Firstly, it depends on whether there will only be 128-bit data registers or address registers.
The first option only means problems for programs that use shifts and overflows.
The second option means that problems with abuse of address arithmetic will also suffer.
BUT: personally, I think that if this message is not a complete BS, it most likely means that MS is trying to make its kernel more agrostatic to support existing 32-bit and 64-bit architectures (ARM comes to mind) not some non-existent 128-bit architecture.
They did this before targeting the i860 during the development of Windows NT. The resulting hardware abstraction layer allows them to support x86, MIPS, Alpha, etc.
elder_george
source share