Watercooled GPU but HOT backplate

[newtekie1]

Look at the image below and explain to me how the VRM is being efficiently cooled. Somehow, the image strikes a different impression in my mind:

The metal block touches the VRM. The fluid flows through the metal block. Heat moves from the VRM into the block, then from the block into the fluid. I'm not sure how this is hard to understand.

 

[Carsomyr]

ok... now let's compare it with this:

 

[EarthDog]

A red X... nice.......

 

[MrGenius]

The metal block touches the VRM. The fluid flows through the metal block. Heat moves from the VRM into the block, then from the block into the fluid. I'm not sure how this is hard to understand.

Fluid flows directly over the part of the block that comes in contact with the GPU die. There's a reason for that. So just reverse your frame of logic. What if the fluid channel only flowed over the VRM, and not over the GPU die. Would that be an effective way to cool the GPU die? OBVIOUSLY not, or that would be a popular method of block construction. Would that be an effective way to cool the VRM? OBVIOUSLY so, since the same reasons apply as to why the GPU die is better cooled that way.

It's a piss poor design for cooling the VRM effectively. It doesn't take a genius to figure that out.

 

[Carsomyr]

A red X... nice.......

sorry bout that. Fixed it.

BOTTOM LINE =
---> ZOTAC Arcticstorm 1080 ti = VRM "passively" Cooled (and very inefficient at that)
---> EKWB = VRM "ACTIVELY" Cooled

 

[EarthDog]

Nope.

 

[Athlonite]

As you can clearly see there is no water flow over the VRM section so the only way to suck more heat out is to increase the RPM of the pump more water going through the block the more heat that'll be taken away... that block is just poorly designed for heat transfer from the VRM section which is leading to the heating up of the backplate

 

[newtekie1]

Fluid flows directly over the part of the block that comes in contact with the GPU die. There's a reason for that. So just reverse your frame of logic. What if the fluid channel only flowed over the VRM, and not over the GPU die. Would that be an effective way to cool the GPU die? OBVIOUSLY not, or that would be a popular method of block construction. Would that be an effective way to cool the VRM? OBVIOUSLY so, since the same reasons apply as to why the GPU die is better cooled that way.

It's a piss poor design for cooling the VRM effectively. It doesn't take a genius to figure that out.

Yes, the reason for that is the GPU is producing by far the most heat. That is also why it has its own separate block with micro-fins on it that the water passes over.

View attachment 92919

As you can clearly see there is no water flow over the VRM section so the only way to suck more heat out is to increase the RPM of the pump more water going through the block the more heat that'll be taken away... that block is just poorly designed for heat transfer from the VRM section which is leading to the heating up of the backplate

The speed of the fluid really has no affect, unless it is so slow the fluid is getting super hot sitting in the block. But that isn't happening.

Just because there isn't fluid flowing directly over a section of the block, doesn't mean that part of the block isn't cooling the card.

If the block isn't getting hot, then the fluid is pulling the VRM heat away from it effectively. Period.

 

[Carsomyr]

Well this was fun. We all disagree lol.

Although I agree that the block Will indeed absorb heat because it is "connected" to a section that "touches" the VRM... If yall are right...
Then why in the world would EK (one of the most known companies out there for water cooling) bother to include water channels over the VRM section in their waterblocks ?!?!
Oh wait... it must be they thought it'd be a nice "decoration"... or ......... maybe they just felt like doing all that extra work and design for naught.... ridiculous...

There is one thing though I forgot to mention: In all of my research I have also found out that apparantly the zotac cards (1080ti) VRMs are not very power efficient. Apparantly they waste a lot of energy and this might be the reason for which Zotac cards specifically seem to output way more heat in regard to the VRM section... Although that remains to be proven

 

[Athlonite]

FUCK it I give and apparently so does the OP he's gotten a refund so it's a moot point about who's right and who's wrong so lets just all agree to disagree and be done with it


PS: it's a shit block design full stop

 

[Carsomyr]

FUCK it I give and apparently so does the OP he's gotten a refund so it's a moot point about who's right and who's wrong so lets just all agree to disagree and be done with it


PS: it's a shit block design full stop

lol. In case someone would be looking for another good reason to support my decision to return the card, here's a REALLY good one:
In my area it would be hard to sell the zotac card when the time to upgrade the gpu would come for me... Whereas if I have the gigabyte aorus model "modded" by installing the EK waterblock. I can always revert back to installing the air cooler on it and sell the card in its "original state" and sell the water block seperatly. I think my odds would be more favorable that way.

 

[eidairaman1]

What a circus this thread has become, do what you want, your money, not mine.

/thread

 

[newtekie1]

Then why in the world would EK (one of the most known companies out there for water cooling) bother to include water channels over the VRM section in their waterblocks ?!?!

Because they've been in the business long enough to know that if they didn't, they'd get a bunch of people that don't understand physics complaining about their blocks not cooling the VRM.

 

[Carsomyr]

Because they've been in the business long enough to know that if they didn't, they'd get a bunch of people that don't understand physics complaining about their blocks not cooling the VRM.

lol... what's it gonna take for you guys to see that something's clearly wrong with zotac's VRM cooling solution?
FLIR pictures of it vs FLIR pictures of the EK ... cause that's apparently what its gonna take isn't it lol

 

[newtekie1]

lol... what's it gonna take for you guys to see that something's clearly wrong with zotac's VRM cooling solution?
FLIR pictures of it vs FLIR pictures of the EK ... cause that's apparently what its gonna take isn't it lol

That would really just show that the EK is better than the Zotac, no one is arguing that.

All you'd have to provide is VRM Temps, or proof the Zotac block is getting too hot and the heat isn't moving into the fluid.

 

[eidairaman1]

lol... what's it gonna take for you guys to see that something's clearly wrong with zotac's VRM cooling solution?
FLIR pictures of it vs FLIR pictures of the EK ... cause that's apparently what its gonna take isn't it lol

Instead of griping send the card back or deal with it...

 

[Carsomyr]

Yup. Card's on it way back and I've received the EK waterblock. Hopefully that resolves the small issues I was having
I'm still greatful for all your input. One thing I've learned about PC and Overclocking... you're never done learning

 

[Carsomyr]

Alright I'm back with the results:

So using the EK waterblock and backplate I'm getting a better temp on the backplate. It still gets hot sometimes, but not nearly as much as it did with the zotac. To be fair, this is on a different card though (the aorus xtreme edition 1080 ti)

One thing I just HAVE to mention:

---> The aorus under water is BEASTING! I couldn't believe my eyes. The core clock maxes around 2076mhz stable although, I found out that if I lower it to 2063mhz (+52mhz instead of +64mhz) at max voltage (+100) I can get the memory to +675mhz stable (at least in valley benchmark) !!!

That's flipping 12.6 memory (6300mhz) Insane lol. I'm like omg, this is just insane memory performance, I'm thinking I should lower it down to the +500 range to avoid cooking the card. Although the temps seem fine. Water's not condensating anymore and no more bubbles in my loop

 

[Ferrum Master]

The aorus under water is BEASTING!

IMHO every 1080Ti under water does that... including my FE.

 

[Carsomyr]

well... in the end. the backplate gets super hot (even with EK)
I'm at 1.81v 2012mhz core clock without even overclocking the memory... and Every single thermal pad touches the block correctly. No air in the loop.

Apparantly the super heat is unavoidable on the card. I'm starting to wonder if that is the curse of the gtx 1080 ti. Maybe that gpu heats up more than say a 1080... or its vrm.
I'm clueless now lol.

I was certain that good water cooling with a good water block would mean a fanless PC (except for the rads of course)

Found an interesting thread on the same subject here on TPU:
https://www.techpowerup.com/forums/threads/solved-gpu-backplate-gets-too-hot.210444/page-1

 

[infrared]

I'm pretty sure you made a typo on the voltage, that's LN2 voltage!

It's not a "curse of the 1080Ti", it's just a reality of running a 250-300w card. It's not the same die as the 1080, it's physically a lot bigger with a lot more transistors, the stock TDP is 250w istead of 180w.

"I was certain that good water cooling with a good water block would mean a fanless PC (except for the rads of course)" You know better now, you always need some airflow. There's a lot more that produces heat than you think, even if you have the gpu, memory, vrm watercooled, there's plenty of other components on the card, and even the power traces in the PCB itself will produce heat, and it needs to go somewhere.

Now of course the back plate will get hot, it has to heat up in order to actually transfer the heat to the surrounding air. But unless you have zero airflow, it shouldn't get dangerously hot. Too hot to touch, sure, too hot for the card, no way. What I'm trying to say is that it's a GOOD thing that the backplate is getting hot, because it means it's transferring heat away from the card, if it was only lukewarm under heavy load I'd be more concerned.

I'm curious as to what you're using to load it up for your testing as well, furmark? I hope not, but if you are then the temps you're seeing are absolute worst case scenario.

 

[Carsomyr]

I'm pretty sure you made a typo on the voltage, that's LN2 voltage!

It's not a "curse of the 1080Ti", it's just a reality of running a 250-300w card. It's not the same die as the 1080, it's physically a lot bigger with a lot more transistors, the stock TDP is 250w istead of 180w.

"I was certain that good water cooling with a good water block would mean a fanless PC (except for the rads of course)" You know better now, you always need some airflow. There's a lot more that produces heat than you think, even if you have the gpu, memory, vrm watercooled, there's plenty of other components on the card, and even the power traces in the PCB itself will produce heat, and it needs to go somewhere.

Now of course the back plate will get hot, it has to heat up in order to actually transfer the heat to the surrounding air. But unless you have zero airflow, it shouldn't get dangerously hot. Too hot to touch, sure, too hot for the card, no way. What I'm trying to say is that it's a GOOD thing that the backplate is getting hot, because it means it's transferring heat away from the card, if it was only lukewarm under heavy load I'd be more concerned.

I'm curious as to what you're using to load it up for your testing as well, furmark? I hope not, but if you are then the temps you're seeing are absolute worst case scenario.

Well... I just learned that the ek water block for the aorus xtreme edition 1080 ti DOES NOT cover the memory VRM section...!?!

What the hell were ek thinking lol
See picture here
https://imgur.com/gallery/QExpP

Here's the block I should have bought instead... Man the obstacle just never stop adding up when doing pc watercooling lol
https://imgur.com/gallery/ZnXK2

 

[infrared]

You're looking at the wrong part, that block (and the last one) do cool the VRM.



Edit, the part you're looking at is the memory VRM which isn't under anywhere near as much stress.

Aorus xtreme edition 1080Ti pcb breakdown:

 

[newtekie1]

Well... I just learned that the ek water block for the aorus xtreme edition 1080 ti DOES NOT cover the memory VRM section...!?!

What the hell were ek thinking lol
See picture here
https://imgur.com/gallery/QExpP

Here's the block I should have bought instead... Man the obstacle just never stop adding up when doing pc watercooling lol
https://imgur.com/gallery/ZnXK2

But the water channel doesn't even flow over the memory VRM section of the "proper" block. So it wouldn't be cooled either by your logic, right?

 

[Carsomyr]

But the water channel doesn't even flow over the memory VRM section of the "proper" block. So it wouldn't be cooled either by your logic, right?

Ooh. The knives are flying low today haha
If indeed zotac's block would be doing it's job, the backplate would not turn into a furnace. And that is MY opinion. Nobody forces you to share it or even to agree with it. But a hardcore perfectionnist like me expects a waterblock to cause not just the gpu core to be at much lower temps, but the backplate/pcb as well. If you are fin with frying your pcb (to which you'll respond that wouldn't ever happen... Which might be true) but when I say fry, I mean: backplate gets so hot that it affects everything else. Air temp inside the case, temperature in the room, water temp... And at the price I pay, that bothers me.

I do concede to you I was wrong about the cooling physics. Flow Vs block touching. I'd bet the zotac block doesn't touch the Memory VRM as well

The memory vrm DOES get super hot. Why else would every other waterblock company bother to make sure it touches the block? Bykski, heatwave, kryo... U name them... They all cover said memory vrm

You seem to accuse me of exaggerating facts and I accuse you of minimizing their importance...

The TRUTH is probably somewhere in the middle. And that truth is: you bother to jump through all the hoops so you can get a silent gaming Pc that is water cooled when you could very well be just fine with air cooled... But you do it because you value the silence and the overclocking potential not to mention the aesthetics...?

YES sir you are indeed completely reasonable to expect the flipping god dang card to be cool

Thank you case closed, period


here, right here:
.

Period