GPU running extremely hot even while idle

[Majornachos]

Hi everyone,

I've been trying to do some research on this but can't seem to find what I'm looking for. I know there are several posts about graphics cards running hot but mine has been running at 90c+ even while sitting here idle.

I have an AMD Powercooler Radeon HD myst edition 7870

I've downloaded speed fan and jack the thing up to 100% to get it to cool down while gaming (or while watching netflix, or while just sitting here typing this question!). Today I even took the thing apart and did the whole thermal paste replacement thing. I haven't seen any improvement in temperatures. My fans are fairly optimized to suck hot air out of the system. The fan on this thing sounds like it's about to liftoff and it doesn't seem to have speeds (at least that I can manipulate) under 90%.

Anyway, what I'm asking is, is all this my card's funny little way of telling me it's time for me to take it out back and put it out of its misery or can something still be done?

 

[amtseung]

Have you tried blasting/sucking the dust out of the heatsink itself? The HD7870 is not very new, so I assume you've had it for a while now.

 

[Majornachos]

Yeah I've had it for awhile, I bought it basically when it came out. And I did clean the dust from the heatsink when I took it all apart.

 

[amtseung]

Since you've replaced the thermal paste (my primary suspect), and you've dusted it off (my secondary suspect), assuming you reassembled the whole thing properly, the last thing I can think of is the chokes and VRM's have decayed over time, and are showing their age. I wonder what other, more knowledgeable people have to say.

 

[Majornachos]

Hmm...so it may be time for me to invest in a 1080. I was also wondering if this overheating gpu would possibly aid in failing a hard drive. I had an older that died slowly, so I ordered a new one which started to show the same signs of failure as the last. Any insight into this would be helpful.

 

[amtseung]

Woah woah woah hold your horses there buddy. I don't know what resolution and refresh rate you're playing at, but I question the necessity of a 1080. Maybe a 1070?

Hard drives will certainly fail faster when exposed to high heat, but even then, the HDD's inside home DVR's are required to run 24/7/365 at like 40-50C, and last for at least 5 years, if not longer. How hot are your hard drives getting? I kinda doubt a new hard drive would die that quickly unless you've dropped it or dropped a magnet on it or something.

 

[Majornachos]

Haha, you're probably right about the 1080/1070, though if I had to justify it I would use the words "future proofing"......even if that future never comes .

Anyway, unfortunately I didn't see what temps my hard drives were getting to when they were installed. The new one I sent back for an rma (which I'm not able to get due to a piece of the plastic that goes over the pins on the sata connector being broken. Either by me or before I got, I can't say) and the old one is just sitting around being useless. I may reinstall it just to see what temps it gets, but honestly I can tell you, even my CPU gets pretty hot when the GPU is overheating though it's able to cool down much better because of the CPU fan I have installed. With that in mind, I could assume that before I was fully aware of what was happening in my case and that it was a problem, everything was likely pretty hot.

As it is, I'm hesitant to buy another hard drive (a replacement for my replacement unfortunately) with my current setup. I've started thinking about a new case along with replacing that graphics card.

 

[amtseung]

Hmmmmm.

I tried trawling the interwebs for a picture of that myst edition cooler design, and couldn't really find any meaningful pictures with the cooler fully disassembled, but I question whether or not the VRM's are actively cooled. If you're overclocking, that would explain the rapid overheating of the card, especially with its low fin density heatsink and single fan. I would say, grab yourself a couple of screwdrivers, yank the thing apart (again), and check the PCB for unexplained discoloration, especially around the power delivery (the power plugs and the VRM's). Beyond that, I don't know what else to do. My HD7950 was a victim of having a crap cooler, so it's now half dead with an entire bank of VRM's removed entirely from the PCB.

Have you tried undervolting and/or underclocking the card?