Maybe it's related to the motherboard.
Maybe it's related to the motherboard.
I recently upgraded my computer by swapping the CPU, GPU, and PSU to fully utilize the AM4 platform. I replaced my R7 3700x with a R9 5900x, upgraded my 3060 Ti to a 3090, and changed the RM850x to a RM1200x. The problem I faced was that during intense GPU-heavy games, the PC would shut down automatically after about 20 minutes. After investigating, I found that the R7 3700x and 3090 worked well together, though there were performance limits in certain titles. However, the R9 5900x and 3090 pairing caused frequent crashes while using the GPU. I have an AORUS B550 Elite motherboard (version F19d, 2024 release) with 32GB of 3600MHz RAM. My question is: Could this be a motherboard issue, making it struggle with such a powerful setup? I updated the BIOS and installed all drivers, but the crashes persisted. The error in the event log read: WHEA log 46, linked to the CPU’s memory cache. I’m curious if the motherboard might be the root cause rather than the CPU itself. Thank you for your help, and feel free to ask if you have more questions.
It might involve several factors, though I suspect the 5900x adjusts a CPU cap, letting your RTX 3090 consume sufficient power to trigger OCP or similar limits. To check this, use tools like MSI Afterburner and lower the TDP% from 100% to around 80%. If it still doesn't work, the issue likely relates to high power draw from the card—possibly overloading the PCIe bus (max 75W) or a faulty power supply. A power supply problem could stem from internal faults or overheating, which can be verified by ensuring proper airflow and checking for unusual noises or temperatures. This scenario might resemble a WHEA Log 46 behavior, leading to a BSOD rather than a normal restart.
I've tested two different power supplies, which suggests the issue isn't with power supply itself. I also attempted the 80% TDP setting, but it caused crashes. Out of 15 crashes, two were BSoD, one was too fast to notice, and another showed a WHEA_UNCORRECTABLE_ERROR. I've switched to PCIe 3.0 before, as it resolved USB port problems with my 3060 Ti. All my tests were on PCIe 4.0, though I tried PCIe 3.0 once. No overheating signs were found—CPU max was 83°C, GPU 78°C, RAM 43°C on Memtest. Overdrawing the PCIe bus seemed a potential cause I hadn't considered.
The 5900X offers similar performance to a 3700X in gaming, the RM850X stays stable before reaching 1000W+, and it handles sudden current spikes well. I’d prefer the CPU to be bad if the system runs smoothly with the 3700X.
Think about it, simply changing the CPU in Windows isn't always smooth. Have you reinstalled the chipset driver after installing the 5900x? It's possible you're fine with the old one, but not with the new one. Try removing the previous driver first and then installing the newest version.
I updated the system with fresh Windows and removed any outdated drivers. I followed all the recommended driver installations from the AMD support site, but still faced some issues. I’m thinking about sending a return merchandise authorization for the CPU, as the motherboard has been functioning well for over two years.
Additionally, the motherboard recognized the CPU adjustment and reinitialized the FTPM.
You're worried about bending the pins on the AM4 CPUs, especially with the 5900X. It's understandable—those pins are delicate! Make sure you're careful and avoid any stress that could cause damage. If you notice bent or missing pins, it might be best to avoid removal until they're fixed.
Every connection works flawlessly. In another article, someone mentioned having faulty silicone on their CPU leading to problems when the GPU, CPU, and RAM were active together. I've run Prime95 on the CPU with RAM at full load without issues. The GPU has been tested for hours and showed no problems. The only times the system restarted were during intense gaming sessions like Ghost of Tsushima or Vermintide 2. On Skyrim and AC3, everything ran smoothly since the GPU was only at around 30%.