disk transfer speed problem

initial data: one brand new intel bonham motherboard (i815) , 900 MHz celeron processor
2 new 80 GB Seagate Barracuda used as raid1, 7200 RPM,ext3 filesystem, RedHat 2.4.9-21 kernel from RH 7.2 updates.

settings (identical for hdc and hdd), as detected by the kernel (except for I/O support = 3):
#hdparm /dev/hdc

 multcount    = 16 (on)
 I/O support  =  3 (32-bit w/sync)
 unmaskirq    =  0 (off)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 nowerr       =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)

geometry = 23989/16/63, sectors = 156301488, start = 0

#hdparm -i /dev/hdc

Model=ST380021A, FwRev=3.05, SerialNo=3HV080KH
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% } RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4 BuffType=unknown, BuffSize=2048kB, MaxMultSect=16, MultSect=16 CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=156301488 IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120} PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5

#dmesg|grep UDMA
hda: 14668290 sectors (7510 MB) w/418KiB Cache, CHS=913/255/63, UDMA(66) (this is the boot disk)

hdc: 156301488 sectors (80026 MB) w/2048KiB Cache, CHS=155061/16/63, UDMA(100
hdd: 156301488 sectors (80026 MB) w/2048KiB Cache, CHS=155061/16/63, UDMA(100)

The IDE cables have 80 wires. I am not sure that they are ATA100, but I think they do.

The problem:
#/sbin/hdparm -tTf /dev/hdd

/dev/hdd:
Timing buffer-cache reads: 128 MB in 1.17 seconds =109.40 MB/sec
Timing buffered disk reads: 64 MB in 3.24 seconds = 19.75 MB/sec

Even after issuing raidstop and umounting the partition placed on the RAID, the speed does not exceed 22 MB/sec.

With hdparm -c1 performances are almost identical

Tests were performed in initlevel 3, with NFS started, but no one but root using the machine (from console). In production, the speed is also very, very slow

Any idea what might limit the transfer speed ?

--
Manuel Wolfshant linux registered user #131416
network administrator NoBug Consulting Romania
Beware the fury of a patient man.

Please post dmesg, /proc/pci and chipset info from /proc/ide. Also, what is result from 'hdparm -tT /dev/hda'?

rgds,
tim.
--

What is more useful is the cat /proc/ide/ide0/config !!!

Oh and for those not reading this email, it is a side note on why the ide proc-pci interface had best be left alone and in tact.

Cheers,

Andre Hedrick
Linux Disk Certification Project Linux ATA Development

Module the fact fact lspci could be easly extendid to provide the same information from user space... wait... we are on unix and we have manual pages:

-x Show hexadecimal dump of first 64 bytes of the PCI
configuration space (the standard header). Useful for debugging of drivers and lspci itself.

-xxx Show hexadecimal dump of whole PCI configuration
space. Available only for root as several PCI devices crash when you try to read undefined por- tions of the config space (this behaviour probably doesn't violate the PCI standard, but it's at least very stupid).

As you can see lspci -x -xxx ALREADY DOES THIS!

And lspci -xxx is insufficent because?

| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

You and I know it is the functional equivalent; however, in development of driver it is more practical and faster to issue

echo PXX:YY > config

setpci XX.YY.Z blah blah ...

So if there is a real issue with its presense, apply a uuid test for access. Everything else has it so this should too.

Cheers,

Andre Hedrick
Linux Disk Certification Project Linux ATA Development

Picking nits, perhaps -- but after giving us data on hdc, you did the test on hdd. I realize you said that it is identical for both, but if you missed something critical, we'll never see it.

Stephen Samuel +1(604)876-0426                [email protected]
		   http://www.bcgreen.com/~samuel/
Powerful committed communication, reaching through fear, uncertainty and
doubt to touch the jewel within each person and bring it to life.

they are 100% identical. except for the fact that one is master and the other slave on the same cable.
I did not include the results for both in order to shorten the message. here we go, just in case:

[root@bugmaster /root]# /sbin/hdparm -tT /dev/hdc /dev/hdd;/sbin/hdparm /dev/hdc /dev/hdd;/sbin/hdparm -i /dev/hdc /dev/hdd

/dev/hdc:
Timing buffer-cache reads: 128 MB in 1.17 seconds =109.40 MB/sec
Timing buffered disk reads: 64 MB in 3.01 seconds = 21.26 MB/sec

/dev/hdd:
Timing buffer-cache reads: 128 MB in 1.09 seconds =117.43 MB/sec
Timing buffered disk reads: 64 MB in 2.96 seconds = 21.62 MB/sec

/dev/hdc:

 multcount    = 16 (on)
 I/O support  =  1 (32-bit)
 unmaskirq    =  0 (off)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 nowerr       =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)

geometry = 23989/16/63, sectors = 156301488, start = 0

/dev/hdd:

 multcount    = 16 (on)
 I/O support  =  1 (32-bit)
 unmaskirq    =  0 (off)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 nowerr       =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)

geometry = 23989/16/63, sectors = 156301488, start = 0

/dev/hdc:

Model=ST380021A, FwRev=3.05, SerialNo=3HV080KH
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% } RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4 BuffType=unknown, BuffSize=2048kB, MaxMultSect=16, MultSect=16 CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=156301488 IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120} PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5

/dev/hdd:

Model=ST380021A, FwRev=3.05, SerialNo=3HV07SDN
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% } RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4 BuffType=unknown, BuffSize=2048kB, MaxMultSect=16, MultSect=16 CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=156301488 IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120} PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5

Mr Davidovac Zoran mentioned normal transfer speeds with kernel 2.4.17. Should I try ? Are there significant differences between 2.4.9-21 (or -31) and 2.4.17 in the IDE drivers ?

-
This is Linux Country. On a quiet night, you can hear Windows NT reboot!