On 05/11/2011 08:33 PM, Joe Zeff wrote: > On 05/11/2011 12:19 PM, Bryn M. Reeves wrote: >> And have it all chucked onto the swap device when we're under memory pressure >> and reclaiming user pages that haven't been touched in a while? How does that help? > > It's much less likely, especially if you're using that file often enough No it isn't - it's dependent on configuration and workload. There are tunables for this sort of thing. See the swappinness sysctl: http://kerneltrap.org/node/3000 I think either Linux Kernel Development 1st/2nd ed[1]. or Understanding the Linux Kernel[2] give a straightforward explanation of the algorithm used (possibly somewhat outdated now but conceptually still mostly accurate - I have not read either since I stopped teaching kernel developer courses). > to worry about saving time on accessing the index. And, although I'm no > expert on the subject, I'd guess that data in a disk cache would be > swapped out long before data that the program's written into its own > data space. And, of course, there's the advantage that you're not > depending on system calls that may or may not be honored the way you want. Guessing doesn't help when you want the data in memory "at all cost" - if you want determinism it's better to use the APIs that provide it. Regards, Bryn. [1] http://www.amazon.com/Linux-Kernel-Development-Robert-Love/dp/0672325128 [2] http://oreilly.com/catalog/9780596005658 LKD is now up to a 3rd ed. btw but I've not read it at all: http://blog.rlove.org/2010/07/linux-kernel-development-third-edition.html -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines