Noctua NT-H1 Update offers excellent insulation properties, making it highly non-conductive.
Noctua NT-H1 Update offers excellent insulation properties, making it highly non-conductive.
The problems showed up and then vanished. Adjusting the GPU a bit with the GPU helped. It worked well for about a month. Once, it popped up at 710 once. Everything functioned normally afterward. There were no DRAM timing issues that time. It stayed trouble-free for another month. Recently it happened again. I found some YouTube videos on fixing GPUs. One of the tips is to wipe it with contact spray. They say dust and similar stuff can alter the dielectric properties and lead to signal loss. I removed the cooler, cleaned everything, and even sprayed around the interposer on the GPU chip. Surprisingly, a lot of grey cooling paste came out. Now it’s running perfectly again. Hope this is the final update on using that old silent cooler. :-D
Yep, record time without issues. But happened again today. Cleaned out again with contactspray. This time, also removed the plastic backplate and cleaned out the other side of the PCB. So the 1 000 000$ question. What causes this failure interval. Do not think it is software (bios etc.) related. Is it the cleaning? Or is it the process of removing the card and re-inserting it? - Did that today. Removed and reinserted card without any other action. Did not resolve the issue. Blowing it clean with air. Nothing. - When I clean it, it is out for a long period. Full powerdown? Can't be it. Computer is completely disconnected from the wallsocket at night. Should do the same. So looks like it is related to dismanteling it and reassembling it. The process: - Unscrew 4 screws of the cooler. - Disconnect cooler from card - Remove cooler - contact spray (but it was still very clean). (Under interposer and DRAM chips) - Cleans out dirt, similar to compressed air - Prevents condens on the surfaces. (No condens issues in this house) - Dissolves oils etc. - Clean up excess contact spray with paper towel. - Mount back on the cooler. Not easy to allign the screws to the mounting holes. Takes some shifting around. - Damaged solderball on the interposer? Pressure makes contact good again? Powercylcing causes temperature cycles that slowly breaks the connection again? Explains the slow degrading. Weeks nothing happens. Then the long boot times start to occur occasionally until it does not boot at all... - Next time, the card goes into the oven at 200°C. Risky move. Anyway, currently one way to circumvent the issue is putting the machine in sleep rather than shutting it down. See me in a month or so.
I solved it. I added new RAM and the issues came back. This time it wasn’t the GPU. It was a loose SATA connector. I figured it out by removing extra parts one by one. Playing around with the GPU might have shifted the cable, causing a poor connection, but not severely. Fascinating!