Error detected with RTX 4090 graphics card in a water-cooled system.
Error detected with RTX 4090 graphics card in a water-cooled system.
Alright, let's talk about this issue together. I'm here to help you figure out what's going on with your watercooled PC. You shared a Reddit post with all the details, which is great. The problem started after reassembly and running for about ten minutes—everything seemed normal at first. But then the monitors stopped working and the RGB lights froze up. After a restart, the mainboard error light stayed on near the VGA port. You tried fixing it by adding the mainboard speaker and adjusting the sound settings, but the video card error kept happening.
It seems like the watercooling setup might have caused some issues, possibly due to leaks or improper connections. The good news is that the fittings didn’t leak, so there were no water spots on the PSU header or card. You also mentioned the APU (7950X) had an APU, and you removed the power from the card before installing the HDMI. Even after doing that, the error persisted.
The GPU isn’t showing any picture, which is confusing since everything worked fine at first. The thermal paste and pads were changed during assembly, but you used high-quality materials to avoid damage.
I’m not sure what’s causing the video card error, but here are some ideas:
- Check for any loose connections or moisture in the cooling area.
- Ensure the power supply is properly seated and clean.
- Try a different watercooling unit to rule out compatibility issues.
- If possible, inspect the motherboard for any signs of damage or corrosion.
Let me know if you want help with troubleshooting steps or more details!
Everything appears intact, no visible leaks detected. This seems like a potential board issue. Can you try updating the BIOS or reverting to an older version? I'll check what options are available. The APU isn't responding well either—this isn't a straightforward fix, but booting with just a single RAM stick might help. Do you have any spare hardware you can run tests on? Rebuilding the tubing could be quite challenging.
The white sections you notice appear to have an issue with density. It looks like water flows through under high pressure but leaks very little. Initially, I placed the clothing underneath it; this should resolve the problem once fully new pass-throughs are installed. However, nothing dripped down and the clothing remained dry. The APU seems to be the oddity here too. I tested it with just one RAM module, which still failed. I can't understand why the APU wouldn't function when the GPU is in the PCIe slot, but it didn’t work even when I held it out of the slot. (When I first assembled the PC, I had to install the Nvidia drivers for the card to work, and then the APU functioned while the GPU was installed.) Yes, I got an air-cooled SFF-Build with a 2070S. For quick GPU testing it doesn’t need cooling—just remove it from this system and put it in another. Does that make sense? The testing rig is about 120km away, but I’ll head there tomorrow. Since the APU isn’t working, I suspect something else is amiss. Rebuilding the tubing will be a pain because this happened recently. I have no idea what to do next.
I hope this outcome isn't so bad; the disappointment could be huge and I might never go back to building PCs at a watercooler again. Overall, I really enjoyed planning and constructing the project. It was even more satisfying than assembling an air-cooled system—probably because there are more things to consider, haha. I expect to test the GPU on another setup in about two to three hours, and I hope it functions there.
Are there any concerns about using too many thermal pads? Even though the XG7 is heavily taped with pads and I just replaced them, the card has a lot of VRAM and a large chip, which explains why so many are needed. I looked up compatible options and chose 1mm and 1.5mm sizes, thinking it should work. However, I used high-quality Odyssey 12W/mK pads that shouldn’t melt, especially after a while.
Thanks for your reply! I'm still a bit delayed because of illness and some technical problems, but hearing this is reassuring. From checking the posts, I noticed others mentioned the 1.5mm pads are too thick, so I might switch them to 1.0mm now. At least the GPU seems to be functioning again—this was caused by a mix-up with three cables (the Strimer didn’t supply enough power and two HDMI/DP cables somehow triggered the issue, which no one predicted). Right now, I’m experiencing high temperatures again, so I’ll replace the pads.
We're aiming for the right device to monitor those readings. Which sensor model are you checking?