New 9850X3D + X870E-P: crashes twice on cold boot, then rock-stable rest of the session, stumped after weeks of diagnostics
Built a new PC about a week ago and it has a very consistent, weird pattern: on cold power-on, it restarts once before the Windows logo shows up, boots into Windows, then crashes/restarts a second time within 5-10 minutes of reaching desktop. After that…
Built a new PC about a week ago and it has a very consistent, weird pattern: on cold power-on, it restarts once before the Windows logo shows up, boots into Windows, then crashes/restarts a second time within 5-10 minutes of reaching desktop. After that second restart, it's completely stable for the rest of the session until the next cold boot, when the same two-crash pattern repeats. I rarely even see a visible BSOD, it mostly just restarts. Specs: CPU: Ryzen 7 9850X3D Mobo: MSI PRO X870E-P WiFi, BIOS E7E70AMSI.2A52 RAM: Corsair CMH32GX5M2M6000Z36, 32GB DDR5-6000 (currently JEDEC 4800MT/s, EXPO off) GPU: MSI RTX 5070 Ti Gaming Trio OC PSU: MSI MPG A850GS 850W SSD: Samsung 990 Pro 2TB Bugcheck codes seen (all point to memory corruption, no common driver): 0x1A MEMORY_MANAGEMENT - most frequent, "corrupted PTE detected" 0x1000007E SYSTEM_THREAD_EXCEPTION_NOT_HANDLED 0x139 KERNEL_SECURITY_CHECK_FAILURE - "corruption of a critical data structure" What I've tried/ruled out: MemTest86, 4 full passes, 0 errors - RAM seems fine even at stock JEDEC timings Driver Verifier running 24+ hrs with all 3rd-party drivers flagged, never caught an offending driver across several crashes Temps via HWiNFO - CPU tops out at 80°C under MemTest86 load, ~67°C normal use, zero thermal throttling logged anywhere SSD - healthy SMART, chkdsk found only NTFS metadata corruption (0 bad sectors), looks like fallout from crashes not a cause CPU voltage - Vcore was spiking to 1.376V stock/Auto. Applied Curve Optimizer (-10 all-core negative), dropped peak to ~1.345V — didn't change the pattern at all The "crashes exactly twice on cold boot, then perfectly stable" pattern feels like it should be a clue but I can't place it. Almost seems like a memory training/POST issue rather than a sustained-load issue, since it never happens once the system's been running a while. Anyone seen this specific double crash on cold boot then stable pattern before? Sound like DRAM training/POST rather than a sustained stability issue to anyone with more expertise? Known issues with 9850X3D + X870E-P specifically? Already filed support cases with AMD and MSI in parallel, but wanted to see if this rings a bell for anyone here before I commit to an RMA.
Collected discussion
Did you try updating your bios, and resetting your bios settings to default and keeping it default settings (no OCs, voltage changes, underclocks etc..)
Yes, BIOS is updated. It crashes with default settings as well.
Did u also try each RAM stick individually? Another thing is since u bought the PC a week ago, r u still in the refund window? That's better than doing RMA since motherboard will be a refurbished model. CPU will be brand new, but idk how RAM RMA looks like