been wanting to give my hand at doing a full custom loop but honestly i don't know if I'd want to spend nearly $1000 on something like that just yet.
Bykski and/or barrow for everything but quick disconnects.
Well it's not supposed to be cheap if you want to fit it, and forget about it. It's expensive to get into, because you're paying for reliability. Ease of installation seems to be a secondary consideration for even the best open loop custom water-cooling solutions, so don't expect much from entry level! And expect to make mistakes.
Just get an AiO to start, with how big the coolers of modern GPUs are, they are cold enough. Or get a Hybrid card if you have the room in your case. You should plan your build with watercooling in mind, like how many rads you can have in your case vs how many you actually need, most Waterblock comes down to which one you like more visually as they all perform good enough, 1 - 5 degrees from eachother, don't cheap out on watercooling stuff either, no need to buy the most expensive either. Don't need Dozens of fitting nor hard tubbing either. Though if you are just cooling the CPU with it, jist get an AiO like I mentioned, they are good enough, cheap vs custom, easy to install, no need to refill unless it is a crappy one... just set and forget kind of deal. Make sure the pump block is not the highest in the loop, same with custom, always have the pump at lowest part or right next to the water reservoir and don't mix metals either if you are doing custom, copper block with copper rads or aluminum block with aluminum rads, aluminum parts are usually cheaper... though is up to you, EK starter pack is aluminum based and is very cheap, decent startin point, don't know if that is available yet as it has been a while...
I feel like a zig-zag pattern would be more ideal for coldplates. But that is more difficult to mass-produce. This time of thin structure is easier, since you can just make a 'shaving comb' to manufacture the fin-stack once you've milled down the surroundings to right depth.
How ever, I do believe their water distribution is quite good. But I think their current design is more suited for lower flow rates than higher. A higher flow rate in such a non-restricting CPU block simply harms its performance, I reckon. It would be interesting if you added a 'valve' before the flow sensor. So you can play around with the flow rates and see what happens if you start on low flow rates and increase.
I understand that shaved is less thermally efficient than milled for fins, some change in the electro-mechanical structure under shaving.
@@MyklCarlton I am taking a bit of inspiration from the Aquarium hobby and how people look at the use of UV bulbs. Light transmission and Heat Transmission isn't too different. Generally, in Aquariums, you want slower moving water across a heating or UV bulb because it provides more time for the energy to attach itself to the water column. In Aquariums, the complexity is less but the flow is generally unhindered through the components. The current design of this Corsair block is similar in that it doesn't really restrict flow.
Therefore, I believe a slower moving water across the CPU block might be better for it. In fact, I believe it would be better if there was a slower moving water current over the CPU with a pass-through, that just returns to the radiator. Basically, high flow through Radiator, Low flow through CPU.
Would like to see perf differential of these blocks with higher vs lower flow rates
Is the inlet/outlet of the flow direction to get about 4 "chiplet" coverage? As chiplets won't generally have high temps in the centre, unlike traditional chips?
Another advantage to that mounting type also is that you can change it from say intel to amd without having to drain the loop if you're not running QDC's for that :P
Although depending on when you changed your coolant last time, it might be time to change it... I have mine in for ~1,5y and starting to get yellowy-green (definitely needs to be switched out...)
@@cromefire_ I run ZMT tubing typically and change mine every 2 years on average and it's fine, though if you run clear soft tubing this isn't recommended, I'm more thinking as I change hardware frequently for fun to play with a new platform so that's actually pretty handy for me haha
What is your take on placement of intake/outtake and how it loops vs where the cores are? Potential differences would be something I would be interested in.
Make sure to ask Corsair before you auction it
Don’t forget to keep the names of the buyers
Ouch 😂
Ohh no.....ohhh nooooooo
(lets make @sparkymotive comment the top comment)
Damn.... brutal, but deserved
😂
I'm surprised Corsair hasn't taken Aquacomputer's business, they already have desktop software so the hard part is already done. Temp sensors are dirt cheap, they could put one at water inlet, one at outlet, and include a few air temp sensors, and feed it all into their USB hub. Mid-tier motherboards don't include a temp sensor input anymore so Aquacomputer's market got bigger.
With the latest hardware using so much power it seemed like water cooling might start making sense, but instead of innovating they've just raised prices. Rather than do maintenance on my loop I recently got a $40 Thermalright air cooler and my 13900K gets 40K Cinebench vs ~40.5K on water.
Any update on the Thermal Grizzly Direct Die Waterblock for 13th gen?
I recently bought an XD7 RGB. When I took it out of the box to look it over, I noticed an extra port that I had not seen in any pictures or videos. It was not mentioned in any of the included documentation. I emailed Corsair and they sent me an updated user manual that marks the extra port as a "temperature probe port". They are not great at updating the manuals.
A minimal flow rate can be estimated based on the max sustained CPU heat output, the heat capacity of the block, the heat capacity of the water in the loop, the radiative heat of the cooler equipment, the ambient temp, and some other variables I am forgetting!
Or find it empirically 🎉
Would they perform differently with more head pressure or flow rate? You might have hit the sweet spot where they perform identical with the rate you're running.
I just finished a build with this and didn't even think that one should be inlet and the other outlet. Did you figure it out after taking it apart? How big of an impact would there be if they were switched?
about the noise from the pump, in my case the Hydro X XD5 I found out that some higher speeds made less noise, there's a question of resonance both inside the pump-res and mounted on your pc I'm sure...anyway, instead of running it at mid speed which is the worse for me too try a higher rpm desktop use I'm around 2000rpm but for "silent" gaming I run 3800rpm below that I can hear it, it's the beauty of using corsair commanders and icue...open your side panel, set speed to percentage and slide the line up and down until you find a speed that sounds right for you, I also did that with my 10 fans so am using weird rpms like 1537
Are research job made into a tech video, you have to love this!
So how long until we see a der8auer silent pump with build-in water temperature meter and flow meter?
For future videos, it might be a good idea to buy/build a dummy heater so that you can increase the precision of the heat load being put into the block. I think that Peltier modules might be an economical way to do this, but you might have to strap a heatsink onto the cold side to give it good access to fresh heat.
I forget how much Steve paid for the one GN uses but I'm sure he would be happy to help Roman if he decides to get one too.
Can we use a very fluid oil instead of water , to avoid having a conductive liquid near electronics in case of a leak ? and the performance will be worse or better ?
The thermistor touches the coldplate and not the water. It's sensing the cold plate temp. It's not sensing in the middle, but that's impossible without interfering with the cooling. What they did is ideal. I dunno what you are expecting them to do.
It will be too slow to run pump speed correctly, and probably too quick for fan speed. If it were closer to the center of the coldplate and bonded to it properly, you could get a delta T vs the die to confirm a good mount, but as it is it does nothing. Meanwhile, if you want a sensor for fan speed, putting it at the radiator inlet is ideal.
no.
One thing you might want to try and figure out is how much your flow sensor impedes flow. 150l/mn is prob not the kind of flow where it matters a lot, but it might be interesting to compare a block's performance with and without the sensor in the loop
That is a super nice looking water block.
I have a 14900K stock with no lapping. Would it be better to run the Velocity2 without a contact frame or something like the Heatkiller with a contact frame? I see EK offers a contact frame for the velocity2 but at a hefty price. I should mention I have never done a custom loop I have always ran AIO's and I'm running a 1/4 HP chiller with no rads or fans in this new setup. I guess I'm asking what block would work well for me? I'm still not sure on what pump to use because the chiller requires 1000 L/H and to get that out of a D5 would be too loud if at all possible from my limited understanding.
Hi derBauer as you say there are no flow indicators for the block could you show the difference for the block if it was wrongly connected. Also can you show how many Millilitres of water each block holds - that will be a good indicator of the thermal capacity of the block.
Hi, quick question, planning to delid my 13900kf to get better temp, current setup with asus z690 glacial and thermal grizzly contact frame, do i need anything else to fit the delided cpu to the motherboard waterblock ? Thank you
Need to measure the resistance/load on the pump motor when testing. The flow can stay the same but the load on the motor could be much higher?
Actually the Finnish site is not inside the city limits of Helsinki. It's located a bit north from Helsinki in the city Tuusula.
Will you be checking out the new optimuspc CPU block for am5 that costs even more than this one?
My personal flow is between 1-1,5 L/min with 3 thick 360 radiators and it's more than enough. Zero noise and I have a 3090 and 13700k with very low temperature.
Heya, a thought that keeps coming to me, has there been any advancement in what fluid people use in liquid based cooling? Or any changes to immersion cooling ? :)
Immersion cooling is definitely seeing advancements. There is a cloud provider in the UK called PeaSoup that uses immersion cooling to cool their servers. They don't use mineral oil but a new kind of dielectric liquid.
immersion check 3M's novec, I think the channel host has shown some demos of it as well. It's mostly datacenter targeted.
Great video
Wow, those labeling (flow direction) and instruction (GeForce 7900 XTX) mistakes seem rather glaring coming from a company as big as Corsair. Thank you for the review!
when you pay for aesthetic you don't pay for performance, always remember it
Does this use corsair's new connector? Or just standard ARGB and pump header connectors... Might be important to include that kind of information in the review
Maybe, the difference is seen with a little higher flow? The changes might need that to do better.
For performance I have i my pump to bairly audible. When I stream I focus on a flow rate of around 100 to 105ish. Find that to be enough for cooling when streaming and it's whisper quiet. Find these days that the performance you get from components these days it doesn't matter much with them temp undervolting and and clocking...
Also will note I have a massive amount of surface area on my rads x2 480 alpha monta rads with push pull so the fans don't have to spin up almost at all...
Now that you have a flow rate meter, I would love to see some more up-to-date testing on the impact of flow rate. The hard-core water cooling community is working on data that is over 10 years old, and while physics didn't change, the environment has. We are dealing with much higher heat loads and I wonder what impact that has on the old stance of targeting 1 gallon per minute (~220 l/hr) before diminishing returns.
220l/hr is pretty slow lol. Im running dual D5's. Im sitting around 1200L/hr and thats not even my max.
It says nylon on the website, is it different from acrylic?
Do they have a screen flow indicator that doesn’t have rgb vomet on it? I don’t need another program bogging down my computer just so the rgb doesn’t look like garbage
That water block on the end...what? They had XT/X versions of the 7900, not just GT, GTX? Hope it fits on the Ultra!
Corsair also said the new water blocks cover 30 series but so far I do not see any 30s available yet.
Flow rate is measured as a maximum (directily proportional to pipe diameter), in order to measure a restriction it would require a pump flowing water at its the control loops maximum speed, and then adding something to the loop in-order to observe a restriction.
An example of this for memory would be: a single drop of water can move through a 1mm tube and a 30mm tube at the same speed, but a LAKE cannot move through tube at the same speed as another tube.
At the control loops maximum speed* (//refering to the flow all the fittings and pipes and sustain(not dictated by the pump (//besides the diameter restrictions on a pump inlet&&outlet))
Can*
This is why flow meters are known to "restrict" flow rate, because they take up some of the area available for water to flow through at one time.
My apologies for phone thumb typing, big fan, hope you get everything figured out
So if your pump is pushing like 20000000l/h, then a block is attached the rlow meter might display like (256l/h). The exact physical maximum flow of the lowest diameter (volume of area) in the loop.
You can keep the test system for all future waterblock tests :)
Not sure which is worse, Asus tax or the Corsair tax. As long as people keep splashing the cash prices will continus to rise.
I would understand that as.. built in sensor FOR the cold plate
flow sensors in fuel systems on aircraft have to be calibrated in a long process after installing them, is that type calibration not needed in the flow sensor you bought?
BTW is your cat on valium?
I just noticed you are a Stargate fan. I have all the episodes for SG1, Atlantis and Universe along with the 3 movies.
where do you get your qdc's from ???!!!
roman, can you show how you would go about delidding a 7800x3d? that's something I want to do but have no confidence in cutting/filing the direct die frame.
secret tip: dont
@@quintit that seems to be the consensus
can you try like all of the good/deacent cpu and gpu water blocks
I’m watching with bated breath. I just bought one alongside your contact frame for 13th gen
RX Nvidia GeForce 7900 XT(X) .. just can't wait to see first benchmarks ;) ( corsair is testing printing on stickers ? )
these water blocks not have a motherboard backing plate?
perfect for a test bench but if your not going to build a high end pc to show it off its not worth it cause you cant see it anyways like you o11 dynamic case the motherboard is on the right side so if you have it on a desk it has to be on the right side of you to be able to see it so its one spot to put it and if you have it in a computer desk that has the desktop tower holder on the bottom left you cant see it anyway. so what im saying is for the price u can get it without the screen and just get a flow meter like what der8auer has then u can run the line were u can see it.
I probably sound like an old man saying this but Corsair to me feels like the car equivalent of those brands that make cosmetic vents and hub caps with spinning LEDs. It’s hard to take them seriously :P
how the hell do i connect the lcd block, does it connect to the motherboard or link?
Is there Monoblock Rgb with wireless or integrated wireless?? Suggest AIO rgb with wireless or integrated wire please for btf future build
200 bucks USD is 270 bucks cad, ouch so 300 dollars for me. Just for the block. OUCH, not sure if i would go for it at that price. I do like the idea of a LCD screen on the cooler. I'm on the fence. For 300 bucks CAD i can get an AIO with a LCD screen.
Hmmm, I'll wait until it goes down in price or on sale if i was going to do a custom loop.
Kinda wish it was black, grey kinda looks off for me IMO.
What is the name of the flow meter? I love to see it has conductivity, water quality and also temp
High flow next from Aquacomputer. It’s sold out quite often. Conductivity is the measurement for water quality. I’ve been using mine for over a year and one of the LED’s yellowed. Otherwise it’s been perfect.
@@Simon_Denmark thanks :)
I wouldn’t be worried about flow numbers the numbers you should be worried about is what is the actual water temperature is the water temperature actually dissipating the heat or is the water flowing too fast and it’s not dissipating the heat
Video idea: How flow affects temperatures.
Question: is water-cooling really quieter than a Noctua air cooler?
not usually. depending on the radiator you' may end up having more fans especially if you do push-pull setup and then you got to add the pump noise in to the equation. water coolers being quieter is a myth. edit: and air cooling is way less maintenance. i got in to water cooling for a while and still the one i built with custom loop as a second pc. but i went back to air on my latest build. noctua d15s no regrets
Do they tell you how much torque to tighten the fittings? Need to be careful with those acrylic threads.
I think you'd be better off spending your money on an auxiliary system display to mount in your case. (or elsewhere)
That could be used in any new system build and in any configuration, water, air, prayer, magic cooling dust . . . .
Well done sir. You have so much hands on experience with blocks and cooling. Why dont you put your knowledge to work and see about designing a few prototype water blocks. See if you can get an improvement with all the features and design protocols you deem good and important? It seems this would be the natural progression for you.
And send the prototype to LTT for testing! 😂
@@acidcrow4051 Um NO! :D
It'd be nice to see multiple pump speeds. Maybe 30, 60, 100. I run a DDC at 100% because my system is under my desk and the fans are still louder than it. So a D5 at 30% would give me almost no useful info about block choice as my flow rate is much greater.
That's a pretty big cat.
Finally a good LCD CPU block
can't wait to get my hands on geforce 7900 xtx
why a cold plate sensor rather than a water temp sensor?
LOL, just by seeing the title, I KNEW it was a Corsair product... I wonder why
Hook it up to a Volkswagen Golf III water loop for a better flow rate.
I normally would never agrue with you Roman, but that temperature sensors is probing the cold plate... so calling it the cold plate probe is fair. Please compare water temp from flow rate sensor to the cold plate temp under load. (It should be different)
Perhaps you should measure power draw on the pump as well as flow rate. If power draw increases on the pump, it could indicate restriction.
Not how that works.
How so?@@helium5912 If the pump speed is pwm controlled. The pump setting is a percentage of max rpm if I am not mistaken. This means that with resistance to the flow, the pump will work harder to maintain the same rpm/flow, and thus draws more power.
I feel like this RGB game won't stop until Luther Corp designs a 10000000 Lumen batman projector, which you can put in your PC for those (completely burning out eyes) RGB fanciness...
Might be a dumb question here guys, but can I use this for direct die cooling? 14900k asus z790 hero.
Thanks 😁
Trying this now. Any advice?
Compared to stupid prices of EK block, that corsair block looks quite reasonable(even with their crap iCue bloatware).
Considering the number of changes to the design, I'm surprised you didn't find much difference in the temperature
He did mention the optimized flow state, which could allow you to more efficiently connect multiple cooling devices to the same loop, like gpu cooling, without seeing any compromising bottlenecks in the water flow's ability to carry heat away from the components.
It can also be things like long-term durability increases and production cost optimization. Lots of small details add up when combined, but not necessarily in a manner that is directly observable for the end user. Either way, it's mostly the display that is driving up the cost.