Which issue is it: GPU or Mobo?
Which issue is it: GPU or Mobo?
Hey there, I'm not completely sure where this fits—please forgive if it goes somewhere unexpected. I have several questions about this new setup I just assembled. There have been some problems that aren’t clear to me, and none of my friends agree on what’s happening or what caused it. Sorry for the long explanation; I’m really anxious about this.
First, here are the details: The system includes a Ryzen 7 7800X3D with ASRock Steel Legend X670e, 2x16 DDR5-6000MHz RAM, Scythe Fuma 3 Sapphire 7900XTX Pulse (switched to ASUS TUF 3070ti after some issues), an NVMe 2TB SSD (2TB crucial), and a Samsung 500GB SSD. The case is Lian Li Liancool 216. I built the PC on December 19th, and everything worked fine then.
When I first booted, the system reported OK. I changed the RAM speed to 6000 in BIOS, saved, and exited. The PC went dark but didn’t restart or shut down. I thought it was strange and couldn’t fix it, so I tried turning it off completely by holding the power button. Then I checked connections—case fans weren’t connected. After a second boot, loading took ages, but the manual said it should be fine on first try (this was my second attempt).
Overall, there were three attempts that took a long time or forced me to shut it down. Now it boots normally and loads without problems. I noticed before any big issues that the GPU temperature difference between global and hotspot was only 20–30°C during games—usually around 95°C max. The main concern is this: a few days later, I’d pause mid-RDR2 for 1–2 minutes, come back, and the PC would be off. No RGB, no lights, even in sleep mode. I tried turning it back on, but nothing happened. I pressed the power button a few times to clear static, then tried again. Sometimes it makes a tiny *click* before failing completely.
Third time it worked—no issues, just a faint *click*. After that, everything was fine except for a burning smell in the room. The PC still smells like smoke but no smoke comes out. I plugged the GPU into my integrated graphics on the 7800X3D, and Windows started up. Still, there’s a lingering smell.
HWMonitor shows normal temps, but it can’t detect the 7900XTX anymore. I turned off the PC, removed the 7900XTX, and swapped it with an old video card (R9 360). The new one worked—PC started up normally, but the screen had weird multicolored pixelation before the ASRock logo appeared. After that, everything functioned properly with HDMI connected to the R9. HWMonitor just says “generic microsoft graphics.”
After being really worried, I went ahead and ordered a refund for the 7900XTX and a replacement card for the Steel Legend (which I canceled soon). My friends who are more experienced said it was probably a problem with the 7900XTX itself, not the motherboard. I didn’t see any obvious damage on the board, and there was no smoke. The smell came from the GPU, which confirmed when I put my nose close to it.
I tried reducing heat by undervolting and underclocking in MSI Afterburner, but the problem returned after a while. I’m considering keeping the 3070ti for a month or so to see how it handles this build before deciding on a GPU upgrade. What do you think?
The instructions need updating; posting after adjusting memory settings can take a long time because AM5’s memory optimization process is slow. If you increased the memory speed, that might be the cause. Quick question—did you just change the memory speed or activate XMP/EXPO? It’s better to enable XMP/EXPO since it sets timings and voltages for improved performance. Regarding the system not shutting down after BIOS changes, this isn’t uncommon, and a BIOS update should likely resolve it. As for why the GPU failed, I’m pretty sure it was a defective card. There’s usually no way a motherboard can damage another component directly, as PCIe slots don’t handle major voltage shifts. Even if it happened, the 3070 Ti and R9 390 worked fine without issues, making this unlikely. With most parts (or complicated systems), there’s a risk of early failure—often due to manufacturing flaws. If a card dies when idle, it’s rare but possible, especially with faulty power stages. Components failing at idle are uncommon, though a bad batch of power stages could cause problems for certain GPUs.
I believe I was only concerned with speed and didn’t search further. My immediate thought was no, but it stopped working during gameplay. When I returned, the system was off. It’s unclear if this is typical or just a rare case. In this situation, it was likely under heavy use—especially with ultra settings in RDR2. Your observation makes sense; the GPU seemed to cause the shutdown.
It seems there might have been a misunderstanding earlier. A malfunctioning part failing under stress is actually quite frequent (though not extremely rare), which aligns better with the scenario you described. The reason I interpreted it differently was about trying to reactivate the system after it shut down, which points to a specific kind of issue. Still, now I’m more certain it’s just a dead GPU.