virtual memory 4482
possibly you are thinking of simulation of LRU for things like database caches ... rather than processor paging that has hardware support for reference and change bits ... and paging least recently used replacement...
On Mon, 15 May 2006 19:25:53 +0000 (UTC) in alt.folklore.computers,
The accessed bit should be in the hardware just like the mod bit. FIFO avoids tracking anything to do with *use*, it just tracks pages in the current working set, and you already know which those are; it tells you nothing about which pages are currently actively used. It's obvious that it won't work as well as any other (i.e. better) real LR*U* approximation with reclaim.
ref: besides the numerous different variations on LRU approximations, "true" LRU implemented in simulators ... there was also belady's "OPT" ... basically with a full trace of all storage accesses ... OPT chose the page...
Compare with other (disk based) hierarchical storage management systems which migrate unused data to offline storage. They would be useless if they didn't track access, which requires an additional disk write, but seems to be considered acceptable overhead on most file systems, even where migration is not done.
On one system where we used migration to alleviate a disk space crunch, we found files unaccessed for three weeks could be migrated with nearly zero impact; we got about one retrieval request a year for migrated data. Using FIFO would be equivalent to using the created or modified date as our migration criteria: there are a lot of files which are only read after creation and migrating those would have lead to mbuttes of retrieval requests, requiring other files to be migrated first, and the situation would have looked very much like thrashing; so I'd expect systems using FIFO paging to have difficulty being able to distinguish thrashing from SOP.
-- Thanks. Take care, Brian Inglis Calgary, Alberta, Canada
fake address use address above to reply
Alt Folklore Computers from Newsgroups