What is the reason for changing the default 1 core per socket configuration? This can cause performance issues, and generally should only be done if licensing absolutely requires it. Microsoft no longer needs this configuration for Windows, SQL, etc, so why mess with it and potentially cause issues with vNUMA?