Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 19 Jul 1999 21:48:03 -0400 (EDT)
From:      Mikhail Teterin <mi@aldan.algebra.com>
To:        Greg Lehey <grog@lemis.com>
Cc:        scsi@freebsd.org
Subject:   Re: SEAGATE ST34371W on an NCR -- slow :(
Message-ID:  <199907200148.VAA07328@misha.cisco.com>
In-Reply-To: <19990714163904.17891@mojave.lemis.com> from Greg Lehey at "Jul 14, 1999 04:39:04 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
Greg Lehey once wrote:

> > I'm getting a dissapointingly low performance from the only disk on the
> > ncr0:
> >
> > ncr0: <ncr 53c875 fast20 wide scsi> rev 0x37 int a irq 21 on pci0.13.0
> > ncr1: <ncr 53c875 fast20 wide scsi> rev 0x37 int b irq 22 on pci0.13.1
> > [...]
> > da0 at ncr0 bus 0 target 3 lun 0
> > da0: <SEAGATE ST34371W 0440> Fixed Direct Access SCSI-2 device
> > da0: 40.000MB/s transfers (20.000MHz, offset 15, 16bit), Tagged Queueing E.
> > da0: 52MB (8496884 512 byte sectors: 255H 63S/T 528C)
> >
> > This  is a  dual  PII-300MHz system  with 64Mb  of  RAM running  FreeBSD
> > 3.2-STABLE from Wed Jul 7.
> >
> > The iozone's numbers are:
> >
> >         File size set to 80000 KB
> >         Time Resolution = 0.000015 seconds.
> >         Processor cache size set to 1024 Kbytes.
> >         Processor cache line size set to 32 bytes.
> >         File stride size set to 17 * record size.
> >
> > 						    random  random
> >       KB  reclen   write rewrite    read    reread    read   write
> >    80000       4    3885    1460     4842     4474     415     208
> >
> >     bkwd  record  stride
> >     read rewrite    read   fwrite frewrite   fread  freread
> >      988  135635     543     5235     1533    4674     4803
> >
> > Big things, like  Netscape and KDE take very long  to start up, probably
> > even longer then they used to take  on my older P90 system with the same
> > amount of RAM.
> >
> > What should I be tuning? Thanks!
> 
> I'd be  interested to see  what results  rawio shows. It  bypasses the
> buffer  cache  and measures  raw  disk  performance.  Take a  look  at
> ftp://ftp.lemis.com/pub/rawio.tar.gz.

root@guest:~ (148) rawio -I "Dual-One-SETI@Home" /dev/rda0s1e
           Random read  Sequential read    Random write Sequential write
ID          K/sec  /sec    K/sec  /sec     K/sec  /sec     K/sec  /sec
Dual-One-SETI@Home Child 0 Bad read at 1511671296: (null) (671939037)
Child 1 Bad read at 382283264: (null) (671939037)
Child 2 Bad read at 1870251520: (null) (671939037)
Child 3 Bad read at 1451056128: (null) (671939037)
Child 5 Bad read at 1095859200: (null) (671939037)
Child 4 Bad read at 2002068480: (null) (671939037)
Child 6 Bad read at 559927808: (null) (671939037)
Child 7 Bad read at 1127489024: (null) (671939037)
     0.0     0    4394.9   268

I wonder what this "Bad read" s are about....

The device is typicly used as my /tmp. THe second run gives:

root@guest:~ (149) rawio  /dev/rda0s1e
           Random read  Sequential read    Random write Sequential write
ID          K/sec  /sec    K/sec  /sec     K/sec  /sec     K/sec  /sec
anon     Child 0 Bad read at 1890011136: (null) (671939037)
Child 1 Bad read at 1000460800: (null) (671939037)
Child 2 Bad read at 475668480: (null) (671939037)
Child 3 Bad read at 230313472: (null) (671939037)
Child 4 Bad read at 1622021632: (null) (671939037)
Child 5 Bad read at 460020224: (null) (671939037)
Child 6 Bad read at 1155239424: (null) (671939037)
Child 7 Bad read at 748593664: (null) (671939037)
     0.0     0    3976.9   243

root@guest:~ (150) rawio -I all -a /dev/rda0s1e
           Random read  Sequential read    Random write Sequential write
ID          K/sec  /sec    K/sec  /sec     K/sec  /sec     K/sec  /sec
all      Child 0 Bad read at 1311602688: (null) (671939037)
Child 1 Bad read at 518595584: (null) (671939037)
Child 2 Bad read at 1612716032: (null) (671939037)
Child 3 Bad read at 1718660608: (null) (671939037)
Child 4 Bad read at 934173696: (null) (671939037)
Child 5 Bad read at 365423104: (null) (671939037)
Child 6 Bad read at 1737638400: (null) (671939037)
Child 7 Bad read at 622154752: (null) (671939037)
     0.0     0    4153.1   253  Child 0 Bad write at 712240128: (null) (671939037)
Child 1 Bad write at 2029816832: (null) (671939037)
Child 2 Bad write at 1479277568: (null) (671939037)
Child 3 Bad write at 1101784064: (null) (671939037)
Child 4 Bad write at 741783040: (null) (671939037)
Child 5 Bad write at 1680343552: (null) (671939037)
Child 6 Bad write at 1918504448: (null) (671939037)
Child 7 Bad write at 700973568: (null) (671939037)
     0.0     0    1830.9   112

And finally, with the -v 2 setting:
root@guest:~ (151) rawio -v 2 -I all -a /dev/rda0s1e
Test name:                   all
Transfer count:            16384
Record count:              16384
Process count:                 8
Device size:          4350404608
Test    ID              Time        KB/sec          /sec %User    %Sys  %Total Reads    Writes
Child 1 Bad read at 754815488: (null) (671939037)
Child 2 Bad read at 1117575168: (null) (671939037)
Child 3 Bad read at 2136927744: (null) (671939037)
Child 4 Bad read at 1750441472: (null) (671939037)
Child 0 Bad read at 838010368: (null) (671939037)
Child 5 Bad read at 1213815808: (null) (671939037)
Child 7 Bad read at 792023552: (null) (671939037)
Child 6 Bad read at 1504394240: (null) (671939037)
RR      all         0.049070         845.2           41    0.0    40.5    40.5 20
SR      all        64.628578        4153.5          254    0.1     2.6     2.6 16384    0
Child 0 Bad write at 656580096: (null) (671939037)
Child 3 Bad write at 1575190528: (null) (671939037)
Child 2 Bad write at 1815015936: (null) (671939037)
Child 4 Bad write at 1145159680: (null) (671939037)
Child 5 Bad write at 1926565888: (null) (671939037)
Child 1 Bad write at 696373760: (null) (671939037)
Child 7 Bad write at 1483964928: (null) (671939037)
Child 6 Bad write at 158612480: (null) (671939037)
RW      all         0.020574        1916.2           97    0.0    83.1    83.1 02
SW      all       142.680023        1881.4          115    0.0     1.1     1.2 016384

Without the Seti@home, which was running on one of the CPUs at idprio 10 during the
previous tests:

root@guest:~ (155) rawio -v 2 -I NO-SETI -a /dev/rda0s1e
Test name:               NO-SETI
Transfer count:            16384
Record count:              16384
Process count:                 8
Device size:          4350404608
Test    ID              Time        KB/sec          /sec %User    %Sys  %Total Reads    Writes
Child 0 Bad read at 1404747264: (null) (671939037)
Child 1 Bad read at 1404747264: (null) (671939037)
Child 2 Bad read at 1538743808: (null) (671939037)
Child 3 Bad read at 1702750208: (null) (671939037)
Child 5 Bad read at 1468262912: (null) (671939037)
Child 6 Bad read at 670607360: (null) (671939037)
Child 7 Bad read at 1359508992: (null) (671939037)
Child 4 Bad read at 957671424: (null) (671939037)
RR  NO-SETI         0.042906         716.0           47    0.0    61.7    61.7 20
SR  NO-SETI        74.679314        3594.5          219    0.0     2.3     2.3 16384    0
Child 1 Bad write at 1859974144: (null) (671939037)
Child 0 Bad write at 2059030528: (null) (671939037)
Child 3 Bad write at 1325156352: (null) (671939037)
Child 2 Bad write at 1457619456: (null) (671939037)
Child 4 Bad write at 839265280: (null) (671939037)
Child 5 Bad write at 1203342336: (null) (671939037)
Child 6 Bad write at 2026992128: (null) (671939037)
Child 7 Bad write at 162286592: (null) (671939037)
RW  NO-SETI         0.016841        1550.5           59    0.0   124.1   124.1 01
SW  NO-SETI       143.285309        1873.4          114    0.0     1.2     1.2 016384

Thanks for ideas... Yours,

	-mi


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-scsi" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199907200148.VAA07328>