All of lore.kernel.org
 help / color / mirror / Atom feed
From: Alexey Kopytov <alexeyk@mysql.com>
To: linux-kernel@vger.kernel.org
Subject: Followup to random file I/O regressions
Date: Thu, 24 Jun 2004 18:58:23 +0400	[thread overview]
Message-ID: <200406241858.23356.alexeyk@mysql.com> (raw)

Hello!

As a follow-up to my previous benchmark results posted a while back, I ran 
some more benchmarks on 2.4.27-rc1 and 2.6.7. Basically I was trying to spot 
a reason for remaining regressions. Summing up the results, 2.6.7 performs 
much better on this workload as compared to previous 2.6 kernels, but it 
seems like large file sizes and large numbers of files cause 2.6 to perform 
slower than 2.4 in the sysbench fileio tests.

Test setup was an IDE hard drive with an ext3 partition with data=ordered, 
anticipatory scheduler for 2.6.7, LinuxThreads. The filesystem was remounted 
before each test run. All results below are average execution times of 3 
sequential test runs with the same parameters.

16 threads, 128 files, variable total file size:
        2.4.27-rc1      2.6.7
1 GB    73.78           75.75
2 GB    86.8            88.09
4 GB    99.02           108.99
8 GB    115.07          124.53
16 GB   134.69          136.63

It's interesting that 2.6 shows a significant regression on 4 and 8 GB, but 
only a minor regression on 16 GB.

1 thread, 4 GB total file size, variable number of files:
        2.4.27-rc1      2.6.7
1       94              93.02
2       95.03           93.37
4       94.27           93.94
8       95.55           94.28
16      95.67           95.36
32      95.87           96.93
64      97.39           99.39
128     99.36           101.89
256     103.51          103.8

Here 2.6 shows better results for small number of files but worse for 32 and 
above files.

1 file, 4 GB total file size, variable number of threads:
        2.4.27-rc1      2.6.7
1       94              93.02
2       96.6            93.26
4       94.93           90.16
8       89.49           87.45
16      86.25           86.2
32      88.14           85.13

Here 2.6 performs better in all tests.

Formatted results and graphs are published on the sysbench homepage at 
http://sysbench.sourceforge.net/results/fileio/20040623.html

Regards,
Alexey.

-- 
Alexey Kopytov, Software Developer
MySQL AB, www.mysql.com

Are you MySQL certified?  www.mysql.com/certification

                 reply	other threads:[~2004-06-24 15:00 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=200406241858.23356.alexeyk@mysql.com \
    --to=alexeyk@mysql.com \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.