Subject: Re: IDE controllers (on Alpha)
To: Arto Huusko <armihu@utu.fi>
From: Richard Rauch <rkr@olib.org>
List: netbsd-users
Date: 11/14/2003 11:10:16
Re. http://mail-index.netbsd.org/netbsd-users/2003/11/14/0006.html

About 2 years ago, I had a hard drive with terrible seek times.  It
was a Maxtor 20GB running under control of a VIA motherboard's (i386)
IDE.  One of the things that I tried was dropping in an ATA100 controller.
The controller didn't help.  (Though it worked.)  The drive was replaced
under warranty.  The new drive was fast again---but even faster with the
ATA100 controller.  (Less than a year later, the new Maxtor hard drive
also began to develop terrible seek times.  In disgust, I didn't even
bother with the warranty, and just bought a Western Digital.  That was
the last Maxtor drive that I expect to buy for a good long while.)

(I know that Maxtor didn't repair-and-return the first drive, but that
the second Maxtor was a completely separate drive.  I know this because
I kept the misbehaving drive in my system until the replacement drive
arrived.)


Anyway, the ATA100 controller is a Maxtor drive with a Promise chipset.
dmesg lines are:

pciide1 at pci0 dev 9 function 0: Promise Ultra100/ATA Bus Master IDE Accelerato
r (rev. 0x02)
pciide1: bus-master DMA support present
pciide1: primary channel configured to native-PCI mode
pciide1: using irq 10 for native-PCI interrupt


Once in a great while, I've known the system with this card to lose an
interrupt on the disk.  Since having more time for NetBSD, I haven't seen
the error repeat, I think, so I don't think that I've ever filed a PR on
it.  It happens very rarely.

(No, I didn't send the old Maxtor drive to seek for the interrupt that
was lost on the disk.  Just to forestall any clever remarks.  (^&)

If I needed assurance of absolutely flawless performance, I would avoid
it.  I can't remember what the consequences were of the lost interrupt,
though I vaguely recall that it was indicative of the kernel/card entering
some kind of mode where errors were frequent until I rebooted---but once
booted cleanly, it runs without problems.

For my practical needs, I'm pretty happy with the card.  (My system is
generally up 24/7, hosts a low-volume nameserver, acts as my home LAN's
fileserver, compiles, plays music, compiles stuff, runs OpenGL (Mesa)
in an X server, ...)


YMMV, and it's an ATA100 not ATA133, but still...(^&


-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/