Global Replacement Policy removes a page from physical memory without regard to who's page it is. When all processes are working i.e. they are active and using there all pages heavily, and say now a new Process enters a new phase of its program and demanded new set of pages leading to a page fault, now for bringing new page we want free frames The Global page replacement algorithm removes pages and we bring the required page. Now the process of which the page was removed Faults and takes away another frame leading to migration of processes to Device queue(majorly Hard disk bufffer). Due to which now the ready queue will start to become empty leading to decrease in degree of multiprogramming. Now the long term scheduler then gives new set of processes to Ready queue which again Demands for new pages leading to faults again and again ----------> Thrashing
So S1 is true always as when ready queue start to become empty the degree of multiprogramming will be increase,
Coming to S2 : Since every device or resources used are Finite ; the Size of the buffer or the number of request a device can handle is finite, when Thrashing happens the degree of multiprogramming will increase upto a point, because after that the resources(Buffer size,Requests) overwhelms after that we cant move processes to device queue, so we cant increase degree of multiprogramming after that as the processes cant be given to the Device Queue. In this case now the performance will plunge and Graph decreases. So now we need to start blocking out the Processes to free up the Resoucres.
So according to me S2 is FALSE.