Here’s what I decided to do: I just created a test win10 qube using this command, which worked beautifully. After everything was done, I decided to add the PCI passthrough to it to see what would happen. Unfortunately, when I started it, I got a blank window, and eventually qrexec timed out and killed the qube.
From examining the logs, it appears I am having the same problem outlined in this thread:
Basically, Xen’s “BIOS-provided physical RAM map” has a reserved area:
Xen: [0x00000000000a0000-0x00000000000fffff] reserved
and the GPU’s BAR 6 attempts to allocate space in that area:
pci 0000:04:00.0 reg 0x30: [mem 0x000c0000-0x000dffff pref]
…
pcifront pci-0: claiming resource 0000:04:00.0/6
pci 0000:04:00.0: can’t claim BAR 6 [mem 0x000c0000-0x000dffff pref]: address conflict with Reserved [mem 0x000a0000-0x000fffff]
pcifront pci-0: Could not claim resource 0000:04:00.0/6! Device offline. Try using e820_host=1 in the guest config.
…
[00:09.0] xen_pt_realize: Assigning real physical device 04:00.0 to devfn 0x48
…
[00:09.0] xen_pt_register_regions: Expansion ROM registered (size=0x00020000 base_addr=0x000c0000)
…
[00:09.0] xen_pt_config_reg_init: Offset 0x0030 mismatch! Emulated=0x0000, host=0xc0002, syncing to 0x0002.
(there are a bunch of these, but this one seems to be syncing differently than the others, which sync to the correct host value)
Am I running afoul of some AMD incompatibility nonsense?
PS: qvm_create_windows_qube is pretty amazing! 
PPS: I noticed that that reserved memory space is the 640KB-1MB EMS/HiMem range. The 80s called, they want their memory limits back. XD I mean, really, is that still a thing now?