Messages in this thread |  | | Date | Tue, 11 Feb 1997 14:58:45 -0500 (EST) | From | Floody <> | Subject | Re: Linux VM subsystem (Was: Big mallocs, mmap sorrows and double buffering.) |
| |
-----BEGIN PGP SIGNED MESSAGE-----
On Tue, 11 Feb 1997, A.N.Kuznetsov wrote:
> In article <3300980A.1F62@raleigh.ibm.com> you wrote: > : A.N.Kuznetsov wrote: > : > Linux vm manager is the dumbest one in the world 8)8) > : > I tested my program on SunOS-4, SunOS-5, FreeBSD and Digital Unix, > : > Linux's VM looks really pathetic. > > : What exactly is missing/misdesigned? > > As I understand, Linux always makes random page swapouts, > that is ideal (though I am not sure) for desktop > applications. But for applications running in background > and using large volumes of virtual memory this algorithm breaks. > Linux practically does not coalesce swapins/swapouts.
Not exactly sure what you mean by "random page swapouts." The algorithm that Linux VM uses is *FAR* from "random", in fact it seems quite intelligent. Pages are aged, weighted and even scaled per process when swapping (scaled == number of pages swapped from a process is affected by the number of page faults the process has recently had, so that a heavily running process will have less swapped out than a "light" process). It would be interesting to see your test code, as well as hear feedback from you on VM techniques and where Linux may be flawed or need improvement.
+-------------------------------------------------------------------+ + -- Finger: flood@evcom.net for my PGP public key -- + +-------------------------------------------------------------------+
-----BEGIN PGP SIGNATURE----- Version: 2.6.2
iQCVAwUBMwDPeBsjWkWelde9AQG77QQAwbpNyln97lJi/ELr/+OGRGOj5PwAdqvY 2AOOJA66hvYBjbidvTHLLft1vkNCyRRbBj/zSb2vU3z46CXj0ZEgIpe8WU+Beo8T DOzDEbdLv+WymGqJHeklJs7b0QHehzHddJ3FbD9il/xHeaJlLl88gjAehypTcAUd m8RwLzs4VxI= =lh+i -----END PGP SIGNATURE-----
|  |