tertiary ide controllers

tertiary ide controllers

Post by David GJ Rober » Sat, 13 Sep 1997 04:00:00

Or even quadtenary (however you spell it).

Now work this one out:  Two Ide controllers on mainboard, and a standard
controller on an ESS Soundcard (port 168h & irq 12) (odd?) Well, plug in the
CD-ROM drive (which does work okay), and it causes the server to lock up.
Unplug it and all is fine. Surely this shouldnt happen? By the way this i/f
has only a 4x CDROM attached. I'll try the 18x one tonight (if I get back
from hounslow...). More specs attached below..

1st master - 7540AV Maxtor         (3
1st second - Western Digital 1170  (0
2nd master - ST5540A               (4  and very thin
s/c ide    - Samsung CS-36R CDROM  (0

Award 586 plug and pray bios rev 7/4/97

whoami:   David J Roberts                -----------------------------------
HOSTNAME: hayley.demon.co.uk             | 686L-166+, 16M/512k 540/540/170 |

httpd:    http://www.hayley.demon.co.uk  |   (including smb & mars_nwe:)   |


1. tertiary (soundcard) IDE controller Q

What's the level of support for a "tertiary" IDE controller? -- typically,
these are found on sound cards primarily for the support of CD-rom drives.
In fact, I saw two sound cards at Fry's -- the $14.95 one didn't have the
IDE interface, the $15.95 one did -- close inspection of the cards reveals
the only difference appears to be the one with the interface simply has the
header-pins soldered in the board.  (I believe it's an OSS based sound

I suppose that the "cost" of me soldering in a set of pins would easily
exceed $1 just waiting for the iron to heat up, so there's no point in
buying the cheaper one and "doing it myself"

However, the real reason I'm interested in this is because I intend to use
the two on-board IDE controllers to do a proper "raid-0" stripe set -- as I
understand it, I should place the two drives to be "striped" on seperate
controllers to maximize throughput AND I should keep the "slow" devices
(older IDE drive and CD-rom) off of the busses.  

Is this 100% accurate?  Can modern-day uDMA devices work somewhat like SCSI
devices in that the "control commands" can be sent to one drive, then while
the "drive" is working on that, additional command(s) can be sent to the
secondary drive on the same channel? (a feature known as "disconnect", I
believe)  I'm getting about 12mb sustained for a single drive with all the
hdparm options set that I can (32bit interface, DMA support, multi-sector
reads, etc.)  Since in theory the interface is 33mbs, two devices running
at 12mbs should run fine without using up the full capability of the
interface (and yes, I realize that "33mbs" is the "burst" rate, but with
two drives interleaved, the "burst" will appear to run longer...)

Also, is there really a performance penalty for non-UDMA devices simply
being "on" the chain if they are not being accessed?  I expect the only
times I'll actively use the CD-rom drive would be to install software or
read something from the LDP, at which point speed is not a critical issue.
Likewise, the older non-uDMA IDE drive is "only" 1 gig (I can tune it up to
about 3mb/sec), so this will be relatively low usage/priority files [maybe
archived source files?]

This brings up the last question -- can the tertiary controller be used to
WRITE to (or control) "normal" disk drives in the first place?  I suspect
the answer is no, but I'm willing to accept the fact that perhaps that's a
"windows" limitation... ;)

[I suppose I could just spend the 15 bucks -- at the very least, I'd have
an incentive to make it work if indeed this proves "problematic"...]

2. snm recource????

3. Tertiary ide port undefinable with 1.2.13(elfs) using non-LILO boot?

4. ?Can Batmon 'sleep' button include screen lock?

5. Tertiary and quaternary IDE interface?

6. SSH2 Problems

7. tertiary IDE drive, finding base address

8. comp.unix.aix Frequently Asked Questions (Part 1 of 5)

9. Linux installation of TERTIARY ide channel???

10. Lilo troubles (tertiary IDE)

11. tertiary IDE port? /dev/hd2x?

12. Tertiary (E)IDE Ports and Instalation

13. Fix for promise RAID controllers not working as IDE controllers in 2.4.19