Further notes:

The Aquantia 10GB controller, the Intel wireless controller, an add-in PCIE 3.0 board in the PCIE x1 slot,  and the USB 3.1 controller (as compared to the 2 x USB 3.0 controllers, which do passthrough) all won't passthrough-- most of these perpetually "need reboot".  I've read this can be caused by being behind a non-ACS compatible PCIE switch.  Setting the parameter to disable ACS check to TRUE results in a non-bootable host.  All BIOS parameters that I can identify (including "enable ACS") are enabled, and I am using the beta bios with AGESA 1.0.0.6.  Stumped, although none of this is absolutely essential to me.  Another possibility (at least for the 3.1 controller) is that it seems to be in the same IOMMU group as a PCIE bridge that can't be enabled for passthrough (greyed out.) The 3.1 controller is slightly different, VM won't power up with a message that a device isn't passthrough capable.

of 4 GPU's in the system (1080 FE, 2 x 1070Ti FE, 1 x 1060 3Gb), all can be passed through.  One of the 1070 Ti's gives the dreaded error 43, even with the hypervisor.cpuid flag set to FALSE.  The others work very nicely.

I can try replacing one of the video boards with an AMD, and using software USB passthrough from the two working controllers, but it would be really nice not to have to do that (passing through USB between VM's means that VM's become dependent on eachother  for their keyboards to work).  And the error 43 just gets my goat because it is so poorly documented on the NVidia side.

Will post further updates, would appreciate any advice.  Thanks