Thursday, November 25th 2021

AMD Radeon RX 6900 XT Liquid Cooled Edition Now On Sale in Europe

AMD first announced the Radeon RX 6900 XT LC featuring a pre-binned Navi 21 XTXH GPU in June with the product only available to select system integrators and OEMs. These reference design cards were not expected to be available in the DIY market but listings for the cards appeared on some Indian retailers after the announcement. The card appears to be receiving a limited release with PowerColor's model (16GBD6-MW2DHC) now available to purchase from German retailer Mindfactory for 1749 EUR. The AMD Radeon RX 6900 XT LC features an increased boost clock speed of 2435 MHz from the 2250 MHz on the RX 6900 XT along with a memory speed boost from 16 Gbps to 18 Gbps.
AMD will be making a limited number of Radeon RX 6900 XT liquid cooled reference design graphics cards available to select system integrators. System integrators are a critical part of the gaming ecosystem and offering this this model through their channels helps ensure that as many of these products as possible get into gamers hands.
Sources: VideoCardz, Mindfactory
Add your own comment

84 Comments on AMD Radeon RX 6900 XT Liquid Cooled Edition Now On Sale in Europe

#26
Makaveli
kane nasend of... the monthly salary
damn what a deal lol

here is the Canadian pricing at newegg


Slightly better deal here Canada computers.

Posted on Reply
#27
Guwapo77
ZubasaThe Type-C itself is also a DP, just get an adapter / cable.
Well damn I didn’t even think of that especially since there is an adapter for basically every other plug in the world. Thanks!
Posted on Reply
#28
looniam
Kohl BaasThermodynamics. If you aten't restricted to keep the water under 35C to keep the CPU and other parts cool, you can triple the cooling performance without sweat. That's how all these single 120 AIO solutions work. And they handle occasionally 400-500Watts of power.
yes heat transfer increases with temp delta. however you'll still end up w/a hotter part.
Posted on Reply
#29
Mussels
Freshwater Moderator
Caring1At that price I want a 240mm radiator and gold plated shroud on the card.
Shroud was off playing CS:GO, PubG and Fortnite was was unavailable for modelling for the cards artwork

(Oh god remember the days of gargoyles and vampires and stuff on GPU boxes? Imagine erotic semi-nudes of male pro gamers. Paypal me money to stop me suggesting this idea to them)


Ah i was wondering why there was no fan + AIO design, this is for OEM's and SI's.
It's a limited edition looks product, thermals aint part of its design.
Posted on Reply
#30
Valantar
MusselsAh i was wondering why there was no fan + AIO design, this is for OEM's and SI's.
It's a limited edition looks product, thermals aint part of its design.
If this is anything like AMD's previous LC flagships (Fury X, Vega 64 LC) then it has a quasi full cover block through the use of base plates and tubing leading the liquid to cool all heat generating onboard components. Sadly I can't find any pictures of this under the shroud, but both the Fury X and V64 LC had rather complex cooling layouts.
Posted on Reply
#31
nguyen
These 6900XT LC has stock TDP of 350W, PCgameshardware already has the review.
350W on a single 120mm rad is just asking for premature pump failure, I highly doubt these AIO would even last 3 years.
If it cost the same as regular 6900XT then it makes sense to buy this 6900XT LC and put a full-cover WB on it.
Posted on Reply
#32
Mussels
Freshwater Moderator
ValantarIf this is anything like AMD's previous LC flagships (Fury X, Vega 64 LC) then it has a quasi full cover block through the use of base plates and tubing leading the liquid to cool all heat generating onboard components. Sadly I can't find any pictures of this under the shroud, but both the Fury X and V64 LC had rather complex cooling layouts.
I meant like the ugly abomination on my GTX 1080, where it has a 92mm fan cooling the VRM's and VRAM, with a block just for the GPU. We've seen quite a few "hybrid" cards split that way - this one looks pure water, but a 120mm really seems too small for that without a hidden fan... unless SI's and OEM's built the entire PC around it with directed airflow


It better not be for mining only loads -.-
Posted on Reply
#33
Jism
nguyenThese 6900XT LC has stock TDP of 350W, PCgameshardware already has the review.
350W on a single 120mm rad is just asking for premature pump failure, I highly doubt these AIO would even last 3 years.
If it cost the same as regular 6900XT then it makes sense to buy this 6900XT LC and put a full-cover WB on it.
Stop.

Such products are extensively tested, and if i'm correct the warranty is there as well.

The 295X2 had a TDP of 500W easily and they managed to run to this very day with the same AIO. Or at least a 120mm setup.



60 degees load for a dual GPU setup on a 120mm radiator.. is'nt what you call bad really. The 300W is nothing compared to what they can do.

The price is a bit high tho; i like the prestige in this thing; it's a exclusive card and it does hold some value being a high binned part.
Posted on Reply
#34
Valantar
MusselsI meant like the ugly abomination on my GTX 1080, where it has a 92mm fan cooling the VRM's and VRAM, with a block just for the GPU. We've seen quite a few "hybrid" cards split that way - this one looks pure water, but a 120mm really seems too small for that without a hidden fan... unless SI's and OEM's built the entire PC around it with directed airflow
IMO that's mainly because those cards are typically made at the lowest possible cost by AIB partners, essentially strapping on a ready-made AIO (at best with a larger cold plate for RAM cooling) and then attempting to address the sudden problem of "oh crap, there's still heat generating components on here, what do we do now?" (Often making use of base plate designs from air cooled cards to coo the VRMs and even VRAM at times.) AMD's collaboration with Cooler Master is probably close enough that they can have these semi-custom AIOs built relatively cheaply, even compared to an off-the-shelf design like an AIB partner would use. I would actually expect OEMs to prefer a design with an added blower fan on the card if that was an option, as fan mounts are always rare on OEM cases - that's likely why this has a single 120mm rad as well, as OEM cases with 240mm rad mounts are really rare (as depressing as that is). But even on a GPU like this, the VRMs and other stuff don't put out enough heat for there to be a need for it as long as you can get even a little water flowing across it. I mean, the Fury X famously just had a copper pipe laid across its VRMs, which worked excellently.

(image from Anandtech)
nguyenThese 6900XT LC has stock TDP of 350W, PCgameshardware already has the review.
350W on a single 120mm rad is just asking for premature pump failure, I highly doubt these AIO would even last 3 years.
If it cost the same as regular 6900XT then it makes sense to buy this 6900XT LC and put a full-cover WB on it.
Why would it lead to that? These aren't un-cooled DDC pumps running at full speed, they are low power AIO pumps. They do not generate a lot of heat. Even if the fluid is hitting 40-45° there's no reason to expect this to cause premature pump failures. It's perfectly safe to run a setup like this for normal consumer usage for years and years with no expectation of rapid failure.
Posted on Reply
#35
nguyen
ValantarIMO that's mainly because those cards are typically made at the lowest possible cost by AIB partners, essentially strapping on a ready-made AIO (at best with a larger cold plate for RAM cooling) and then attempting to address the sudden problem of "oh crap, there's still heat generating components on here, what do we do now?" (Often making use of base plate designs from air cooled cards to coo the VRMs and even VRAM at times.) AMD's collaboration with Cooler Master is probably close enough that they can have these semi-custom AIOs built relatively cheaply, even compared to an off-the-shelf design like an AIB partner would use. I would actually expect OEMs to prefer a design with an added blower fan on the card if that was an option, as fan mounts are always rare on OEM cases - that's likely why this has a single 120mm rad as well, as OEM cases with 240mm rad mounts are really rare (as depressing as that is). But even on a GPU like this, the VRMs and other stuff don't put out enough heat for there to be a need for it as long as you can get even a little water flowing across it. I mean, the Fury X famously just had a copper pipe laid across its VRMs, which worked excellently.

(image from Anandtech)


Why would it lead to that? These aren't un-cooled DDC pumps running at full speed, they are low power AIO pumps. They do not generate a lot of heat. Even if the fluid is hitting 40-45° there's no reason to expect this to cause premature pump failures. It's perfectly safe to run a setup like this for normal consumer usage for years and years with no expectation of rapid failure.
Yeah sure talk about how many FuryX AIO pump die within 3 years LOL
There are lots of reports about those FuryX pump whine and failure, just google them.

here is an example
Posted on Reply
#36
Caring1
nguyen
No such thing as impossible.
Posted on Reply
#37
Valantar
nguyenYeah sure talk about how many FuryX AIO pump die within 3 years LOL
Got any examples?
nguyenThere are lots of reports about those FuryX pump whine and failure, just google them.
Whining and pump failure are two entirely unrelated things. The whining in the Fury X was due to a poorly vibration dampened pump motor. I should know, seeing how I replaced my own whiny Fury X AIO with an EK waterblock when I found one on sale for around €30, as it was bothering me sufficiently to go custom loop. So, again, care to show some examples of actual premature pump failures?
nguyenhere is an example
If you actually watched the video, you'd notice that he says - at less than two minutes in! - the following:
I've got one of these guys, a Fury X, and it lead me to think that it's a pretty good video to make - on the reasons why you shouldn't buy a hybrid card, nor should you produce one.
So, he's not as much making a video on the Fury X as he is making a video on how permeation and evaporation from AIOs is a reason to avoid them, as they can't be refilled easily. This is universally true for all AIOs (and whether one sees it as a valid reason to avoid them or not is quite separate from what you're arguing). (Of course in the case of the video he entirely misses the point that he's getting air stuck in his pump constantly because he doesn't have the radiator mounted to his case and higher up than the GPU but has it flopping around his desk, leading to air rising up to the card and getting stuck in the pump in a way that wouldn't happen as much (or at all) in a normal installation. And as such he's having much more trouble with the card than he would have if it were installed in his case - and as he himself makes clear, it still works perfectly once he gets the stuck air out of the pump. But never mind that right now.)

But again: this is not proof of premature pump failures. It might be indicative that the Fury X cooler design was too exposed to permeation and/or evaporation. Or it might be a few units. A single example neither proves nor disproves anything. But fluid loss and pump failure are two distinct things. Yes, fluid loss can lead to pump failure, but there's no sign of that in this case.

And, while I didn't bother finishing the video, it did seem like he managed to refill it? Though he did seem to really play up the theatrics and make it as complicated as possible for himself. Refilling an AIO is entirely possible, it just isn't easy. It can certainly be done with less mess and hassle than he did.
Posted on Reply
#38
nguyen
ValantarGot any examples?

Whining and pump failure are two entirely unrelated things. The whining in the Fury X was due to a poorly vibration dampened pump motor. I should know, seeing how I replaced my own whiny Fury X AIO with an EK waterblock when I found one on sale for around €30, as it was bothering me sufficiently to go custom loop. So, again, care to show some examples of actual premature pump failures?

If you actually watched the video, you'd notice that he says - at less than two minutes in! - the following:

So, he's not as much making a video on the Fury X as he is making a video on how permeation and evaporation from AIOs is a reason to avoid them, as they can't be refilled easily. This is universally true for all AIOs (and whether one sees it as a valid reason to avoid them or not is quite separate from what you're arguing). (Of course in the case of the video he entirely misses the point that he's getting air stuck in his pump constantly because he doesn't have the radiator mounted to his case and higher up than the GPU but has it flopping around his desk, leading to air rising up to the card and getting stuck in the pump in a way that wouldn't happen as much (or at all) in a normal installation. And as such he's having much more trouble with the card than he would have if it were installed in his case - and as he himself makes clear, it still works perfectly once he gets the stuck air out of the pump. But never mind that right now.)

But again: this is not proof of premature pump failures. It might be indicative that the Fury X cooler design was too exposed to permeation and/or evaporation. Or it might be a few units. A single example neither proves nor disproves anything. But fluid loss and pump failure are two distinct things. Yes, fluid loss can lead to pump failure, but there's no sign of that in this case.

And, while I didn't bother finishing the video, it did seem like he managed to refill it? Though he did seem to really play up the theatrics and make it as complicated as possible for himself. Refilling an AIO is entirely possible, it just isn't easy. It can certainly be done with less mess and hassle than he did.
Another example


Your google skills must be very lacking or you just feint ignorance, it doesn't take me that long to find some examples....
Posted on Reply
#39
Valantar
nguyenAnother example


Your google skills must be very lacking or you just feint ignorance, it doesn't take me that long to find some examples....
I haven't even googled it, but then it's not my responsibility to prove your arguments, is it? If the best you can come up with is "I'm right, google it", then you've somehow misunderstood some pretty basic tenets of how a discussion works. The onus is on you to prove your own points, not anyone else. And again: single examples neither prove nor disprove anything. Anecdotal evidence is not evidence. If this is indeed a systemic problem, can you for example show an article talking about it as a major problem? A forum thread of more than a handful of people saying this has happened to them - say, from AMD's community forums? While the Fury X was by no means as popular as the 980 Ti, it still sold reasonably well for its time. If this was a widespread problem, something more substantial than individual examples ought to exist. I'd be more than happy to accept it as a systemic problem if you could actually provide some proof to back this up.

Oh, and as you are arguing that this is specifically down to them using a 120mm AIO for a high powered card, I assume you're able to show the same for every other hybrid GPU out there with a 120mm AIO as well? IIRC those have been around from various AIB partners since at least 2015, both for Nvidia an AMD's high end cards. Remember: if this is a systemic problem but specific to the Fury X, that doesn't prove your initial statement of a 120mm AIO on a high powered GPU being problematic - it might then be down to a poor design, a cheap pump (which it definitely had!), or any other of a number of variables unique to that GPU. To prove your point, you need to show it across a variety of GPUs. Because, just to remind you, this was the core of your original argument:
nguyen350W on a single 120mm rad is just asking for premature pump failure
So: either show some broadly applicable proof for that statement, and not an individual GPU, or please stop spreading FUD.
Posted on Reply
#40
nguyen
ValantarI haven't even googled it, but then it's not my responsibility to prove your arguments, is it? If the best you can come up with is "I'm right, google it", then you've somehow misunderstood some pretty basic tenets of how a discussion works. The onus is on you to prove your own points, not anyone else. And again: single examples neither prove nor disprove anything. Anecdotal evidence is not evidence. If this is indeed a systemic problem, can you for example show an article talking about it as a major problem? A forum thread of more than a handful of people saying this has happened to them - say, from AMD's community forums? While the Fury X was by no means as popular as the 980 Ti, it still sold reasonably well for its time. If this was a widespread problem, something more substantial than individual examples ought to exist. I'd be more than happy to accept it as a systemic problem if you could actually provide some proof to back this up.

Oh, and as you are arguing that this is specifically down to them using a 120mm AIO for a high powered card, I assume you're able to show the same for every other hybrid GPU out there with a 120mm AIO as well? IIRC those have been around from various AIB partners since at least 2015, both for Nvidia an AMD's high end cards. Remember: if this is a systemic problem but specific to the Fury X, that doesn't prove your initial statement of a 120mm AIO on a high powered GPU being problematic - it might then be down to a poor design, a cheap pump (which it definitely had!), or any other of a number of variables unique to that GPU. To prove your point, you need to show it across a variety of GPUs. Because, just to remind you, this was the core of your original argument:

So: either show some broadly applicable proof for that statement, and not an individual GPU, or please stop spreading FUD.
It's just common sense, which you are severly lacking.
AIO always have higher failure rate than air cooler, it would be even less reliable when you put very high thermal load on an inadequate AIO.

But sure, just defend your theory, when you yourselve already experience pump whine on your Fury X, which could be an indicator of water permeation that could lead to early pump failure down the line.

There are so few FuryX owners that a few anecdotal evidences is probably all of them :roll:
Posted on Reply
#41
Valantar
nguyenIt's just common sense, which you are severly lacking.
AIO always have higher failure rate than air cooler, it would be even less reliable when you put very high thermal load on an inadequate AIO.

But sure, just defend your theory, when you yourselve already experience pump whine on your Fury X, which could be an indicator of water permeation that could lead to early pump failure down the line.

There are so few FuryX owners that a few anecdotal evidences is probably all of them :roll:
... sigh. My Fury X had pump whine from day 1. It did not have permeation issues, as there was very little air in the loop, and in the two years or so I ran it before switching to the custom loop it didn't change in any noticeable way. This is easily identified by moving it around, so ... yeah.

Also, "it's just common sense" is just another way of saying "I have run out of arguments and/or data to back up my arguments".

And btw, why are you resorting to personal attacks here? I'm just asking you to back up your claims. Is that so offensive to you?
nguyenAIO always have higher failure rate than air cooler,
Have I ever said anything against this?
nguyenit would be even less reliable when you put very high thermal load on an inadequate AIO.
This is the key point though: this is a new claim, which requires new data to prove it. While heat can indeed kill pumps, there isn't necessarily a significant thermal pathway between an AIO pump and its fluid (there typically isn't, actually - the impellers typically have plastic axles or are entirely decoupled with magnetic drives), so thermal effects of running the loop hotter on pump failure are - at least until someone tests this and can show some proof - unknown.

As it seems you need to be spoon fed this: I am entirely open to this being a systemic problem. But so far, you have shown exactly zero proof as to this being the case. A couple of examples does not make a systemic problem. And as I said, my standards here are pretty low: even an article reporting on this being widespread, or a forum thread with more than a handful of people saying their pumps have died would be enough for me. But a couple of youtube videos or tweets showing examples of failed pumps? Not even close.

Judging by that "joke" at the end you also seem to have missed all the 980s , 980 Tis, 1080s, 1080 Tis and other hybrid GPUs with 120mm AIOs ... As I said: for your point to be true, these would also need to be affected by widespread premature failure.
Posted on Reply
#42
Mussels
Freshwater Moderator
ValantarIMO that's mainly because those cards are typically made at the lowest possible cost by AIB partners, essentially strapping on a ready-made AIO (at best with a larger cold plate for RAM cooling) and then attempting to address the sudden problem of "oh crap, there's still heat generating components on here, what do we do now?" (Often making use of base plate designs from air cooled cards to coo the VRMs and even VRAM at times.) AMD's collaboration with Cooler Master is probably close enough that they can have these semi-custom AIOs built relatively cheaply, even compared to an off-the-shelf design like an AIB partner would use. I would actually expect OEMs to prefer a design with an added blower fan on the card if that was an option, as fan mounts are always rare on OEM cases - that's likely why this has a single 120mm rad as well, as OEM cases with 240mm rad mounts are really rare (as depressing as that is). But even on a GPU like this, the VRMs and other stuff don't put out enough heat for there to be a need for it as long as you can get even a little water flowing across it. I mean, the Fury X famously just had a copper pipe laid across its VRMs, which worked excellently.

(image from Anandtech)


Why would it lead to that? These aren't un-cooled DDC pumps running at full speed, they are low power AIO pumps. They do not generate a lot of heat. Even if the fluid is hitting 40-45° there's no reason to expect this to cause premature pump failures. It's perfectly safe to run a setup like this for normal consumer usage for years and years with no expectation of rapid failure.
I'd never seen inside the fury X before. My life is complete.
Posted on Reply
#43
nguyen
Valantar... sigh. My Fury X had pump whine from day 1. It did not have permeation issues, as there was very little air in the loop, and in the two years or so I ran it before switching to the custom loop it didn't change in any noticeable way. This is easily identified by moving it around, so ... yeah.

Also, "it's just common sense" is just another way of saying "I have run out of arguments and/or data to back up my arguments".

And btw, why are you resorting to personal attacks here? I'm just asking you to back up your claims. Is that so offensive to you?

Have I ever said anything against this?

This is the key point though: this is a new claim, which requires new data to prove it. While heat can indeed kill pumps, there isn't necessarily a significant thermal pathway between an AIO pump and its fluid (there typically isn't, actually - the impellers typically have plastic axles or are entirely decoupled with magnetic drives), so thermal effects of running the loop hotter on pump failure are - at least until someone tests this and can show some proof - unknown.

As it seems you need to be spoon fed this: I am entirely open to this being a systemic problem. But so far, you have shown exactly zero proof as to this being the case. A couple of examples does not make a systemic problem. And as I said, my standards here are pretty low: even an article reporting on this being widespread, or a forum thread with more than a handful of people saying their pumps have died would be enough for me. But a couple of youtube videos or tweets showing examples of failed pumps? Not even close.

Judging by that "joke" at the end you also seem to have missed all the 980s , 980 Tis, 1080s, 1080 Tis and other hybrid GPUs with 120mm AIOs ... As I said: for your point to be true, these would also need to be affected by widespread premature failure.
Heat kill electronics (Pump have IC), simple logic, I don't have to prove anything.

DDC and D5 pumps have maximum water operating temps (Typically 60C for D5 and 50C for DDC), how do you think AIO pump is exempted from this?

I built 2 exact custom loops for a friend and I, he has 50C water temp and his Bitspower S1 pump failed after 2 years and my pump still kicking after ~4 years at 40C water temp (I use Noctua NF-A12x25 fans while my friend use EK Vardar 1200rpm fans). Yes I did warn my friend about possible early pump failure but he ignored it.



Coincident that my friend S1 pump failed and mine did not? hardly.

Mind you that the liquid temp inside those 6900XT LC would easily reach above 50C with room temperature at 25C, the longevity for these AIO is just dubious.
Posted on Reply
#44
Mussels
Freshwater Moderator
Sample sizes of 1 is not good science
Posted on Reply
#45
Zubasa
nguyenWell Covid has 1% rate of fatality, I guess there is no need to take Covid too seriously do we :D
1% just means 1/100 of something, the samples size being the number of patients who were infected is in the hundreds of millions.
Posted on Reply
#46
usiname
nguyenWell Covid has 1% rate of fatality, I guess there is no need to take Covid too seriously do we :D
If you build custom loops in the same way you calculate statistics its surprisingly that the two loops lasted that long
Posted on Reply
#47
Valantar
nguyenWell Covid has 1% rate of fatality, I guess there is no need to take Covid too seriously do we :D
Even with this being meant as a joke, the conflation of two wildly different things here (sample size and findings from a sample size of hundreds of millions) is an apt illustration of why we're having this nonsensical argument in the first place. To be able to make that joke requires an ability to conflate radically different things and treat them as the same. And that's what this discussion is all about.
nguyenI'm talking about statistic, if hardware are working to their limits, chances of failure are much higher. If people want more proof then I will just kindly ask them to learn some common sense.

Same with Covid, I'm not taking my chances even though the chance of dying is ~1%, which a lot of people deem a normal flu.
This just underscores that you don't seem to understand the concept of sample size. Either that or you're just doubling down after being called out on making a nonsensical comparison. Regardless of conditions, a sample size of one is never proof of a systemic problem. It can be, but you can never know that it is unless you increase the sample size.
nguyenHeat kill electronics (Pump have IC), simple logic, I don't have to prove anything.
Yes you do, because that extremely broad point isn't what you're arguing. You are specifically arguing that a 120mm AIO on a high powered GPU is inherently significantly more likely to fail than, well, any other cooler, be it air or water with a bigger radiator. But not only that, really: you claim was that this is such woefully underpowered cooling that "this is just asking for premature pump failure". That's pretty specific and strongly worded. And that was the claim made in your original post that I responded to. If you've changed your argument since then, we'll, then you need to make that clear.
nguyenDDC and D5 pumps have maximum water operating temps (Typically 60C for D5 and 50C for DDC), how do you think AIO pump is exempted from this?

I built 2 exact custom loops for a friend and I, he has 50C water temp and his Bitspower S1 pump failed after 2 years and my pump still kicking after ~4 years at 40C water temp (I use Noctua NF-A12x25 fans while my friend use EK Vardar 1200rpm fans). Yes I did warn my friend about possible early pump failure but he ignored it.



Coincident that my friend S1 pump failed and mine did not? hardly.

Mind you that the liquid temp inside those 6900XT LC would easily reach above 50C with room temperature at 25C, the longevity for these AIO is just dubious.
It's likely not a coincidence, no, but are your two loops using the same pump? You weren't clear on that. Also, DDC(-style) pumps are notorious for poor longevity due to them not being directly cooled by the fluid (unlike D5s, which have metal-to-liquid contact and are built to be cooled by the passing fluid) and thus needing heatsinks and some airflow to keep cool. A DDC or similar pump without airflow, even with a heatsink, will die relatively quickly. I recently saw someone put thermal sensors on the PCB in a DDC with a decently good DDC heatsink (an EK one IIRC), and saw PCB temperatures in the 90s. Considering that, any increase in ambient temperatures, including adjacent but not directly coupled liquid temperatures, will push things towards failure. But the thing is, AIO pumps are much weaker than DDCs, and don't output even remotely as much heat. Most AIO pumps pull 5W or less in normal operation, DDCs range from 10-18W. Electric motors put are all roughly equally efficient, so heat output scales mostly linearly with wattage. Thus, assuming a similarly poor thermal design to a DDC, it will still take a lot more to cook a 5W pump than a 10W one. Btw, if you were using the Bitspower S1, did you add a heatsink? And were/are you and your friend running them at full speed? 'Cause that flat top they have on it looks woefully insufficient for cooling unless there's significant airflow across it. Oh, and for that comparison between you and your friend: were they in the same case? Did you have the same number of fans? Without that, they're not comparable. And tbh, even different fans means they aren't, as the low airflow from those slower fans causing less direct cooling of the pump might be just as much of a cause of the pump failing as the high fluid temperature. Most likely it's a combination of both - it was likely heat-induced failure, but pinning that to the fluid alone isn't supported even by your anecdotal data. A combination of hot fluid, a hot case, insufficient airflow across the pump, and insufficient pump heatsink surface area is a more likely explanation.
Posted on Reply
#48
nguyen
ValantarEven with this being meant as a joke, the conflation of two wildly different things here (sample size and findings from a sample size of hundreds of millions) is an apt illustration of why we're having this nonsensical argument in the first place. To be able to make that joke requires an ability to conflate radically different things and treat them as the same. And that's what this discussion is all about.

This just underscores that you don't seem to understand the concept of sample size. Either that or you're just doubling down after being called out on making a nonsensical comparison. Regardless of conditions, a sample size of one is never proof of a systemic problem. It can be, but you can never know that it is unless you increase the sample size.

Yes you do, because that extremely broad point isn't what you're arguing. You are specifically arguing that a 120mm AIO on a high powered GPU is inherently significantly more likely to fail than, well, any other cooler, be it air or water with a bigger radiator. But not only that, really: you claim was that this is such woefully underpowered cooling that "this is just asking for premature pump failure". That's pretty specific and strongly worded. And that was the claim made in your original post that I responded to. If you've changed your argument since then, we'll, then you need to make that clear.

It's likely not a coincidence, no, but are your two loops using the same pump? You weren't clear on that. Also, DDC(-style) pumps are notorious for poor longevity due to them not being directly cooled by the fluid (unlike D5s, which have metal-to-liquid contact and are built to be cooled by the passing fluid) and thus needing heatsinks and some airflow to keep cool. A DDC or similar pump without airflow, even with a heatsink, will die relatively quickly. I recently saw someone put thermal sensors on the PCB in a DDC with a decently good DDC heatsink (an EK one IIRC), and saw PCB temperatures in the 90s. Considering that, any increase in ambient temperatures, including adjacent but not directly coupled liquid temperatures, will push things towards failure. But the thing is, AIO pumps are much weaker than DDCs, and don't output even remotely as much heat. Most AIO pumps pull 5W or less in normal operation, DDCs range from 10-18W. Electric motors put are all roughly equally efficient, so heat output scales mostly linearly with wattage. Thus, assuming a similarly poor thermal design to a DDC, it will still take a lot more to cook a 5W pump than a 10W one. Btw, if you were using the Bitspower S1, did you add a heatsink? And were/are you and your friend running them at full speed? 'Cause that flat top they have on it looks woefully insufficient for cooling unless there's significant airflow across it. Oh, and for that comparison between you and your friend: were they in the same case? Did you have the same number of fans? Without that, they're not comparable. And tbh, even different fans means they aren't, as the low airflow from those slower fans causing less direct cooling of the pump might be just as much of a cause of the pump failing as the high fluid temperature. Most likely it's a combination of both - it was likely heat-induced failure, but pinning that to the fluid alone isn't supported even by your anecdotal data. A combination of hot fluid, a hot case, insufficient airflow across the pump, and insufficient pump heatsink surface area is a more likely explanation.
Our system are almost identical except the fans: 8700K + 2080Ti (~350W combine load), both run S1 pump at max speed without heatsink. Even now after I swapped to a 9900K + 3090 for a full year, my S1 pump is still running fine.
A combination of hot fluid, a hot case, insufficient airflow across the pump, and insufficient pump heatsink surface area is a more likely explanation.
Same case (Lian Li O11D), same pumps running max speed all the time without heatsink and no airflow over the pumps (no air access to the back), the only difference is that I run my fans at 2000rpm vs 1200rpm on my friend's built leading to much lower liquid temp on my case.

Though you are just gonna disregard any evidence as anecdotal evidence aren't you? Surely you are not running anything in your systems at temperature that you deem could cause potential failure? Like running your 6900XT at 180W undervolted instead of 300W+ that it could run without issue?

I guess you don't know that plastic is not a thermal insulator, the pump IC temperature is not decoupled from the liquid temperature. The specs for DDC pump also have maximum liquid temperature of 60C (I thought it was 50C earlier)

So yeah, couple with the fact that higher water temp will lead to faster rate of water permeation, it's not unfounded that cooling 350W with a single 120mm AIO will lead to EARLY pump failure. Well the AIO might live long, if you live in Alaska that is.

Not sure how legit this is but it seems the general consensus for max water temp for AIO is also 60C
Posted on Reply
#49
gorillacookies
ValantarJudging by that "joke" at the end you also seem to have missed all the 980s , 980 Tis, 1080s, 1080 Tis and other hybrid GPUs with 120mm AIOs ... As I said: for your point to be true, these would also need to be affected by widespread premature failure.
Both the 980ti and 1080ti were 250 watt cards under extreme loads and the others much less. Maybe instead of constantly clogging up this thread demanding proof, you provide some of your own because that's how High School Debates work.
Posted on Reply
#50
nguyen
Woot, seems like I should have pointed out from the review
If you put the RX 6900 XT LC under sustained full load, the temperatures rise to up to 80 ° C (GPU) or 90 ° C (hotspot). The delta between the hottest and middle temperature is relatively small, which speaks for a good heat emission and distribution - despite the high heat density of the 284 watt radiating core. How, among other things, with the help of the HW Infocan be read out, the temperature of the coolant reaches around more than 60 ° C, which is still within, but on the edge of common specifications
>60C water temp? I guess I should be asking for proofs that AIO can reliably work at this temperature instead of giving proofs :roll:
Posted on Reply
Add your own comment
Mar 15th, 2025 22:12 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts