HP 3000 Manuals HP 3000 Manuals
Demand-Paged Virtual Memory (new) MPE XL employs the mapped file technique for performing file access. It is an improved version of the disk-caching capability of MPE V/E. File access efficiency is improved when portions of code and data files required for processing reside in memory. Accessing memory is faster than performing physical disk I/O operations. The mapped file technique can eliminate file system buffering and optimize global system memory management. File Mapping File mapping uses MPE XL demand-paged virtual memory to make a large amount of virtual memory directly available to you. When a file is opened, it is logically mapped into virtual memory. Each byte of each opened file has a unique virtual address. The system actually stores files page by page at physical addresses. The operating system keeps a map, a table of correspondences between virtual addresses and physical addresses, and translates between them as needed. System hardware resolves addresses through the Translation Lookaside Buffer (TLB). This translation process is transparent to the user. You reference an open file and its contents by virtual address. The MPE XL memory manager fetches several adjacent pages directly from disk as required. It places them in the user's area in memory. If you program in a language with pointers, you can access mapped files directly. You can write programs that address files through virtual memory, instead of calling file system intrinsics for disk reading and writing. Using a pointer data type, you can open and close user-mapped files with with faster LOAD and STORE on file references. You can get the advantage of file system naming and data protection for accessing array-type structures and developing specialized access methods. File mapping improves I/O performance without imposing additional CPU overhead or sacrificing data integrity and protection. HP-PA uses the Translation Lookaside Buffer to resolve addresses in the hardware. Traditional disk caching schemes for increasing I/O performance impose a CPU overhead penalty. The 900 Series hardware and system architecture allow MPE XL to perform file mapping without this penalty.