Re: Pages swapped out even when free memory available...

Daniel Egger (degger@fhm.edu)
20 Aug 2002 16:11:01 +0200


--=-cCIanc71ivzeDFOT9CdC
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

Am Die, 2002-08-20 um 12.28 schrieb Jan Hudec:

> Don't do it. In user-land, you simply don't have the info kernel does.
> Gimp tries to do things like this and results are truly horrible.
> It forces machine in deep swap if you set the cache large, while with
> small cache the actual size does not matter much as kernel takes care of
> the real caching.
=20
> You should rather open a file you want to swap in, mmap it and only keep
> minimum data in other memory. Only reason to do even this may be, that
> the data may get too large to fit in swap. For all other reasons you
> should just let kernel do it's work.

Interesting you mention it. A week ago I started a thread exactly for
that reason; I implemented GIMPs tilemanagement using mmap but
unfortunately the performance sucks even more than GIMPs "clever"=20
swapping. That having said it has at least one more severe disadvantage:
The maximum amount of mmapable memory is far less than the maximum
filesize with large file support.

--=20
Servus,
Daniel

--=-cCIanc71ivzeDFOT9CdC
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: Dies ist ein digital signierter Nachrichtenteil

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQA9Yk31chlzsq9KoIYRAvK1AJ0Qlef8IG9XD+X0/EceKK8v8/IqMwCfSyFX
mo4o7FvZN7sWfiv0swPC3B4=
=r2Ej
-----END PGP SIGNATURE-----

--=-cCIanc71ivzeDFOT9CdC--

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