Jaime,
Here goes.... :-)
Short Version:-
Quick check of your BIOS setting should show:-
Primary Master Auto Large or CHS
Primary Slave Auto LBA
Set both to LBA for extra speed.
I'd also look at putting each drive on separate cables, that is if your
CD-ROM is UDMA or better. It may also slow things down. I don't know your
system...
Long Version:-
The 8G limit old BIOS's had was Cylinders (C) couldn't be above 1024, Heads
(H) above 16, Sectors above 64. Giving 1024x16x64 = 1048576 x 8K sectors =
approx 8G
LBA just goes ahead and gives whatever the driv is set to, as you can see
from the bad drives H geometry.
bad one geometry = 2434/255/63, sectors = 39102336, start 0
good one geometry = 38792/16/63, sectors = 39102336, start 0
As the logical geometry of 'good drive' is the same as the physical
geometry, there is fewer overheads to writing the data.
Regards,
Craig F.
-----Original Message-----
From: Jaime Nebrera [mailto:[EMAIL PROTECTED]]
Sent: Saturday, 10 March 2001 1:01 AM
To: [EMAIL PROTECTED]
Subject: [e-smith-devinfo] IDE disks and hdparm
Hi everyone,
I have just installed a stock 4.0.12-r56 on a brand new test computer.
The exact configuration needed for this problem is:
AOpen MX3S motherboard
2 Seagate UDMA-100 20 GB Hard Disks
Ok, the board supports ATA 100 and so do the disks. I am being told the HD
are exactly the same. One HD is installed as the primary of the first IDE
channel and the second is the primary of the second IDE channel.
I install the system from an ISO image (4.0.12-28) and updated it
(4.0.12-56). Now I format the second HD with the order:
mkfs -t ext2 /dev/hdc
And edit fstab (template that is) to see it.
I start optimizing the first HD till I reach an optimum of 9,75 MB/s with:
hdparm -c3 -m16 -d1 /dev/hda
If I try more risky ones the system hangs.
For the second HD I run the same order and get 23,10 MB/S !!!! Thats more
than 2 times improvement !!! I don't understand why I'm seeing this
difference in results. I run hdparm to see the configuration of each one and
see only one difference:
bad one geometry = 2434/255/63, sectors = 39102336, start 0 (deffault
esmith after optimization)
good one geometry = 38792/16/63, sectors = 39102336, start 0 (default
mkfs after same optimization)
Please, some help?
-----
Jaime Nebrera Herrera -- [EMAIL PROTECTED]
Area de Nuevas Tecnologias -- INDEMAN S.L.
--
This list is archived
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]