[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Sparc has slow disk access



Several things.  One, you left out a ton of information, as this
could be the result of many, many things.  Where to start?

What kind of disk/controller for each machine/disk?

What is the results you get from running bonnie?

Can you give the output of hdparm for each disk?

Also notice the the x86 box you are talking about is a considerably
faster processor/memory combination, and no, I'm not talking about
bogomips.

Try the solaris test again, Binkey, and this time use a filesystem
other than /tmp for the output.  Try setting up a ufs-log file
system and see if the performance doesn't exceed that of linux.

That's your homework for today.  Tomorrow we'll get to advanced disk
performance considerations. ~:^)

a


"Tille, Andreas" wrote:
> 
> Hello,
> 
> I´ve got a Sparc E250 server and I´m runing Debian GNU/Linux on it.
> I now started to make some performance comparisons against a default PC.
> Here are the configurations I used:
> 
> E250:
>   ~> cat /proc/cpuinfo
>   cpu             : TI UltraSparc II  (BlackBird)
>   fpu             : UltraSparc II integrated FPU
>   promlib         : Version 3 Revision 16
>   prom            : 3.16.1
>   type            : sun4u
>   ncpus probed    : 2
>   ncpus active    : 2
>   Cpu0Bogo        : 799.53
>   Cpu1Bogo        : 799.53
>   MMU Type        : Spitfire
>   State:
>   CPU0:           online
>   CPU1:           online
> 
>   2 GB RAM
>   4x36 GB SCSI
> 
>   ~> uname -a
>   Linux bse 2.2.18pre21 #1 SMP Tue Dec 19 22:53:19 CET 2000 sparc64 unknown
> 
> PC:
>   ~> cat /proc/cpuinfo
>   processor       : 0
>   vendor_id       : GenuineIntel
>   cpu family      : 6
>   model           : 8
>   model name      : Pentium III (Coppermine)
>   stepping        : 6
>   cpu MHz         : 795.576
>   cache size      : 256 KB
>   fpu             : yes
>   fpu_exception   : yes
>   cpuid level     : 2
>   wp              : yes
>   flags           : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse
>   bogomips        : 1589.24
> 
>   256 MB RAM
>   20 GB IDE
> 
> With this setup I noticed that the PC is about two times faster in
> inserting data into a postgresql database.  I´ve got a hint to test
> disk speed and I did so with the following script:
> 
> #!/bin/bash
> date
> ls -l /home/tillea/hd-test.file
> i=0
> while [ $i -lt 100 ] ; do
>    echo -n "$i "
>    let i="$i+1"
>    dd if=/home/tillea/hd-test.file of=/tmp/hd-test.file 2> /dev/null
> done
> ls -l /tmp/hd-test.file
> date
> ### end of scrip
> 
> Here are the results:
> 
> PC:
> 
> $ uname -a
> Linux wr-linux02 2.4.5 #1 Mit Mai 30 08:24:06 CEST 2001 i686 unknown
> 
> $ hd-test
> Don Aug 23 12:21:56 CEST 2001
> -rw-r--r--    1 tillea   admin    25600000 23. Aug 11:50 /home/tillea/hd-test.file
> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 -rw-r--r--    1 tillea   admin    25600000 23. Aug 12:22 /tmp/hd-test.file
> Don Aug 23 12:22:47 CEST 2001
> 
> ----> 51s  for the PC
> 
> Sparc:
> 
> $ uname -a
> SunOS bse 5.8 Generic_108528-08 sun4u sparc SUNW,Ultra-250
> 
> $ hd-test
> Thu Aug 23 13:23:38 MEST 2001
> -rw-r--r--   1 tillea   admin    25600000 Aug 23 13:13 /home/tillea/hd-test.file
> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 -rw-r--r--   1 tillea   admin    25600000 Aug 23 13:26 /tmp/hd-test.file
> Thu Aug 23 13:26:49 MEST 2001
> 
> -----> 191s  for the Sparc running Solaris 8
> 
> $  uname -a
> Linux bse 2.2.18pre21 #1 SMP Tue Dec 19 22:53:19 CET 2000 sparc64 unknown
> 
> Don Aug 23 13:55:37 CEST 2001
> -rw-r--r--    1 tillea   admin    25600000 23. Aug 11:50 /home/tillea/hd-test.file
> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 -rw-r--r--    1 tillea   admin    25600000 23. Aug 13:57 /tmp/hd-test.file
> Don Aug 23 13:57:16 CEST 2001
> 
> -----> 99s    for tge Sparc running Debian GNU/Linux
> 
> So the relation is about:
> 
>    PC/linux    :      Sparc/Solaris     :      Sparc/Linux
> 
>       1        :          3.7           :         1.9
> 
> That´s not good for the Sparc machine (and even worse for Solaris by the way
> but this is of less interest because the machine should run under Linux).
> 
> Any hint if there some optimisations could be done?



Reply to: