@c4ppy If it was faulty, I would expect it to fail completely. I would try different monitors and graphics cards.
It appears the R43SL gets all the 3.3V power from the M.2 slot. Wikipedia says 3.3V is always 3A (9.9W) max for PCIe. The PCI Express M.2 Specification Revision 1.0 says 2A for M.2 socket 1 (Key A, Key E, SATA), and 2.5A for M.2 socket 2 (Key B, WiFI) and socket 3 (Key M, NVMe, AHCI, PCIe x4). I guess the 500 mA difference only matters if the GPU needs more than 2.5A from 3.3V. Even if it does, the M.2 socket 3 has nine 3.3V pins, 500 mA each, so maybe more than 2.5A can work (but, about the 4 extra pins, the spec says "it is not intended to increase the current sinking capability of the Module. The intention is to further reduce the IR drop of the power under extreme high current cases and increase the robustness of the SSD devices.").
If the problem is not power, then could it be signal integrety? I don't know. If a signal is corrupted, then I think PCIe would retry it instead of accept the bad data...