For the past three or four months, my Windows 11 PC has been crashing with seemingly random BSODs anywhere from once a week to once a day. It has happened while using Premiere, opening a web page or Windows window, switching browser tabs, and occasionally while gaming. There doesn't seem to be a consistent trigger, and the error code often changes. I've seen IRQL_NOT_LESS_OR_EQUAL (0xA), 0x1E, 0x133, 0xF7, 0x20001, and 0x3B, with 0xA and 0x133 appearing more than once.
My system has a Ryzen 7 3700X, MSI B450 Gaming Plus Max motherboard, 64 GB of Corsair Vengeance LPX DDR4-3200 memory, a Sapphire RX 7800 XT, a Seasonic 750 W power supply, a 1 TB NVMe SSD, two SATA hard drives, and a 2 TB SATA SSD. The CPU, motherboard, primary SSD, and first memory kit were purchased around 2019–2020; the graphics card, power supply, and second memory kit were added last year.
After installing the second 32 GB memory kit, I could not run all four modules at their advertised 3200 MHz speed, so I reduced the memory speed to 2133 MHz. The crashes did not begin until several months later. I also occasionally experience graphics driver crashes in certain games, especially when ray tracing is enabled, although driver updates have not resolved that issue.
CrystalDiskInfo reports no obvious drive problems, with the main SSD showing 74% health, and Windows Memory Diagnostic found no errors. I have several recent minidumps available. What would be the best way to determine whether the problem is caused by defective RAM, the CPU's memory controller, storage, drivers, or something else?
3 Answers
Since the second memory kit never worked reliably at 3200 MHz, I would treat the four-module setup as a major suspect even if the crashes started later. Two separate kits that look identical are not guaranteed to work together at their rated speed, and filling all four slots puts more stress on the memory controller. For troubleshooting, use only the original matched pair, keep the BIOS at conservative default settings, and update the motherboard BIOS if it is significantly outdated.
If the crashes stop with the original pair, add the other kit back only after confirming stability. You may ultimately need to run all four sticks at a lower speed or replace them with one matched 64 GB kit.
The different bug checks still point toward memory corruption. That does not automatically mean the RAM itself is bad—possible causes include a faulty module, an unstable four-DIMM configuration, the CPU's memory controller, or a memory-related driver. Windows Memory Diagnostic can miss intermittent DDR4 problems, so passing that test does not rule RAM out.
Remove any overclocking or undervolting, including memory profiles, and test with only one stick installed at a time. Run the system normally with each module separately and test different motherboard slots if necessary. If one stick consistently causes crashes, it is probably defective. If both sticks crash independently, the motherboard or CPU memory controller becomes more suspicious.
The storage drives are less likely based on the dump pattern; storage-related crashes usually identify a disk or storage driver more consistently. The graphics crashes may be a separate issue, especially since they are tied to particular games and ray tracing. For the BSODs, first disable all CPU and RAM tuning, install current chipset and motherboard drivers, and test each memory configuration for several days.
If you want a stronger memory test, use a bootable test such as MemTest86 and let it complete multiple passes, but also rely on real-world testing because intermittent faults can escape synthetic diagnostics. If the machine crashes with known-good memory at default settings, the CPU's memory controller or motherboard becomes a stronger possibility.
I had not considered that a previously stable memory configuration could become unreliable later. I will test the original matched pair by itself and compare the results before changing anything else.