Link to home
Start Free TrialLog in
Avatar of Aikema
Aikema

asked on

Hard drive access speed

I just installed SuSE 8.1 and I'm noticing that the access times seem a little slow.

I ran hdparm -t /dev/hda and got
/dev/hda:
 Timing buffered disk reads:  64 MB in  9.68 seconds =  6.61 MB/sec

6.61 MB/sec seems a little bit on the slow side to me (drive is a Seagate 7200 rpm 80 gig drive I picked up a week or two ago)

I noticed, using hdparm that DMA was off so I did the following:
#hdparm -d1 /dev/hda

/dev/hda:
 setting using_dma to 1 (on)
 HDIO_SET_DMA failed: Operation not permitted
 using_dma    =  0 (off)

I grabbed the following from dmesg:
Uniform Multi-Platform E-IDE driver Revision: 6.31
ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
VP_IDE: IDE controller on PCI bus 00 dev 89
VP_IDE: chipset revision 6
VP_IDE: not 100% native mode: will probe irqs later
VP_IDE: Unknown VIA SouthBridge, contact Vojtech Pavlik <vojtech@ucw.cz>
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide0: probed IRQ 14 failed, using default.
ide1 at 0x170-0x177,0x376 on irq 15
ide1: probed IRQ 15 failed, using default.
ide-floppy driver 0.99.newide
ide-floppy driver 0.99.newide
scsi0 : SCSI host adapter emulation for IDE ATAPI devices
reiserfs: checking transaction log (ide0(3,6)) for (ide0(3,6))
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)
ide-scsi: hdd: unsupported command in request queue (0)

(Motherboard is also only a week or two old)
Avatar of Aikema
Aikema

ASKER

Strange... I've got a cd-rom and a cd-writer hooked up to the system.

I attempt to mount the cd-rom and I get an error message, and I attempt to mount the cd-recorder and the cd-rom is mounted instead.

The installer program appears to have detected both, so why aren't they both working now?
The cd writer uses ide-scsi emulation. Try "mount /dev/scd0 /mnt/cdrom". That should mount your writer. "mount /dev/hdc /mnt/cdrom" should mount your cd-rom drive.


As for your IDE chipset, you might want to try to compile support for your chipset directly into your kernel if it is there. I suspect that linux is using generic support at the moment. It is peculiar that you can't enable DMA on your hard drive.

Larry
ASKER CERTIFIED SOLUTION
Avatar of ahoffmann
ahoffmann
Flag of Germany image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of Aikema

ASKER

Are there any workarounds?
Avatar of Aikema

ASKER

# mount -t iso9660 /dev/hdc /media/cdrom/
mount: wrong fs type, bad option, bad superblock on /dev/hdc,
       or too many mounted file systems
       (could this be the IDE device where you in fact use
       ide-scsi so that sr0 or sda or so is needed?)

# mount /dev/scd0 /media/cdrecorder/
... this mounts the cd-rom, not the cd-writer

Avatar of Aikema

ASKER

Well, I just made a copy of about 500 megs of data.  Took about 3 minutes... and the load average shot up to about 4.

I've increased the points.

ahoffman: I looked around in the SuSE support database but was unable to find anything similar
1. I think you should use /dev/scd0 and /dev/scd1 for CD drives since you installed ide-scsi

2. I've unhappily played with the kernel and IDE devices lately, with lots similar problems on some bigger machines. Also I've monitored closely the kernel lists and what I understood is that we'll probably have to wait till kernel 2.4.20 will be available. 2.4.20-pre2 looks quite ok for me, but I'm not feeling comfortable using pre kernels on production machines.
Avatar of Aikema

ASKER

hdd=ide-scsi is passed to the kernel on boot... but perhaps, if the other ide stuff is a little crazy, it might have picked up hdc as well.
Aikema
I think simply enabling ide-scsi will automatically "convert" all your IDE-CDROM devices to SCSI devices, making it necessary to access them by /dev/scd0,1,2 etc.
The reason why it's mounting the wrong cd, is because the symlink "/dev/cdrom" is linked to the incorrect scsi device, you can change this in /etc/fstab. Just remove the symlink in the /dev/folder, and re-create it , pointing it at the correct scsi device.

As for your hard drive problems:
I'm convinced that you have a problem with an unsupported chipset here. You might want to go to VIA's website and see if they have any modules for the chipset for Linux, alternatively, you can do as previously suggested and play around with your kernel settings and see if you can find a more compatible chipset to compile into the kernel. On my RH8.0 system there seems to be quite a couple of tweaks available for VIA and DMA.

.
Avatar of Aikema

ASKER

I had Mandrake 9.0 installed on the system briefly before just to try it out.  IDE worked just fine on it (it had some other problems though, most notably with sound)

I've played with a number of linux distros before and, IIRC, I always had the cd-writer as the only thing configured as scsi.
Avatar of Aikema

ASKER

Guess I'll be spending some time looking at this (http://www.viaarena.com/?PageID=89#ide) tomorrow
Hi,

I have experienced the same trouble as you did with a DELL computer. Actually, the problem comes from the 2.4.19 kernel (as stated ahoffmann). You better install the 2.4.18 one. It worked for me (until 2.4.20 comes out !).

You should then try
  hdparm -d1 -k1 /dev/hda
and everything should run faster.
Avatar of Aikema

ASKER

Well, 2.4.20 is now out and I built it last night.  I used hdparm and and saw the timing resulted in rates of about 10 times the previous amount.

I'm not very experienced with kernel-building yet, so I need to go clear up a few things with that kernel yet before I put it into use (sound, video, frame buffer, among a few other things).
Avatar of Aikema

ASKER

Well... building 2.4.20 seemed to fix things so I guess you must be right.