meh, split frame rendering has been around since the dawn of multi-gpu gaming, it didn't work well because the workload in different parts of the screen can be vastly different. Nothing has changed, it is just a fundamental limitation. Sure, with DX12 developers probably have better control rather than relying on the drivers to split the screen but that doesn't resolve the situation.
And the shared memory is nice and all except it only has the advantage of preventing duplication of a resource that will only be used by one of the GPUs, which basically never happens. texture and geometry etc are going to be needed on both GPUs because it will be far too slow exchanging resources between GPUs.
This works much better for compute applications.