17 May
2013
17 May
'13
9:05 p.m.
On Fri, 18 May 2013, Dave McGuire wrote:
I agree. At the time it was simply a shortsighted decision, as were nearly
all decisions that brought the PC into existence.
I've yet to find a design decision in the PC world that WASN'T
shortsighted...I even see it in processor designs...
On 05/17/2013 11:32 PM, der Mouse wrote:
Well yes. If you keep a SCSI bus as short as the average IDE bus, it will
usually work unterminated. Electrically, IDE does actually need
terminators...they're just not required by the spec, nor provided for in
drives or cabling.
That'd explain why there's only two drives on a cable maximum with
IDE...yet my SCSI stuff was up to 8 or more. ;)
And
don't get me started on the whole "ATA is better than SCSI
because it doesn't need terminators!" crap that n00bs spill all the
time. ;)
C'mon, be fair; that _is_ one of its (few) advantages over SCSI: it
does not, in practice, need separate terminators. Just because you
(and I, and pretty much everybody else sane) consider it far outweighed
by other differences for most purposes doesn't mean it's not an
advantage. I've seen it said that the spec calls for terminators; even
if so, it does not actually need them in practice, so it's a practical
advantage.
Of course, that's not what the n00bs are *trying to* say, but it _is_
what they *are* saying.
Also, SCSI is pretty robust in that regard; if your SCSI bus is as
short as your typical IDE cable, I suspect it may not actually need
terminators in practice either. I haven't tried that. But I _have_
seen SCSI buses work when triple-terminated; I even saw one
kinda-mostly work when quadruple-terminated.
To be fair, it was never intended as an external bus, and very rarely used
as such, while SCSI was intended to be such from day one.
You're telling me IDE was once used as an external bus by some people?!
No
technical person beyond the basement hobbyist stage would design
such an interface.
Today? No, perhaps not. But, at the time, I'm not convinced it was a
wrong choice. Remember what IDE stands for - it was in comparison to
disks where the smarts were basically all on the interface card; moving
them physically to the drive was a big deal, if only because it meant
that every drive had its own copy of them, with all the expense that
implied.
Note the past tense. Today, the cost of the hardware to implement
those smarts is down in the noise. Then, it was significant, and
that's part of why I hold the opinion I expounded above.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
--
Cory Smelosky
http://gewt.net/ Personal stuff
http://gimme-sympathy.org Experiments