Re: [PATCH] Simple Page Coloring [fixed] (2.3.99pre3 diffs)

From: Joseph A. Martin (jmartin@linux08.mro.dec.com)
Date: Thu Apr 13 2000 - 10:45:51 EST


:: Andrea Arcangeli <andrea@suse.de>
>On Wed, 12 Apr 2000, Joseph A. Martin wrote:
>
>>+ while ((page = alloc_pages(GFP_USER, order)) != 0) {
>
>It has bad complexity problems. An allocation have to keep working in O(1)
>as now.

I don't think I understand. Is "It" the whole thing or the quoted
line? What does "0(1)" mean?

>BTW, that's not either static or dynamic page coloring but it's "random"
>page coloring ;) and it can't work if you spawn multiple tasks at the same
>time.

That depends on the size of the cache and the size of the tasks. In
fact, if the total memory used by the simultaneous tasks is less than
the cache size, then not only will these tasks not compete with
themselves, but also they won't compete with each other for cache. If
the aggregate size of the tasks exceeds the cache size by a lot, then
the competition among tasks will churn the cache with each time slice,
whether or not any one task has nice behavior on its own. That nice
behavior in a single task may be desirable enough to strive for
anyway, and it's true that I'm not attempting to produce it under
every imaginable set of conditions.

Is there a particular applications or suite of applications you had in
mind?

Regards,
\Joe [Joseph.Martin@Compaq.com Joseph A. Martin
                                         Compaq Computer Corporation
                                        200 Forest Street MRO1-2/K20
 (508) 467-2100 Marlboro, MA 01752 U.S.A.
 ]

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sat Apr 15 2000 - 21:00:21 EST