Unusual actions from the PC following the water cooling leak.
Unusual actions from the PC following the water cooling leak.
Yesterday I found a tiny leak while working on the PC, so it wasn't running. Water flowed between the CPU socket and RAM slots, some splashed onto the RAM modules and traveled down the mainboard to the chipset. I repaired the leak and checked for any other leaks. Then I carefully dried it with paper towels and used a hair dryer to finish. Because I didn’t want to take out the motherboard, which would have meant taking apart most of my cooling system, I stopped there. A day later I reinstalled the RAM, CPU, and GPU, then powered it up. Everything worked fine so far. Later, my laziness caught up with me. I got a first BSOD displaying the message UNEXPECTED_STORE_EXCEPTION. A quick search revealed this could stem from storage issues or a GPU not fully seated. I reinserted all internal components and still saw the error, so I removed the RAM and ran MemTest86. It found errors on one of the four RAM sticks, so I replaced that one and completed a full test, which took about four hours. Now the main problem appears to be resolved. The next issue was my second drive, which mostly stores game data. It occasionally stopped working. It still appears in Explorer but fails to open folders when accessed. When I play a game from that drive, it crashes immediately or continues but new files won’t load. After removing both SSDs, reseating them fixed the problem. Edit: Still not resolved. The next occurrence was that every time I opened a 3D application—games or anything demanding GPU power—the PC would reset instantly. No BSOD, no freezing, just a quick restart. The Event Viewer didn’t help much except pointing to an unexpected shutdown. I tried DDU, updated drivers, reset BIOS and chipset, did multiple resets, even a fresh Windows install. The only game I’ve downloaded so far doesn’t crash anymore, and running FurMark with CPU burner also doesn’t cause issues. Hope this helps. Edit: Still experiencing the same problem. The most persistent issue is that, regardless of where I connect the water pump or how I configure BIOS, it only responds to GPU temperature. The strangest part is the many temperature options available for the RPM curve, yet the GPU remains invisible. The BIOS update and removing the CMOS battery didn’t help either. This suggests the main damage from water might be to the motherboard. That explains why so many components started failing, even though they weren’t directly affected by the water."
It looks like a chipset malfunction, mainly involving PCIe connections—your storage and GPU issues suggest NVMe drives are present. You’ve tried moving the drive to a different PCIe slot, which is a good first step. Regarding memory sticks, do they consistently fail in every slot? It’s possible water exposure damaged the northbridge around 2008, linking these problems to that era. You’re likely using hardware that’s significantly newer than that timeframe, so the issue probably lies in a shorter or broken trace near the CPU socket.
My PC is quite recent, with all the specifications listed on my profile. Yes, the RAM modules are causing issues in at least two slots, and both of those slots can accept other sticks. I’m fairly certain the problem lies with one of the RAM modules itself. Since both my drives are NVMe and I only have two slots, I can’t easily swap them out. Doing so might cause my Windows drive to disconnect unpredictably, which could quickly result in a corrupted operating system.
Many issues might be linked to water entering the CPU socket. Did you remove the CPU and let the socket dry? Most northbridge features are built into the CPU itself, not the chipset—though the southbridge was moved to the chipset. This looks like water in the CPU socket, which likely explains the memory and 3D errors since neither component is related to the PCH.
It would be simple for moisture to reach the block. If it reaches the PCB close to the RAM, it could easily pass beneath the socket. Moisture is effective at entering tiny gaps and would only need a small amount to disrupt a pin in a socket.
It seems like a mainboard swap is the right path. The problems with my second drive failing unexpectedly and the PC shutting down during intense gaming also returned. I adjusted some BIOS settings, but even with RPM control disabled for the pump or CPU fans, the GPU still manages the load. That makes me wonder why this isn’t a BIOS setting by default. The motherboard and firmware clearly detect GPU temperatures in real time... Right now I’m also questioning my water cooling choices. For my next full upgrade, I might go back to air cooling.
So I replaced my motherboard with another used one. It's the same model, and so far everything seems to be working again. GPU load doesn't cause a system reset, the SSD doesn't disconnect anymore, and the pump isn't weirdly tied to GPU temperature anymore. Unfortunately, it turns out that one RAM stick is still broken, so it wasn't a motherboard issue. G.Skill already approved my RMA though, so I'll send the kit in and probably get a new replacement. Seriously, G.Skill support has been an amazing experience every time I have had to contact them so far. The last "weird" behavior that still happens is that the pump speed is still not controlled by the CPU temperature, no matter what settings I use in the BIOS. At least it's now decoupled from the GPU and I can set a manual speed, so it doesn't ramp up and down like crazy, but I'd still like it to run a bit faster when my CPU has to work hard. I made another thread about CPU temperatures after my upgrade to a 5800X3D, and people said it's normal for these X3D chips to hit 90°C under full load, regardless of the cooling solution. So maybe it's intentional that the CPU can't control the pump speed anymore? I was thinking that the newer 7000 series CPUs are also designed to use all the thermal headroom, so could it be that AMD doesn't want the pump/fan speeds of the CPU to jump around all the time?