Messages in this thread | | | Date | Sat, 28 Mar 1998 15:16:58 +0100 (MET) | From | Gerard Roudier <> | Subject | Re: 2.1.91 swap performance: jerky. |
| |
On 28 Mar 1998, Linus Torvalds wrote:
> In article <351C2E91.5C23733D@pobox.com>, mlord <mlord@pobox.com> wrote: > > > >Tried it. Less dramatic swings in swapping behaviour > >than before, but still leaves more free memory than I'd > >expect.. 4-5MB "free" at all times, despite continuous swapping. > > This is intentional. The background swapping tries to swap out enough > stuff that we're never really low on memory (and that means that it > tries to keep at least one 128kB _contiguous_ area free, for example - > which in turn tends to indicate a few megs of real free memory).
Without some very smart algorithm that selects pages to swap out, this 128kB contiguous free area goal seems to me close to impossible without trashing LOTS of really used pages. Using a more or less blind algorithm can only lead to some catatonic behaviour, IMO.
What's the real need of contiguous memory?
Modules that need large physically contiguous memory chunks should be loaded very early after system start-up. If we need some 8k contiguous chunks for kernel stacks, why to try the 128kB goal that is, IMO, a LOT harder to achieve ?
Gerard.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |