Miguel Figueiredo Mascarenhas Sousa Filipe
2008-Mar-01 16:26 UTC
[Btrfs-devel] Bonnie++ on linux/ppc btrfs 0.13
Hi there,
I thought I'd share some simple bonnie++ runs I did on a "low end"
system.
I ran bonnie++ 1.03 on a ibook/ppc, with btrfs-unstable (rev e4cd88595ed7 ) and
ext3, on ubuntu/ppc.
This i a ibook G3 800mhz ppc750 , with 640mb of ram, and a 30GB slow ide disk.
Kernel is: 2.6.20-16-powerpc
this is the output of a hdparm -i /dev/hda
/dev/hda:
Model=TOSHIBA MK3018GAS, FwRev=Q3.03 B, SerialNo=X2MH7901T
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=46
BuffType=unknown, BuffSize=0kB, MaxMultSect=128, MultSect=off
CurCHS=16383/16/63, CurSects=16514064, LBA=no
IORDY=no, tPIO={min:120,w/IORDY:120}
PIO modes: pio0 pio3 pio4
UDMA modes: udma0 udma1 *udma2 udma3 udma4 udma5
AdvancedPM=yes: unknown setting WriteCache=enabled
Drive conforms to: Unspecified: ATA/ATAPI-1,2,3,4,5
$ cat btrfs.bonnie
Version 1.03 ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
newton 1256M 4174 81 15443 35 7111 8 4933 89 18301 11 75.1 1
------Sequential Create------ --------Random Create--------
-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 5583 98 +++++ +++ 4536 99 5596 99 +++++ +++ 4070 98
$ cat ext3.bonnie
Version 1.03 ------Sequential Output------ --Sequential Input- --Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
newton 1256M 5337 95 17829 14 7100 8 3498 59 17087 12 63.8 0
------Sequential Create------ --------Random Create--------
-Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 16200 83 +++++ +++ 23956 83 14767 76 +++++ +++ 18247 65
I would like to point out that metadata operations, like create and delete seem
cpu bounded, and a bit low..
Also, btrfs is allready besting ext3 on seeks/sec and input performance :D.
I also had to apply this patch for the kernel module to link and
insmod on this kernel.
diff -r e4cd88595ed7 extent_io.c
--- a/extent_io.c Thu Feb 21 14:54:12 2008 -0500
+++ b/extent_io.c Sun Mar 02 00:25:39 2008 +0000
@@ -2078,7 +2078,7 @@ done:
return 0;
}
-#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
+#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,20)
/* Taken directly from 2.6.23 for 2.6.18 back port */
typedef int (*writepage_t)(struct page *page, struct writeback_control *wbc,
Kind regards
--
Miguel Sousa Filipe
On Saturday 01 March 2008, Miguel Figueiredo Mascarenhas Sousa Filipe wrote:> Hi there, > > I thought I'd share some simple bonnie++ runs I did on a "low end" system. > I ran bonnie++ 1.03 on a ibook/ppc, with btrfs-unstable (rev e4cd88595ed7 ) > and ext3, on ubuntu/ppc.Great, thanks for trying it out.> > This i a ibook G3 800mhz ppc750 , with 640mb of ram, and a 30GB slow ide > disk. Kernel is: 2.6.20-16-powerpc > > this is the output of a hdparm -i /dev/hda > /dev/hda: >[ results ]> I would like to point out that metadata operations, like create and delete > seem cpu bounded, and a bit low.. > Also, btrfs is allready besting ext3 on seeks/sec and input performance :D.Especially on slower CPUs, the btree operations done by btrfs are going to be slower than ext3 for creating and deleting files. There is some low hanging fruit to optimize there, but I'd suggest smaller block sizes (mkfs.btrfs -l 4k -n 4k) on this type of machine. I'm surprised that ext3 did not rewrite the file at full disk speed.> > I also had to apply this patch for the kernel module to link and > insmod on this kernel. >Ok, I'll pull this in, thanks! -chris