Hi everyone, I am losing my mind with my setup and I really need some help from the community to get a stable workstation for production. I am a student VFX artist and my Houdini FLIP simulations are constantly crashing.
https://youtu.be/_5p4NeegKb0 Here's an example of when it crashes. My RAM isnt even saturated, there's like 20+ gb that aren't used.
It seems quite unpredictable, since sometimes it crashes after 5 frames, other times it crashes after 50 or 100+, and sometimes it doesn't crash at all.
I already ruled out a RAM issue since I'm not at maximum capacity, and after monitoring everything with HWiNFO64, I am pretty sure my i9-13900KF is the culprit.
My machine is built around that i9 13900k, a Gigabyte Z790 Aorus Elite AX motherboard, 64GB of DDR5, and Windows 11. When the simulation runs, I can see in HWiNFO that my Core VIDs maximum peaks at crazy numbers like 1.55V or 1.57V, and the CPU package temperatures instantly hit 100°C with massive thermal throttling right before Houdini disappears.
I have already tried a bunch of things to fix this. I flashed the latest Gigabyte BIOS to get the recent Intel microcode patches, and I switched the CPU profile from Extreme to the official Intel Performance Settings. Since that didn't stop the crashes, I tried restricting the power limits PL1 and PL2 to 200W, and I even forced the CPU Over Temperature Protection down to 85°C in the BIOS to act as a thermal shield.
The thermal shield actually worked to prevent the PC from melting, as the CPU throttled down to 3.3 GHz to stay at 85°C, which miraculously allowed one simulation to finish. But on the next sims, the motherboard just kept forcing those massive voltage spikes up to 1.57V anyway, and Houdini started crashing again.
With all of that done, it introduced random crashes to my computer. Yesterday, when I was playing Battlefield 6, my computer crashed like 3/4 times for no reason at all. It worked nicely again when I put the default settings back in the BIOS.
At this stage, I honestly do not care about losing ten percent of rendering speed. I just need a machine that is one hundred percent stable so I can actually work. Has anyone with a similar Aorus and i9 setup managed to completely lock the voltage down to a safe ceiling without the BIOS ignoring it?
Since I don't know a lot about that, all these experiments were made with the help of Gemini, but I can't reach my goal. If anyone can help, I'd be glad! Thanks!