In my defense I worked with arch, but mostly headless or simple i3 systems 
I dont think its the nvidia card, by specifying the cards pci id like this in /etc/default/grub
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash vfio-pci.ids=10de:2684,10de:22ba ... you guarantee that the pci device as a whole is ignored based on the kernel parameter, and this is necessary to avoid the host os initializing the card if you want to pass it thru to a vm as a whole.
As for mkinitcpio, I’m not too familiar I just added the gpu pass-thru related modules but I’m under the impression that kms should be handled automatically, and I also see here that its part of the default setup and I dont have it, I’ll add it if the system breaks again, for now its stable.
my /etc/mkinitcpio.conf
MODULES="vfio_pci vfio vfio_iommu_type1"
HOOKS="base udev autodetect modconf block keyboard keymap consolefont plymouth resume filesystems fsck"
UPDATE
I just came back from lunch and everything was frozen and unresponsive except my Kitty terminal 
after rebooting I got the blank screen again.
Added the kms hook, cleared the dirty flag on the boot drive with fsck, no other errors, and it works again.