Author

Topic: Crater on surface of GPU heat spreader (Read 152 times)

legendary
Activity: 1820
Merit: 1121
August 04, 2021, 04:56:35 PM
#10
I carefull sanded the crater out of the heatsink and VERY carefully sanded the GPU heatspreader using very fine sandpaper.  Completely removed the heatsink crater but left a much smaller pit in the heatspreader as I wasn't game to go too deep.  Then I had to deal with cleaning the microscopically fine metal powder that was undoubtedly left on the card.  Put it all back together and... well definitely an improvement.  Instead of running 10 degrees hotter it's now only about 4 degress hotter.  

Definitely nothing wrong with the fans and I can't see anything amiss with any of the components.  The other 7 cards in the rig are fine so it's not the power supply.  The reported watts is normal and the total power drawn by the rig from the socket is normal and the card had been running 24/7 for a month or two before I got around to cracking it open to see why it was 10 degrees hotter, so I expect it will continue to run fine especially as it's now only 4 degrees hotter.  Appears to be solely a heating issue caused by a defect in the heatsink or heatspreader, maybe an airpocket, who knows.  Up until a few months ago, it had been running pretty much non-stop for 4 years at a normal temperature.  Interestingly, it's winter here so temperature-wise, it actually occurred when it less stressed than 6 months ago.
Change the thermal pad on this video card at the beginning of summer and install additional cooling on the other side of the video card.
This is a little sisit the temperature.
Your radiator may have been made of substandard materials.
I hope that your video card is the last one in the farm, because the last video card always has the highest temperature
full member
Activity: 258
Merit: 104
August 04, 2021, 02:08:38 AM
#9
I carefull sanded the crater out of the heatsink and VERY carefully sanded the GPU heatspreader using very fine sandpaper.  Completely removed the heatsink crater but left a much smaller pit in the heatspreader as I wasn't game to go too deep.  Then I had to deal with cleaning the microscopically fine metal powder that was undoubtedly left on the card.  Put it all back together and... well definitely an improvement.  Instead of running 10 degrees hotter it's now only about 4 degress hotter.  

Definitely nothing wrong with the fans and I can't see anything amiss with any of the components.  The other 7 cards in the rig are fine so it's not the power supply.  The reported watts is normal and the total power drawn by the rig from the socket is normal and the card had been running 24/7 for a month or two before I got around to cracking it open to see why it was 10 degrees hotter, so I expect it will continue to run fine especially as it's now only 4 degrees hotter.  Appears to be solely a heating issue caused by a defect in the heatsink or heatspreader, maybe an airpocket, who knows.  Up until a few months ago, it had been running pretty much non-stop for 4 years at a normal temperature.  Interestingly, it's winter here so temperature-wise, it actually occurred when it less stressed than 6 months ago.
member
Activity: 196
Merit: 11
August 02, 2021, 03:55:23 AM
#8
I suspect bad MOSFET, find a magnifier and look very careful on the GPU maybe you will see bursted or weird looking components, if this isn't happening with your GPU before then something is broken in the GPU parts
legendary
Activity: 3808
Merit: 1723
Up to 300% + 200 FS deposit bonuses
August 01, 2021, 08:03:18 PM
#7
Couple things to try.

I had this happen in the past and there are 3 reasons.


First one is obviously the fans are bad. They are spinning and might display something like 3000rpm but they are spinning barely and you just can’t tell. But if you are 100% sure it’s not the fans then...

Second could be a bad PSU. It might register +12V on idle but under load it might deliver less than 12v and your GPU will compensate by increasing the current and adding extra heat.

Third is something is shorted on your GPU. Some mosfet or capacitor and it’s not completely shorting out but enough to raise the temps and still work.
sr. member
Activity: 770
Merit: 268
August 01, 2021, 01:42:42 AM
#6
maybe you need to find out the reason so you can avoid getting the same thing on your other cards. is there anything different before the crater happened? like, doing some overclock or something.
full member
Activity: 258
Merit: 104
August 01, 2021, 01:25:08 AM
#5
Was it always running hot or all of a sudden?

All of a sudden.  Just suddenly started running 10 degrees hotter with the fan going 100%.  Fans are fine, I replaced the original sleeve bearings with sealed ball bearings years ago and they been fine ever since.  For sure, the crater wasn't there when I changed the fans so it has happened while it's been running.  It's like a teeny weeny explosion, damaging the GPU and heatsink.  But it still runs fine.  I will get some fine emery paper and see if I can smooth it down a bit.
member
Activity: 1201
Merit: 26
August 01, 2021, 12:41:09 AM
#4
One of my cards was running hot so I took it apart to see what was going on and discovered a crater maybe 2mm wide and maybe 50-100 microns deep in both the heatsink and the GPU.  It's like some tiny people let off a tiny stick of dynamite and made this not-so-tiny crater.

Card works totally fine except for getting hot.  How to fix?  Sandpaper?
I seen like that one of asus strix. i fixed with mini grinder.carefully grind it so gpu and heatsink will attach good.
legendary
Activity: 3808
Merit: 1723
Up to 300% + 200 FS deposit bonuses
July 31, 2021, 10:45:05 PM
#3
Was it always running hot or all of a sudden? If its all of a sudden, maybe your fans are failing since most likely the crater wouldn't be created while it was in use. I had this issue before with an CPU. It had a tiny tiny crack and for some reason it still worked. I did some googling and there is some layer between the outside and transistors, so if it works, it'll keep working.

However CPUs and GPU chipsets are frigile. I've broke many of my GPUs when the heatsink was stuck and many CPU people broke their expensive intel i7's when they tried to delid it improperly. So be glad that yours is still functional.

Try some thermal paste and make sure the fan bearings are good. Spin it very fast with finger and listen to any noises. 
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
July 31, 2021, 06:58:46 PM
#2
One of my cards was running hot so I took it apart to see what was going on and discovered a crater maybe 2mm wide and maybe 50-100 microns deep in both the heatsink and the GPU.  It's like some tiny people let off a tiny stick of dynamite and made this not-so-tiny crater.

Card works totally fine except for getting hot.  How to fix?  Sandpaper?

photo would help.

i  would not sand the gpu chip.

maybe the heatsink can be buffed flat and load the groove on the gpu chip with extra heat sink paste.
full member
Activity: 258
Merit: 104
July 31, 2021, 06:28:47 PM
#1
One of my cards was running hot so I took it apart to see what was going on and discovered a crater maybe 2mm wide and maybe 50-100 microns deep in both the heatsink and the GPU.  It's like some tiny people let off a tiny stick of dynamite and made this not-so-tiny crater.

Card works totally fine except for getting hot.  How to fix?  Sandpaper?
Jump to: