“…Use of large pages to increase TLB-coverage [34,35,36,55,57,63,64,66] and additional MMU structures to cache multiple levels of the page-tables [19,24,26] are some of the techniques widely adopted in commercial systems. In addition, researchers have also proposed TLB-speculation [20,60], prefetching translations [47,53,62], eliminating or devirtualizing virtual memory [42], or exposing virtual memory system to applications to make the case for application-specific address translation [16].…”