Has anyone experimented with this command? Did it speed up your system much? Anyone had any bad experiences? Anyone got any recommendations for parameter settings?
Keith ____________
From the first not a thing is. Hui-neng
On Wed, 28 May 2003, Keith Watson wrote:
Has anyone experimented with this command?
Yes.
Did it speed up your system much?
Yes.
Anyone had any bad experiences?
Yes. Reboots due to misplaced hard disks.
Anyone got any recommendations for parameter settings?
Yes: hdparm -c3 -m16 -X66 -d1 -u1 /dev/hda or similar.
man hdparm
http://linux.oreillynet.com/pub/a/linux/2000/06/29/hdparm.html
Caveat emptor.
Andrew.
On Wednesday 28 May 2003 11:26 am, Keith Watson wrote:
Has anyone experimented with this command?
Yes
Did it speed up your system much?
No
Anyone had any bad experiences?
No
Anyone got any recommendations for parameter settings?
Use -t to find currently achievable data transfer rate. Properly set up systems return figures in the 20 to 50MByte/s range
Use -I to find info about drive and settings available. Those with a * against are current.
Most distros set hd up for optimum without user intervention unless htere is a kernel bug (as in rh7.3). If -t returns a reasonable value then leave alone.
Ian
From: Ian Bell On Wednesday 28 May 2003 11:26 am, Keith Watson wrote:
Has anyone experimented with this command?
Yes
Did it speed up your system much?
No
Anyone had any bad experiences?
No
Anyone got any recommendations for parameter settings?
Use -t to find currently achievable data transfer rate. Properly set up systems return figures in the 20 to 50MByte/s range Use -I to find info about drive and settings available. Those with a * against are current. Most distros set hd up for optimum without user intervention unless htere is a kernel bug (as in rh7.3). If -t returns a reasonable value then leave alone.
Had a play with hdparm last night and my experience was the same as Ian. In fact the settings seemed to be at optimum by default, everything I tried just slowed things down.
Regards,
Keith ____________ 'If there is a better solution...find it.' - Thomas Edison
On Thu, 29 May 2003 09:16:39 +0100 "Keith Watson" keith.watson@kewill.com wrote:
Had a play with hdparm last night and my experience was the same as Ian. In fact the settings seemed to be at optimum by default, everything I tried just slowed things down.
I didn't find anything that speeded up the transfer rate as reported by hdparm -t - there was as much variation between any two tests with the same settings as between different settings.
From: Steve Fosdick
On Thu, 29 May 2003 09:16:39 +0100 "Keith Watson" keith.watson@kewill.com wrote:
Had a play with hdparm last night and my experience was the same as Ian. In fact the settings seemed to be at optimum by default, everything I tried just slowed things down.
I didn't find anything that speeded up the transfer rate as reported by hdparm -t - there was as much variation between any two tests with the same settings as between different settings.
From previous experience the DMA setting is the one that makes a huge
difference in performance - without DMA I get 5Mb/s, with it 25Mb/s. DMA is enabled by default but can get turned off in the event of a DMA timeout.
Steve.
Direct reply, I think intended for the list :o)
Keith ____________ A bend in the road is not the end of the road... unless you fail to make the turn.
On Thu, May 29, 2003 at 10:00:04AM +0100, Steve Fosdick wrote:
On Thu, 29 May 2003 09:16:39 +0100 "Keith Watson" keith.watson@kewill.com wrote:
Had a play with hdparm last night and my experience was the same as Ian. In fact the settings seemed to be at optimum by default, everything I tried just slowed things down.
I didn't find anything that speeded up the transfer rate as reported by hdparm -t - there was as much variation between any two tests with the same settings as between different settings.
From previous experience the DMA setting is the one that makes a huge
difference in performance - without DMA I get 5Mb/s, with it 25Mb/s. DMA is enabled by default but can get turned off in the event of a DMA timeout.
Generally another major performance increase you will see by enabling DMA is a much lower cpu usage.
Adam
-
Adam Bower -
Andrew Savory -
Ian Bell -
Keith Watson -
Steve Fosdick