Re: [patch] async writes flushes rewrite

Andrea Arcangeli (andrea@suse.de)
Sun, 19 Sep 1999 19:02:24 +0200 (CEST)


On Sun, 19 Sep 1999, Andrea Arcangeli wrote:

>I would like to get some number comparing 2.2.12 with 2.3.18ac5 and with
>2.3.18ac5+my-async-flush-rewrite. Actually I only checked the level of

If you have a chance I would like to get some number also on my rewrite of
the inode allocation as it allows the dcache to grow to an insane size 8)
and to be shrunk automagically on demand (actually the shrink is not the
best that could be done as right now we always enter the swapout path to
shrink the cache. The right thing to do is to rewrite the
shrink_slab_cache interface to get the information about how many pages
are been freed and to avoid entering the swapout path if enough pages of
the interesting type are been freed, I'll do that soon). In application
that are handling an huge number of small files and if you have plenty of
memory I expect an _huge_ speedup using the new inode dynamic allocation.

If you want you my give a try also to the whole current `my-2.3.18ac5`. It
just includes both the inode and buffer-flush changes plus other
improvements and fixes. My tree also includes the patch from
jean-luc.boyard at Siemens to make shm.c SMP threaded.

ftp://ftp.suse.com/pub/people/andrea/kernel-patches/my-2.3.18ac5/

You should simply download all the patches in the above directory in your
local disk. Then run something like:

cd linux-2.3.18ac5
for i in ../my-2.3.18ac5/*; do patch -p1 <$i; done

and you'll go in sync with the kernel I am running while writing this
email ;).

Andrea

-
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/