Do you guys take your cards apart if they are too hot?

[NitroKoS]

I have 12 R9 290s and a couple of them run a full 10-15 degrees hotter than the others. Would I be correct in assuming that the thermal paste was likely applied poorly at the factory on those cards? I moved them around to make sure it wasn't a positional/airflow issue.

Do any of you with large amounts of cards regularly take the problem card apart to reapply the thermal paste thus dropping temperatures by 10 degrees or more?

Just curious if this is something I should pursue, I've built hundreds of PCs but I've never taken a gpu cooling system off.

[NameTaken]

I have my 280Xs suspended 3.5 inches apart being cooled by natural airflow so I haven't ran into any thermal problems yet.

[AtlantisPlatform]

I have 12 R9 290s and a couple of them run a full 10-15 degrees hotter than the others. Would I be correct in assuming that the thermal paste was likely applied poorly at the factory on those cards? I moved them around to make sure it wasn't a positional/airflow issue.

Do any of you with large amounts of cards regularly take the problem card apart to reapply the thermal paste thus dropping temperatures by 10 degrees or more?

Just curious if this is something I should pursue, I've built hundreds of PCs but I've never taken a gpu cooling system off.

Yes I do, I got 3 r9 280x cards from newegg recently, one card ran way hotter than the others.  I could have sent it in for warranty, but there were no more cards available. I chose to void warranty and repaste  chip. Took 15 minutes and gpu ran cool like the others afterwards.

[kalus]

i prefer to take full advantage of the warranties.  

run 'em hot, and if one keels over, throw a standby card in the rig, and RMA the stock, unmodified, intact card.  

I find there's less hassle with warranties if the cards are left the way they left.  for me, it's more efficient to have a few backup cards than try to get each card running precisely the same.  I would rather pour the effort into accumulating more cards and building more miners than fine tuning existing cards running at 90-95% of the maximum.

[crazyates]

Ya I'll take it apart and repaste it. Maybe throw some new thermal pads on the RAM chips. We usually see a few degC drop, esp on a hot card. RMAing a card is easy, but 45 minutes of downtime is a lot easier to handle than 5-6 days for an RMA (if you're lucky).

Here's the thing, tho: Make sure you use high quality screwdrivers, with hardened tips, and do not strip a single screw. If you're careful, they'll never even know you took it apart. Yes, I've had to RMA cards I've taken the HSF off, and I've never had an issue. You strip half the screws, do a poor job repasting, and re-use half-torn thermal pads on the RAM, and you're sure to get your warranty rejected on you.