severe packet loss with vtnet0 guest, re0 host

From: void <void_at_f-m.fm>
Date: Sat, 06 Aug 2022 21:24:29 UTC
Hello questions@,

On a 13.1-release server there is a 13.1-release bhyve guest vm.
The host (server) is releng/13.1-n250148-fc952ac2212

There's no packet loss direct from the host to the vm. There's high
packet loss to the vm from the DMZ and when pinging from offsite.
There is no packet loss between the host and offsite.

The host's filesystem is (nvme-backed) UFS. The guest is UFS.
The host interface is re0. The guest is of course vtnet0.

The odd thing is that pinging the host from anywhere - zero packet loss.
It's just the guest. Load as reported by top in the guest is nearly 
nothing - 0.02. No swap is in use (though swap is configured) on
either the host or the guest

Are there any issues wrt re0 on host and vtnet0 on guest?
I have no idea where to start with this :(

The same vm but hosted on another freebsd host (also 13-stable)
shows no issues. The differences here are that the problem
occurs on the nvme-backed host. Another difference that might 
be relevant is that problematic host is an AMD Ryzen 5 PRO 3400G but the 
non-problematic hosts are Xeons. Maybe it's an interrupt thing?
The vm is started on (any) host with vmrun.sh

Anyone have any ideas? Thanks