I went ahead and filed a support request for this just to get an answer out of VMware. Here is what I was told over the phone:
The Revision E processors do not support RDTSCP; The Revision F processors do. The mask effectively "hides" the bit from the Guest OS. When the Guest OS queries to see if that instruction is available, it returns no. That way, you have a uniform environment across all your hosts.
If this mask was not in place and a VM was running on your Revision F host, and was VMotioned to the Revision E host, that instruction would no longer be there and the Guest OS could encounter problems (it would probably bluescreen if it was Windows).
They also said that this bit mask is supported and they have not seen any issues in production with this setup.
After speaking with support, i'm pretty confident we won't encounter any issues with the new Revision F boxes.
