I am in the same boat as yourself. I believe it is attributed to that XSAVEC issue (a change has been introduced in kernels 5.19+, see https://forum.parallels.com/threads/linux-5-19-issue-with-cpu-capabilities.358140/). This results in many applications crashing -- the most noticeable on Arch are Chromium and anything that uses its rendering capabilities under the hood, e.g. any Electron app. I am currently surviving by using Firefox with hardware acceleration disabled.
I think you can combat the majority of the graphics issues if you revert back to the prl graphics drivers from the recently introduced virtio-gpu, but I think the proper fix should come from Parallels by addressing XSAVEC issue (which all the competitor fixed like an year ago) and also reviewing how their virtio-gpu is exposing its interfaces (there are some issues with the selected IDs for that device, so some libraries cannot recognise it as a virtio-gpu:
Code:
[galaxy@archlinux ~]$ lspci | grep -i virtio
00:05.0 Ethernet controller: Red Hat, Inc. Virtio network device
00:0e.0 RAM memory: Red Hat, Inc. Virtio memory balloon
01:00.0 VGA compatible controller: Red Hat, Inc. Virtio GPU (rev 01)
[galaxy@archlinux ~]$ cat /sys/devices/pci0000\:00/0000\:00\:01.0/vendor
0x8086
[galaxy@archlinux ~]$ cat /sys/devices/pci0000\:00/0000\:00\:01.0/device
0x2981
[galaxy@archlinux ~]$
The device, IMO, is supposed to be 0x1050 for the device if it tries to mimic Red Hat's virtio-gpu, but I am not that knowledgeable in the virtio space, so I may be wrong here.