On Sat, 2 Nov 2002, jean wrote: > What's the reason behind: > void swapin_readahead(swp_entry_t entry) > in mm/memory.c > > Is it just because that it does not cause seek time? Exactly. > I'm doing a measurement, and surprising found that a > lot of swap in readaheads are wasted. I'm not sure if > the measurement is wrong. Your measurement is probably right. However, you should keep in mind that a disk seek is on the order of 10 ms while disk transfer time is on the order of 40 MB/s. This means that in the time of one disk seek, a disk can read 400 kB. Disk seeks are so much more expensive than linear reads that reading extra data is worth it, even if some of that data is not used at all... regards, Rik -- Bravely reimplemented by the knights who say "NIH". http://www.surriel.com/ http://distro.conectiva.com/ Current spamtrap: <a href=mailto:"october@surriel.com";>october@surriel.com</a> -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
