Re: [linux-audio-user] Ardour, Jack, and 2.6 kernels + XRuns with the Audiophile 24/96 M-Audio

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
To: A list for linux audio users <linux-audio-user@...>
Date: Wednesday, June 2, 2004 - 1:17 am

> I think that I've now actively avoided each major manufacturer three

*LAUGH* Sounds familiar.

> Anyone else remember the stiction problem in the Seagate 105MB drives?

Quantums had the problem as well. I vividly remember it at the time,
because I had to eat some crow for my public position to fellow engineers
about how "computers operated better with a light touch than with a
sledgehammer". Oddly, I've never owned a drive which needed this, but I
have had cohorts who did - and it was always a delight when I showed them
how to "fix" their "broken" drive. The first time, anyway. I had to
restrain some of them who would start to get overenthusiastic.

> And people used to say that SCSI was voodoo.

SCSI was really only voodoo because they ignored the Cardinal Rule of all
Cardinal Rules: NEVER HOT (UN)PLUG SCSI DRIVES. There was an insanely
stupid "urban myth" that you could just do this with abandon, not helped by
the fact that some times when you did this you'd probably not blow any
microfuses (or ones you needed to work properly). But regular violators
would inevitably find themselves stuck with full-on Bad Mojo SCSI "Voodoo".

Symptoms like:

(o) Drives only seem to work when set to certain SCSI IDs.
(o) Drives only seem to work in certain configurations, i.e., behind or
in front of certain other drives.
(o) Drives which only work when they're the only thing on the SCSI bus.
(o) Drives which only work when they're the first device in the chain.
(o) Drives which only work when the termination is removed, even if it's
at the end of the SCSI bus.

I've found that various Macintosh DTP and Service Bureaus would kvetch madly
about SCSI Voodoo, whilst blatantly ignoring the no-hot-reconnect/disconnect
rule.

Note: You can safely power-on a SCSI device that's been already connected
into a live system, however, if it's the end-device, it really should be
powered on [to ensure a proper supply of termination power].

The rules of SCSI are simple: provide termination at both ends of the chain
[most motherboards provided their end, so you only need to worry about one],
make sure there's enough termination power [ideally supplied from both
extrema of the bus cabling], and make sure the IDs are all different.

What I find annoying is how fragemented SCSI became after SCSI-2. So many
connectors, signalling methods, etc. What a mess.

=MB=
--
A focus on Quality.

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
Re: [linux-audio-user] Ardour, Jack, and 2.6 kernels + XRuns..., Malcolm Baldridge, (Tue Jun 1, 10:51 pm)
Re: [linux-audio-user] Ardour, Jack, and 2.6 kernels + XRuns..., Malcolm Baldridge, (Wed Jun 2, 1:17 am)