Mobo doesn't accept my RAM settings. D:
Mobo doesn't accept my RAM settings. D:
Realized I was running a generic 650 watt PSU. Must have switched to a Corsair 850 after a major rebuild. Capacitor was damaged, transformer appeared warped, and the glue turned brown. Heat problems existed before the upgrade, but I kept using the same PSU. It’s clearly failing—no tests were done since I could see the issues. Amazon is sending me a certified 850 watt PSU today, one that’s new, reliable, and has good headroom. Originally, I assembled my system gradually, changing parts occasionally. This setup had been stable for years, but it was always somewhat unreliable. Recently, after playing Tarkov, I noticed random PC restarts—sometimes monthly, sometimes weekly. This occurs 99% of the time when streaming on Discord and playing Tarkov, rarely just during gameplay. After weeks of adjustments, BIOS updates, and tweaks for Discord, I ran an SFC check. Two Bluetooth system files were corrupted, but that was resolved. Memtest86 showed no problems. Still, it keeps happening. Discovery: My motherboard compatibility list lists CMW32GX4M2E3200C16 as valid, but only for 1|2 DIMMs, not 1|2|4. In CPU-Z I saw Slots 1 & 2 are dual, while slots 3 & 4 are single. This seems to be the cause of my instability and occasional slowdowns. Not fully compatible with a four-pack mixed dual/single banks. Could be worth considering a different RAM configuration—maybe 32x2 or another 16x4 setup. I’ll stick to a single RAM package that fits my needs, double-checking the QVL list. I’ll also boost the RAM speed if possible. My specs: CPU Ryzen 5 5600X, motherboard MSI Mag B550 Tomahawk Max (V29 BIOS), RAM Corsair DDR4 (CMW32GX4M2E3200C16, 3200, 16GBx4), GPU Radeon 6650XT.
AMD setups generally prefer two components over four. The four could even fail to reach the claimed performance. However, if Memtest finds no issues, everything seems okay.
I tried the built-in Windows memtest myself. I'm using Rufus to install memtest86 on a USB drive now, and I'll run that version to check performance. I've done some more digging, but it looks like QVL rankings are pretty old and unreliable. I spent several hours trying to find the exact RAM type mentioned in those lists, which was tough. Maybe I should consider replacing my power supply. I've used it with my 3600X, 5600X, and even the 1660 TI Super. I think it's a 750-watt unit (I'll confirm), but that should be fine. If it's failing slowly, now could be a good time to upgrade to a higher wattage one.
Ensure dual-rank sticks remain in the initial positions. Adjust timing settings via the BIOS and verify using Aida64 (SPD). Then apply the configurations in the BIOS.
This was the outcome I reached after digging deeper. I planned to run memtest86, but first I checked my PSU. I didn’t realize that during an upgrade with a 1600x, I required more power, so I swapped in a used PSU from a friend—SolidGear 650 watts. This might be the issue, a second-hand PSU I acquired in 2021 and reused when upgrading from 1600x to 5600x and from an 1660ti S to a 6650xt. I’m puzzled why it held up so long. A new 850-watt MSI PSU will arrive soon. It seems to explain the stability, as my calculations suggested 550 or 600 watts at that point. With a 650-watt unit, pushing the GPU to full capacity and overclocking the CPU adds little margin. Thanks for your help.
Slots 3 and 4 are not single. You likely have issues with one stick on one bank. All 4 sticks should say dual channel . Nevermind, you talking about rank, not channels.
I experienced RAM problems (random crashes, WHEA warnings) and memtest detected no issues even after two days. Reducing speed didn’t fully fix it. Replacing the RAM (B-die) resolved the problem completely—no crashes or warnings afterward.