The Pascal architecture supports the page migration engine required for unified_shared_memory, as indicated by NVIDIA:
The limitation was introduced in D54493 which justified the cut-off by the requirement for unified addressing. However, Unified Virtual Addressing (UVA) is already available with [[ https://docs.nvidia.com/cuda/gpudirect-rdma/index.html#basics-of-uva-cuda-memory-management | sm20 (Fermi. Kepler, Maxwell) ]]. Unified shared memory might even be possible with these, but with migration of entire allocations on kernel startup.
To be sure, I enabled the tests for a Pascal GPU which [[ http://meinersbur.de:8011/#/builders/143/builds/345 | finish successfully ]].