The device I built is still experiencing frequent crashes.
The device I built is still experiencing frequent crashes.
I bought a PowerSpec G715 from Micro Center recently, but the experience has been extremely frustrating. To be honest, I went for an open-box model, so dust accumulation likely played a big role. At first, everything looked good—XMP and Rebar were enabled, no unnecessary software was installed. However, problems quickly emerged. The machine would suddenly freeze, blacking out every 5 to 10 minutes, followed by error messages about AMD drivers. It would crash after about half an hour, then restart normally but act as though nothing happened. Games like Star Citizen, Halo Infinite, and BG3 would launch but then crash immediately, rebooting the whole system.
I attempted several fixes: updated BIOS, reinstalled drivers, replaced all fans with Noctua models, added extra RAM sticks, disabled XMP, tested different memory speeds, reset cables, changed RAM slots, reinstalled Windows, and turned off Rebar. Despite keeping CPU temps under 75°C and GPU under 55°C, the system still failed to launch most games reliably. Even when diagnosing, I couldn’t pinpoint a clear issue.
I’m determined to figure this out before accepting that I’ve made critical mistakes. If anyone has suggestions or solutions, please share them—I’m open to trying anything before giving up.
You handled the issue well! I’d prefer to save my time and send it back to Micro Center, but if it’s a rare opportunity, I’m willing to give it a shot. From what you shared, it seems the GPU might be having problems—there are steps you can try, though I’m unsure if they’ll solve the problem long-term. (For example, adjusting voltages for short-term stability, but eventually the chip will fail.)
Thanks for the update. It’s tough to gauge how much time I have. The trip back took six hours, and I bought the PC for $1124. I can run Windows stable enough to download a game, but it crashes when trying to load it.
Power source, do you have a spare to test? Consider running the CPU with just one active core and no multi-threading to lower power usage. Cut the overall clock speed by about half and attempt it without the GPU. Make only one adjustment at a time, starting with another power supply before making further changes.
I experimented with some OCCT tools and the app kept stopping. Restarting would show recent data but no warnings. I ran a memory test which highlighted many issues. A colleague from Micro Center suggested using MemTest86, which also reported a serious problem. It looks like a memory-related fault. Possibly more than one factor is involved.
Review the precise details of your RAM and adjust the voltage and timings within the UEFI manually. Check whether the crashes persist. The explanation I provided earlier includes the key information you need.