[prev in list] [next in list] [prev in thread] [next in thread]
List: reiserfs-devel
Subject: (reiserfs) The number of bytes memcpy'd on average for reiserfs balancing
From: Hans Reiser <reiser () idiom ! com>
Date: 1999-02-17 18:22:06
[Download RAW message or body]
[Attachment #2 (multipart/alternative)]
It occurred to me that 2k might be an unrepresentative number. Measure it with 3k also.
Thanks Yura,
Hans
--
Don't be locked out of the source, and doomed to life in the slow lane.
Dump NT! Get Linux (http://www.kernel.org) plus ReiserFS
(http://devlinux.org/namesys). If you sell an OS or
internet appliance, buy a port of ReiserFS!
Speed matters. Trees are fast. Go faster!
[Attachment #5 (text/html)]
<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
It occurred to me that 2k might be an unrepresentative number. Measure
it with 3k also.
<p>Thanks Yura,
<p>Hans
<pre>--
Don't be locked out of the source, and doomed to life in the slow lane.
Dump NT! Get Linux (<A HREF="http://www.kernel.org">http://www.kernel.org</A>) plus ReiserFS
(<A HREF="http://devlinux.org/namesys">http://devlinux.org/namesys</A>). If you sell an OS or
internet appliance, buy a port of ReiserFS!
Speed matters. Trees are fast. Go faster!</pre>
</html>
Return-Path: yura@namesys.botik.ru
Received: from select.botik.ru (select.botik.ru [193.232.174.49])
by idiom.com (8.8.8/8.8.5) with ESMTP id GAA06418
for <reiser@idiom.com>; Wed, 17 Feb 1999 06:49:57 -0800 (PST)
Received: from namesys.botik.ru (really [127.0.0.1]) by select.botik.ru
via smail with esmtp (ident root using rfc1413)
id <m10D94E-0006VrC@select.botik.ru> (Debian Smail3.2.0.101)
for <reiser@idiom.com>; Wed, 17 Feb 1999 18:39:14 +0300 (MSK)
Sender: root@namesys.botik.ru
Message-ID: <36CAE29A.DC7C2D6E@namesys.botik.ru>
Date: Wed, 17 Feb 1999 18:39:07 +0300
From: "Yury Y.Rupasov" <yura@namesys.botik.ru>
X-Mailer: Mozilla 4.5 [en] (X11; I; Linux 2.2.1 i586)
X-Accept-Language: en
MIME-Version: 1.0
To: Hans Reiser <reiser@idiom.com>, vs@namesys.botik.ru
Subject: Re: Some non-serious profiling using gdbstub, slightly interesting
References: <36C29532.B35565F1@netcom.com> <36C29FAB.D7F95D2C@idiom.com> \
<36C8636B.EB0A7FD8@namesys.botik.ru> <36C873D8.B191EBFD@idiom.com> \
<36C9B4EA.D26EC16B@namesys.botik.ru> <36CA4099.9BD9322F@idiom.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mozilla-Status2: 00000000
Hello,Hans
> So, my question is, when we cp a small 2k file, on average, how many k of bytes do \
> we memcpy? If this is a hard question, taking more than an hour or two to measure, \
> don't spend time on it....
The answer is 4556.4 bytes on average .
I added to the source code of reiserfs global variable to calc bytes memcpy do.
10 files of 2kb was created, then I did the test :
copy 10 file of 2kb from ext2fs to reiserfs.
With help of gdbstub I cheked value of my glogal variable.
So: 45564 bytes - total for 10 files.
4556.4 bytes - on average.
I did the test 3 times - results was around this value.
best,
Yura.
Hans Reiser wrote:
> "Yury Y.Rupasov" wrote:
>
> > Hello, Hans
> >
> > I removed Volodya's bugs from reiserfs code and ran gdbstub.
> > I tested how reiserfs works with small files.
> >
> > I did some commands:
> >
> > mkreiserfs /dev/hdc3
> > mount -t reiserfs /dev/hdc3 /mnt
> > cd /mnt
> > mk dir1
> > cd /mnt/dir1
> >
> > Then :
> >
> > 1. f1 - file 55 bytes long.
> >
> > cp f1 f2
> > search_by_key 11 times stopped
> > fix_nodes 6 times stopped
> > do_balance 3 times stopped
> >
> > 2. cat f1 f2 > f3
> > search_by_key 15 times stopped
> > fix_nodes 8 times stopped
> > do_balance 4 times stopped
> > 3. cp f3 f4
> > search_by_key 11 times stopped
> > fix_nodes 6 times stopped
> > do_balance 3 times stopped
> > 4. rm f3
> > search_by_key 7 times stopped
> > fix_nodes 6 times stopped
> > do_balance 3 times stopped
> >
> > 5. cp 2_byte_file f5
> > search_by_key 11 times stopped
> > fix_nodes 6 times stopped
> > do_balance 3 times stopped
>
> So, my question is, when we cp a small 2k file, on average, how many k of bytes do \
> we memcpy?
> If this is a hard question, taking more than an hour or two to measure, don't spend \
> time on it....
> Best,
>
> Hans
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic