So how can you prove, with actual data, that this is the fault of the card being unstable and not a coding/design fault of some sort? That's the problem. We have no way that I know to test this, I would be glad if there was one, Nvidia surely must have apps to properly stress test cards. It is very suspect to me that regardless of the stress I put on the card only Dx12 TDRs, always TDR, sometimes the app crashes after losing the GPU device. And there's no dmp in LiveKernelEvent, there should be one, isn't it? I remember this was the case on previous TDR stuff when Manuel asked for the dmp files, but again none were found where they should have been. Annoying. What we need is a proper way to diagnose what's going on so we don't rely on speculation. If you look at forums, basically all the TDR stuff is Dx12 related for some reason.