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?
In the worst situation it remains stable, just a touch slower. Even with the 3D v-cache's reduced reliance on it, performance will drop slightly, though not significantly. In games this difference is unlikely to affect overall speed. It’s the trade-off my friend chose with his 7800x3D setup, facing timing challenges with DRAM.
Now I'm going to prioritize a bit more performance. To be real, the 7950x3d and 4090 are too much for what I play (MSFS2020 in VR), so a few frames per second won't stop me. Just tired from all the testing, trying things out, and the costs. I just want to use it.
I also suggest configuring the CPPC to 'cache' mode. It’s typically found under advanced CPU features in the UEFI. This turns off the trash 3D v-cache optimizer driver and lets you manage it directly through the UEFI. It assigns heavy tasks to CCD0 while lighter ones go to CCD1, but unlocks full operation when a demanding workload appears. Gigabyte’s adjustments here are quite effective; they outperform the automatic setting that relies on the driver. It’s worth testing since software glitches in the driver could also affect stability and parking performance.
Unfortunately, the CL40 didn't assist. I tried booting Steam just for the Steam client and it crashed.
Machine is shutting down unexpectedly? No error screen displayed.
It might be related to 3D accelerated applications. Have you attempted to run without the 4090 and instead used the iGPU? I usually need to enable PCIe 4.0 on the 4090's slot to avoid strange boot problems. My friend also has a 4090 but with different issues. I experienced random restarts or BSODs while streaming YouTube without a game running, possibly due to scheduler problems. Make sure you're using a balanced power plan as well.
haven't tested without just the iGPU. I actually turned it off because I thought there might be problems. One thing I realized was that with my two 2x16GB DDR5 kits, I probably mixed them up. I think you should keep them together. I saw that the two sticks in the system didn’t have consecutive serial numbers. In fact, I had fake serials like 101 and 103 in one place and 102 and 104 elsewhere. I put 101 in A2 and 102 in B2, assuming sequential numbers mean they came from the same kit.
It really shouldn't be an issue, but a 4 DIMM setup can pose challenges based on the motherboard wiring. Some boards use daisy chained connections while others have dedicated traces for each DIMM, making it hard to identify the correct one without tools or manufacturer guidance. If possible, I'd place two sticks in the primary slots—likely the second set—and remove the 4090 to see if a stable setup emerges. You might be dealing with several faults, so reducing variables is key to isolating the problem. Begin with the simplest configuration and gradually add components until instability appears. This approach helps pinpoint the fault rather than aiming for full stability.