Chris Mathey wrote:
> Hi!,
> I just wanted to chew some bandwidth and introduce myself. I've been a
> lurker for awhile and see that the folks on this list are top notch :)
> To give you a little background on me. I am a Sr. level network
> engineer for one of the nations largest banks. Hopefully I can add
> some network knowledge along with the networking experts on this list.
> I've been using Linux since circa 2000 and have played with alot of
> distros\apps.
>
> Ok so.. for my first question
> on my home server I was playing with hdparm. (suse 9.2)
>
> *my main disk reads:
> /dev/hda:
>
> UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5
>
> * and tests @
> /dev/hda:
> Timing buffered disk reads: 100 MB in 3.00 seconds=
> 33.32MB/sec
>
> ** my "storage" disk reads:
> /dev/hdc:
>
> UDMA modes: udma0 udma1 *udma2
>
> ** and tests @
> /dev/hdc:
> Timing buffered disk reads: 70 MB in 3.05 seconds = 22.95
> MB/sec
>
> Both disks are capable of UDMA mode 5, however /dev/hdc only lists up
> to udma2. Why do you think mode 5 is not listed? I haven't checked but
> It's possible a CD-ROM is slaved on the same channel. Could that be it?
>
> Does /dev/hda's timing look around par?
I've found that even newer drives, when used with older parallel IDE
cables (not Ultra-ATA high-density cable where the extra wires are
grounded to improve signal transmission by removing bleedover from
parallel wires) tend to report as udma2 max.
If you see a set of listed speeds, like "udma0 udma1 *udma2 udma3 udma4
udma5", this would mean that the systems controller is actually talking
at UDMA mode2 even though the drive and controller are capable of
talking at up to UDMA mode5.
You can fix the above problem by using the "hdparm -X69" style argument.
It's 64 + the mode number, IIRC.
You can also fix the above by compiling your kernel to use ultra-DMA by
default, and enabling the "automatically detect fastest rate possible"
option. Can't remember what #define that is off the top of my head, but
it's in the list.
If you see a limited list of speeds, like "udma0 udma1 *udma2" on a
newer drive, I would suspect:
1. The wiring.
- Replace any "old" low-density ATA cables with higher-density ultra-ATA
cables.
- Replace any "split" high-density ribbon cables using a plastic
conduit, with a normal solid ultra-ATA IDE ribbon cable.
2. The controller
- Make sure you're using the correct driver for your controller. There
is an older IDE driver that can
be compiled into the kernel (for ancient system), and sometimes multiple
IDE drivers for a given chipset (libsata, etc)
- Make sure that your controller is trying to use the maximum speed
possible to talk to a drive.
- Make sure your system remains stable after making this change.
Yes, the speeds you listed are pretty slow. Depending your drive, the
22MB/s 33MB/s might be as fast as the drive can go, but any recent
system should typically be 40MB/s or higher on a mode5 Ultra-ATA drive
(it's a measure of speed just to the drive buffer, which should be
relatively high).
On a VIA system with "split" high-density ribbon cables at home (I had
to use them to make two drives fit, long story), I still get some
relatively good numbers:
hdparm -t -T /dev/hda
/dev/hda:
Timing cached reads: 488 MB in 2.00 seconds = 243.79 MB/sec
Timing buffered disk reads: 76 MB in 3.06 seconds = 24.81 MB/sec
On an idle Athlon XP system, properly wired, I get better numbers:
hdparm -t -T /dev/hda
/dev/hda:
Timing buffer-cache reads: 128 MB in 0.37 seconds =345.95 MB/sec
Timing buffered disk reads: 64 MB in 1.09 seconds = 58.72 MB/sec
My guess is that you may have a cabling issue here.
- Ian C. Blenke <ian@blenke.com> http://ian.blenke.com/
-----------------------------------------------------------------------
This list is provided as an unmoderated internet service by Networked
Knowledge Systems (NKS). Views and opinions expressed in messages
posted are those of the author and do not necessarily reflect the
official policy or position of NKS or any of its employees.
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 16:17:54 EDT