High "Service" Throughput After Viprion Migration

We migrated our production LTMs from 6900 appliances on BigIP v12.0 to vCMP guests running v12.0 on Viprion 2400s with two 2250 blades each. We added the new LTMs to the existing device group and failed over to them so the config is exactly the same. Before the migration, we ran mostly under 1Gb of throughput as indicated in this 24 hour graph:

After the migration, we are seeing the service throughput at very high numbers as seen below:

Notice graph 1 is in Mb and graph 2 is in Gb. Also, in graph 1 you see the in and out lines while in graph 2 you only see "service" because the in and out are so small in comparison. The network links do not appear to have any additional traffic than before the migration so this appears to be an issue internal to the LTM.

Looking into the network statistics, it appears to be related to the Packet Velocity Acceleration. But again, this doesn't show up in the actual network interfaces. We only have 80G of connectivity enabled per Viprion chassis.

This is the LTM traffic:

This is the PVA traffic:

Has anyone else had an issue like this after moving to vCMP guests or Viprions in general on v12.0? Any ideas as to why the PVA and service throughput numbers would be so high? Everything seems to run fine but these numbers make it hard to monitor the system. Thoughts?