Frequent BSOD issues, application failures, system reboots following RAM replacement and testing.
Frequent BSOD issues, application failures, system reboots following RAM replacement and testing.
Hey! Last year I upgraded my rig by assembling a new PC with individual components, and since then I faced a lot of constant issues (system freezes, total crashes, BSODs, app failures). After trying many solutions—driver updates, different Windows versions, CMD commands, stress tests for CPU, GPU, RAM, BIOS changes, even a new motherboard—I figured my RX 6700 (not an XT) was the problem. I sent it back to the store I bought it from and got a replacement GPU, the RX 6750 XT. After that, the system became much steadier and the issues disappeared. Until now. Last week I found the same RAM I’d bought before at a good price, so I bought another set, intending to add more RAM later (I only had one 2x8GB stick before). RAM issues keep popping up: I install it next to my current ones and it causes crashes and freezes after just two hours. Initially, I thought it was bad RAM, so I ran it through MemTest86. The results were alarming—about 40 errors in a single night. What’s odd is that all addresses and sectors showed errors across every slot, except the last one. After some testing, I removed the faulty RAM and ran another MemTest86 with just this problematic stick. It passed cleanly—no errors. That helped confirm the problem was indeed with the RAM. Now I’m back to my old setup, but things are still unpredictable. On the first day of installing the new sticks, I experienced Windows crashes and system freezes during a two-hour cooldown. Eventually, I realized the RAM was likely the culprit. So I started testing it again with the faulty stick, and after a full night of use, the test passed completely. That’s when I understood: the errors were everywhere, not just in one place. Finally, I tried everything again with the same configuration as before—same slots, same setup—and after 10 hours, it worked flawlessly. Still, I’m left wondering: was the RAM the real issue, or something else? After all these tests, I can’t pinpoint exactly what’s wrong. But I’m ready to give up and stick with the current setup, hoping for a stable fix.
I've gathered the necessary outputs from SysnativeBSODCollectionApp and Perfmon Reports. I've included additional screenshots, two full memory dumps, two mini dumps, and Event Viewer logs exports. The program would get stuck when running Perfmon /report, pausing for 60 seconds. I located the relevant files in "C:\PerfLogs\System\Diagnostics" and copied them as requested. Please let me know if you need anything further. Thank you all.
Summarizing, the device is designed for two sticks only, not four. The timings in the DOCP profiles reflect this setup. You can operate the kit without DOCP(xmp) enabled, using auto speeds instead. If that works, adjust frequency, voltage, and primary timing manually rather than relying on DOCP.
Really? I checked the manual and found a RAM setup chart, but after looking it up online it seemed fine. I didn’t realize it wouldn’t work at 3600 MHz, especially since my rig isn’t that high-end. I’ve also checked sites like Corsair and Kingston, and they listed 4 RAM kits without any issues. This chart comes from the motherboard manual, but it only mentions 3200 MHz and works with older CPUs like Matisse and Renoir. Since my CPU is Vermeer, I thought it was outdated and not very helpful. But when I searched for the RAM model myself, people confirmed it’s supported.
We're discussing the RAM kit, and it seems there might be some confusion. The issue is whether the RAM is compatible with your CPU and motherboard, even if there are available slots. It's unclear what kind of anti-consumer upgrade feature is in play, which makes it hard to determine the cause of the crashes. I plan to test the RAM and run additional diagnostics, but so far the performance remains stable despite long sessions. I haven't received a clear result yet.
The specifications, timing details, and speed settings are precisely what made me choose this exact model. The XMP profiles and timing data work seamlessly across all sticks, which is why I went with the identical kit.
Some factors to consider here. Operating with four sticks, as detailed in the manual, might contribute to the problem. Using four sticks places a significant strain on the memory controller (part of the CPU). I don’t see your setup listed on the motherboard’s memory QVL. I haven’t paid much attention to the QVL since the second generation of Ryzen, when RAM issues were common. Still, with four sticks you should ideally use the QVL if available. It indicates whether the system passed with two or four sticks. They evaluate all kits regardless of stick count, even single or dual sticks. If it isn’t listed, it’s a matter of chance. Purchasing the same RAM model as your existing setup doesn’t guarantee identical performance. Before each production run, the RAM supplier selects the lowest-cost components meeting required speeds and timings. In subsequent runs, the RAM could differ substantially from the original batch. This complicates the QVL because it doesn’t specify which parts were used, leaving you to verify after purchase. The QVL usually reveals the manufacturer of the components tested with your specific kit. With Corsair RAM, they display which parts are printed on each sticker. This information is encoded intentionally to prevent disclosure. On the sticker, you’ll find a version number—not the actual model—but it identifies the part origin and the chipset version used. You can observe in the screenshot that all SK Hynix kits begin with a 5. If they lack this version, they’re less similar to any other random kits. An example of the memory QVL for Corsair sticks on your board is shown alongside your CPU. The Chip field lists the component manufacturer, including the version behind the model, and the number of sticks it supported in the "DIMM Socket Support" section.
For instance, even with this particular RAM (the first one, Samsung C), it appears on the QVL list. If I skip the 4 sticks kit and choose the 2 sticks kit instead, once I have more funds I can purchase another 2 sticks kit. Even though it's marked as verified, since I place the order later from a different RAM batch, I might end up with a Hypnix and still face the same issue.
Mixing is possible but it greatly lowers the likelihood of success. I mainly notice it failing, so my perception of how often it works is likely biased negatively. EDIT: The older unit includes components from Nanya, while the newer ones feature parts from Micron/Spectec. Micron acquired Spectec and it's unclear if they still use the same part codes now. The figures after the period indicate ranks (possibly not exact, just related to memory chip layout and chipset versions like C-Die or M-Die).