[ltp] Re: ThinkPad 600... Xubuntu... Maxxed Out Memory

[Robert Tomsick]

--=-YZidz+IcTHy03ehVJUNg
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Mon, 2010-03-22 at 21:40 -0400, Stefan Monnier wrote:
> > Let me get this straight: The machine doesn't have enough RAM, so you t=
ake
> > some of that precious RAM and make it unavailable to programs, so the O=
S can
> > swap RAM into and out of this reserved RAM, compressing and uncompressi=
ng it
> > when it's needed?
>=20
> That's right.
>=20
> > And it's supposed to be faster?
>=20
> Yup.  Obviously, only in some cases.  If you use little RAM, it
> shouldn't make any difference.  If you use just a bit less than all your
> RAM, the compcache will slow you down because you'll be swapping
> intou/outof the compcache.  But if you need a bit more than all your RAM
> (but not too much more, so it does fit within the compcache), then
> you'll go faster because you swap to compressed RAM rather than to disk.
>=20
> When using a lot of memory a well-designed compcache can also help if it
> ends up compressing the swap space (i.e. you first compress into the
> compcache and then swap from there to disk) and hence increase the
> apparent bandwidth of your swap disk.  IIUC Linux's current compcache
> doesn't really do this quite right yet.

Somewhat OT, but it appears that everything old is new again...  RAM
Doubler (an ancient extension by Connectix for the classic Mac OS) did
this and a bit more: it compressed pages in RAM, paged compressed pages
to disk, and even did some more trickery to make up for the OS's (rather
poor) swap implementation.  It worked quite well and, as long as you
didn't expect it to work miracles, went a long way towards making
RAM-starved machines more useful.

In my experience compcache is the same way: provided you don't expect it
to be as good as simply getting more RAM, it's actually quite handy.

-Rob

--=-YZidz+IcTHy03ehVJUNg
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEABECAAYFAkuoHwgACgkQd6gWl6secBKkAACgzy+96wQb0wEGBjnB1USA+hfb
MXkAnRb1QYMfyk/kIIQtybMwzgcacYFa
=utKW
-----END PGP SIGNATURE-----

--=-YZidz+IcTHy03ehVJUNg--