Question Seized (?) AIO-pump - any hope?

[_Rick_]

So in my build thread from last winter, I added an Inno3D iChill Black 2080 Super - and it's a great card for the living room and my case.
That is until recently, when for some reason the fan-speed got locked at 30% (I'm assuming something to do with using MSI Afterburner for monitoring while doing some machine learning model training), the card started overheating, and increased pump speed to the point where it became very audible - and maybe even started cavitating (is that a thing?).
In any case, after maybe 3 minutes, I found the culprit, fanspeed normalized again, and the model training completed.

Come the next cold boot though, the pump apparently never got going. GPU crashed while I was watching a video - check everything out. Fanspeed was maxed, but no sound at all from the pump.
Pump control is apparently completely handled by the card, with no sensor or signal to read out or set rpm.

I opened an RMA with the vendor - card is only a year old after all, and it shouldn't be able to shake itself apart... Still, opened the RMA on Sunday, but due to high load, still haven't even heard back. Will also open a support ticket with Inno3D - but I haven't found the SN yet, which I guess might be helpful.
Is there anything else I can do to try and get that pump running? vBIOS flash? Shake the whole thing madly, in case there's an air bubble somewhere? Or is it likely game over for some mechanical aspect of the pump?

Edit: And yeah, right now obviously isn't the best time to have a ready-to-use AIO on a GPU fail. I might get my money back, which isn't worth anything in todays market, and there are no similar cooling solutions available anyway, or they will replace the cooler - which will likely mean a trip to the closest service center, waiting for a spare cooler to arrive, etc. Under normal conditions, you could just do a like-for-like exchange. But this was a rare beast to begin with...

 

[_Rick_]

And as I wrote this, the RMA-shipping label just arrived. So I guess, I'll be re-packaging it tonight, unless some genius suggestion is made before then

 

[aigomorla]

u cant mess with AIO's as they are sealed and if you gotten a RMA leave it as it is, and RMA it.
Only solution would require you to hack away at it, and well, that will invalidate your RMA.

 

[_Rick_]

Thanks for the input - went into the mail yesterday. I wonder how this will play out. I'll keep you posted, because it sure isn't a great time to RMA a graphics card, and maybe it'll help someone when searching the forum. Or maybe I'll cross-post to the GPU forum instead.

 

[thesmokingman]

Cavitation in an AIO, yea when pigs can fly! That said, AIO's are not the most reliable thing out there. It sounds like the pump ran dry...

 

[_Rick_]

Cavitation in an AIO, yea when pigs can fly! That said, AIO's are not the most reliable thing out there. It sounds like the pump ran dry...

Thanks for clearing that up.
I was imagining a really tiny pump going really fast, in some hilariously misguidedly designed way. But I guess it would have been unlikely.

And if by dry you mean out of lube (and not out of pumpable medium) then that may well be the case. Potentially overheated and destroyed the grease, which then set once the whole thing came to a stop/cooled down.
I did try to give the whole system a good shake to settle any possible air bubble out of the pump and into the rad - but not a peep out of it.

 

[Fallen Kell]

And this is why I still refuse to use liquid cooling...

Sure, it works great when it works. It just has so many additional failure points (running dry, electrolytic corrosion, biomass/slime buildup, pump failure, leaks...) that it isn't worth it to me.

Hopefully they get you a repair/replacement quickly.

 

[BonzaiDuck]

And this is why I still refuse to use liquid cooling...

Sure, it works great when it works. It just has so many additional failure points (running dry, electrolytic corrosion, biomass/slime buildup, pump failure, leaks...) that it isn't worth it to me.

Hopefully they get you a repair/replacement quickly.

Yeah -- see my recent thread here. I even avoided driving cars with the electric windows, because I imagined the complexity of too many parts and reliance on little electric motors might leave me sweating on a hot day if the AC wasn't working either. But I got over it.

Same thing applies, though. There's nothing simpler than a bunch of heatpipes and aluminum or copper fins. With the right airflow, they will continue to work even if a CPU fan fails.

If only the really good heatpipe coolers were a lot smaller . . .

 

[_Rick_]

Yeah -- see my recent thread here. I even avoided driving cars with the electric windows, because I imagined the complexity of too many parts and reliance on little electric motors might leave me sweating on a hot day if the AC wasn't working either. But I got over it.

You're saying this in the thread by the guy who had his wind-up windows in for repairs regularly

The main issue is that ATX is misdesigned for 150W+ GPUs (let alone 250W GPUs...).
You have all that space to cool a 90W (180 if you're really badass) CPU with a gigantic 12-heatpipe 140 mm fan monstrosity, but that GPU which averages twice the CPU TDP in the average gamer PC has to get away with 4 tiny, weirdly oriented heatpipes, attached to a few grams of aluminum fins, and some weedy fans with real hope of achieving decent airflow across those fins.

There's enough space to cool that CPU essentially inaudibly at100+ Watts, but trying to get a semi-silent GPU is impossible without watercooling, once it hits the power target

 

[Stuka87]

There is nothing inherently unreliable about an AIO. Obviously Company A could have a high quality product and last, while Company B has one that is cheap and fails after a few months.

I have installed some (including on my own machines) and never had an issue, one has been running for six years, and still runs every day.

Is the card still under warranty so you can at least get it fixed?

 

[aigomorla]

Obviously Company A could have a high quality product and last, while Company B has one that is cheap and fails after a few months.

company a and b both have the same oem as AIO's all hold a patent by ASTEK, unless its radically different.

 

[Stuka87]

company a and b both have the same oem as AIO's all hold a patent by ASTEK, unless its radically different.

Many are Asetek, but Coolit also has its own design. Corsair for instance uses both depending on which model cooler you get.

EDIT: GN has a video compare Asetek and Coolit, but at the end also shows the rather large list of suppliers.

 

[_Rick_]

Is the card still under warranty so you can at least get it fixed?

Should be - according to local law, there's 6 month of warranty and 2 years of extended warranty, with the burden of proof switching between the two modes from the vendor to the customer. So unless they find some indication that the card was used outside of its intended usage (it wasn't) they should be required to hand me a fix/replacement/reimbursement.

The card arrived there on Tuesday morning, still waiting for an update on the RMA. But since it took almost a week just to confirm the RMA, I'm guessing patience is what I need.

Also, the appartment block's central heating just broken. The worst moment for a 200 Watt -well-radiated heat source to disappear

This card's cooler is apparently a collab with Arctic Cooling, probably recyling an existing pump/tube/radiator design, and adding the full-cover block.

I hope that this kind of card will continue to be sold, as having the full warranty on the card and cooler makes it much more palatable, than having to swap coolers on an 800-1500 dollar card, and potentially voiding warranty by installing a block or AIO.
Cards still can die from e.g. bad RAM, and losing that money over a voided warranty is going to hurt...

 

[_Rick_]

So, as feared: They refunded me. On the upside it looks like a full refund, which is kind of insane. On the downside....yeah, in the GPU market this isn't consequential money. They can probably make money, by selling the card without a cooler.

Given my desire for a 240 mm radiator for a GPU - any tips how to proceed? I will probanly have to ask Inno3d if they plan to eventually renew their iChill black series- but I doubt their engineering costs were recouped on the 2080 series...

 

[Stuka87]

So, as feared: They refunded me. On the upside it looks like a full refund, which is kind of insane. On the downside....yeah, in the GPU market this isn't consequential money. They can probably make money, by selling the card without a cooler.

Given my desire for a 240 mm radiator for a GPU - any tips how to proceed? I will probanly have to ask Inno3d if they plan to eventually renew their iChill black series- but I doubt their engineering costs were recouped on the 2080 series...

Is the cooler that is mounted on it now available as a standalone part? That way you can at least buy it and get back up and going without having to try and find a new card, which is near impossible right now?

 

[_Rick_]

Card is already gone, but it looks like Inno3D will offer the new GPUs with a similar cooling solution in a few weeks.
Of course, availability will probably be extremely spotty - but I'm hoping demand will slow down at some point...
I mean, eventually everyone who needs a GPU should have gotten one.... Particularly in the consumer market. Enterprise/cloud eats GPU capacity like no other... But I don't think they're on that node yet.

Oh well, I guess now that I had the 2080 super refunded, I can just get a similarly priced 3070, and pretend that I didn't overpay...

Sent from my LM-V405 using Tapatalk