As you can see in the second image, VRAM test is running yet even though the VRAM should be running at 8400MHz, NVIDIA has decided in the name of power efficiency not to allow us to set a fixed frequency and instead allow the card to "idle" itself. The result is, OCCT is acting like its testing my vram, in reality its not. Its just testing the idle vram frequency of 5000MHz which in fact will be stable 365 days straight, What I need to know if its stable is at 8400MHz, which its not actually testing as you can see.
Similar issue with the first picture, the GPU can only be tested up to 0.925mV. Anything higher results in the card hitting power limit, therefore throttling to a lower voltage, and frequency, thus creating a false sense of stability, by essentially running the test at a lower more than 10000% stable frequency. If I try to find the higuest clocks I can sustain at 1.0mV for example, the card trips out will never actually hold 1.0mV in this test as that would put power consumption >400W, limit is somewhere around 260W, so It would just jump around lower voltages, never quite 1.0mV, therefore again never actually testing what I'm trying to test. Another thing I noticed which maybe you guy have a better understanding of, sometimes when I click the test, some type of protection system is tripped and the card quite literally goes into limp mode, throttles to like 0.742mV 120w, 35c basically zero reason to be throttling it just throttles because it feels like it. System restart, puts it back to normal.
All in all what causes both of my issues? The inability to set a fixed gpu core and vram frequency and a fixed set voltage.