7950X3D keeps failing repeatedly without a crash screen – what should I do next?
7950X3D keeps failing repeatedly without a crash screen – what should I do next?
Hey there. I’m trying to figure this out and could use some guidance. Here’s what I’ve got so far:
- 7950X3D PNY 4090 MSI X670E Carbon WiFi (previously swapped, details below)
- Seasonic USA Vertex GX-1000 1000 Watt PSU (also changed from something else)
- 4 Sticks Corsair Vengeance RGB 16GB DDR5-6000 PC5-48000 (DDR5 concerns noted)
- H150i AIO 10 AF120 fans – BIOS and drivers are all up to date
- System has been running for about 3–4 weeks now
- Frequent random crashes, no BSOD or memory dumps, just sudden shutdowns
- Original build uses an ASUS 650E-F motherboard
I’ve experienced issues right after setup. Within a few days the PC wouldn’t power on at all—just a strange clicking noise. I took it to the store and they suggested the PSU might have failed and taken the motherboard with it.
I replaced the PSU with Seasonic and swapped the ASUS 650E-F for an ASUS 650E-E, added four RAM sticks, and tried various settings: Expo/off, PBO on/off, default optimizations, clean Windows install. Still had crashes during everyday tasks. Ran stress tests like Cinebench, OCCT, Memtest—everything passed without problems, but crashes persisted.
Later I switched to a MSI X570E and everything worked until I installed Windows again. When opening iCue it would crash repeatedly. I uninstalled iCue, kept crashing, then reinstalled Windows to ensure it was gone. Then I tested different RAM configurations—from 3200 up to 6000 MHz—with both Expo and PBO settings. Still got crashes.
Performance metrics look normal under load: CPU maxes around 83°C, GPU around 72°C.
I’m stuck. Any advice or alternative approaches would be greatly appreciated. I’ve tried nearly everything over the past weeks and can’t recall what worked last.
He faced similar stability challenges due to RAM speed settings. His setup used a very fast 30-76-76-76 6000MT/sec configuration, which resolved the issues entirely by adjusting timings. This worked on a 7800x3D board but applies to any 7000x3D CPU as well. I typically use a 36-36-36-76 6000MT/sec setup because 3D v-cache is less sensitive to low RAM speeds. I prioritize a higher baseline and steadier frame rates over a marginally better average with more fluctuations. My friend needed CL40 timings to achieve stability on his CPU, though EXPO didn’t help much. The only factor that kept his system running smoothly was the custom high timing settings. His system has remained completely stable, avoiding crashes or BSODs for several minutes.
This specific version isn't included in the QVL. I haven't tested another kit yet. I'm holding off on buying anything else because 1) I'm already over budget and 2) the return period has closed. However, I'm open to it if needed.
This might seem a simple question, but I'm not very familiar with adjusting custom settings or overclocking. Setting the default BIOS options should already configure the timing correctly for stable performance. EDIT: It seems you're referring to the same idea as "optimized defaults" you mentioned earlier.
This would likely be my next step in diagnosing the issue. I’d test a dual-channel 2x32GB configuration from the QVL. It’s worth noting that DDR5 may behave oddly with quad modules on Ryzen 7000 chips, so the kit might not work as expected. You could also adjust timings and memory settings manually, as suggested by @Agall. But if I were using this build, I’d prefer to purchase a fresh kit from the QVL to avoid those complications.
I located a collection of Corsair Dominator (CMT32GX5M2D6000) that's available on the QVL and in stock at my nearby Microcenter. What caught my attention was that all the details match exactly with my existing Vengeance setup, which makes me a bit doubtful. On the QVL it mentions compatibility with 1, 2, or 4 sticks. Even though it's listed there, is it still wise to use four sticks?
I own a 670 board with the 5600 Vengeance memory. Do you have Samsung chips on your system too? (Attached photo of yours) When I was securing it, BIOS updates seemed to assist with stability. Your specific setup matches someone's report on the MSI forum. Unless you're the one experiencing it. Outside of memory concerns—are you sure the power is fully connected to the GPU? Have you looked in Event Viewer for any Windows logs? Did you change power profiles to top performance? Turned off PCIE Bus Sleep? Disabled Ultra Low Power State in the registry for your AMD board to rule out other issues. (Refer to my troubleshooting post for a similar situation)
I have the Samsung chips. Your part number matches mine except for the 5600 versus 6000. * Power cable is connected (I'm worried about overheating). * No errors in the event viewer since these are hard shutdowns (no BSOD). I've checked for correlations with other events, but no clear pattern. * Power profile is set to balanced now; it was balanced at one point before crashing, though it still failed. * I haven't adjusted the PCIE Bus Sleep Modes. As suggested by @Agall, I'm using a CL40 with two sticks to test. If it crashes again, I'll follow your advice.
Quad kits are facing general problems for some reason, starting around the release of 7000. I haven’t spent much time digging into it myself—just rely on professionals and those more connected online who are up-to-date. Right now, I’d suggest going with Dual kits instead of Quads. As for the specs, they might look similar to your current setup, but the internal components differ. Sometimes boards struggle with certain chips like Samsung B-die or Hynix.