View previous topic :: View next topic |
Author |
Message |
vert Apprentice
Joined: 07 May 2002 Posts: 214 Location: Delft, The Netherlands
|
Posted: Mon Jun 02, 2003 9:56 am Post subject: Slow (buffered) disk reads while dma etc is optimized |
|
|
I got a system in which there are 4 hard drives. All report about the same (slow...) disk reads with hdparm. This averages around 155MB/sec for buffer-cache reads and between 34 and 45 MB/sec for buffered disk reads.
Except hda. This is the root drive and hdparm reports around 14 MB/sec for buffered disk reads. What's up with that? I turned on all optimization with hdparm (see below) and still this slow. What could be the problem. The server is currently running 2.4.19-gentoo-r10.
output hdparm /dev/hda
Code: |
/dev/hda:
multcount = 16 (on)
IO_support = 1 (32-bit)
unmaskirq = 1 (on)
using_dma = 1 (on)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 128 (on)
geometry = 1027/255/63, sectors = 16514064, start = 0
|
|
|
Back to top |
|
|
BradN Advocate
Joined: 19 Apr 2002 Posts: 2391 Location: Wisconsin (USA)
|
Posted: Mon Jun 02, 2003 8:18 pm Post subject: |
|
|
Are they all the same model hard drive? It's possible the hard drive is just slower... 14MB/s is kinda low, but seems reasonable for older stuff. My 30GB laptop drive only gets 20MB/s. |
|
Back to top |
|
|
Beltzak n00b
Joined: 02 Jun 2003 Posts: 10
|
Posted: Mon Jun 02, 2003 8:59 pm Post subject: Dma ALI Chipset On kernel 2.4.19 Failed |
|
|
You should upgrade your kernel to 2.4.20 or use an ac kernel.
My trouble was ALI Chipset bug and hdparm gaves not more than 8Mb on disk read ;( Now that's solve |
|
Back to top |
|
|
vert Apprentice
Joined: 07 May 2002 Posts: 214 Location: Delft, The Netherlands
|
Posted: Tue Jun 03, 2003 6:56 am Post subject: |
|
|
Could be indeed the age of the drive. It just seemed so slow...
And I'm planning on updating, but since it's our home server, it is kind of a hassle. I'm actually thinking on upgrading to 2.5 since I'm pretty happy with that on my own system. |
|
Back to top |
|
|
|