SCSI Bus Problem?

From: Julius Sridhar <vance_at_ikickass.org>
Date: Mon Feb 11 18:52:10 2002

On Mon, 11 Feb 2002, Tothwolf wrote:

> > > A narrow device will not slow down a wide bus to narrow width. But a
> > > 10 MHz device will slow down a 20 MHz SCSI bus to 10MHz.
> >
> > You're contradicting yourself here. Either that or the semantics of
> > your sentence are confusing :)
>
> SCSI itself is confusing...basically the bus will operate at the fastest
> rate of the slowest device on the bus.

This is simply not the case.

> Original - narrow, 5Mb/s [used by lots of older devices]
> Fast - narrow 10Mb/s [very common]
> Ultra - narrow 20Mb/s [very common]
> Wide - wide, 20Mb/s [common for a short time]
> Ultra-Wide - wide, 40Mb/s [very common]
> Ultra2 - wide, 80Mb/s [never really caught on]
> Ultra160 - wide, 160Mb/s [starting to become common]
> Ultra320 - wide, 320Mb/s [not common yet]

Ok. First of all, it's MB/s, rather than Mb/s. Second, MB/s is not a
good way to describe the bandwidth of the bus for technical reasons. A
fast/wide 20MB/s bus looks *nothing* like an ultra 20MB/s bus. This is so
because fast/wide is a 10MHz 16-bit bus and ultra is a 20MHz 8-bit bus.
Just because they have the same bandwidth doesn't mean anything. An ultra
device will run at 10MB/s on a fast/wide bus and a fast/wide device will
run at 10MB/s on an ultra bus. Both busses are *capable* of 20MB/s. If
you have an ultra device and an ultra/wide device on an ultra/wide bus,
the ultra device will run at 20MB/s and the ultra/wide device will run at
40MB/s *simultaneously* **on the same bus**.

> (info from memory, might be incomplete, inaccurate, etc, etc)
>
> There are also Differential versions of Fast, Ultra, Wide, and Ultra-Wide.
> These use a "high voltage" (+-12VDC IIRC) signaling that is *NOT*
> compatible with standard devices. You will literally fry any non HVD
> devices if you connect a HVD drive to the same bus. Ultra2 and newer have
> a Low Voltage Differential bus, I'm not sure if there is a HVD
> specification for those.

Signalling voltages are irrelevant to this discussion.

Peace... Sridhar
Received on Mon Feb 11 2002 - 18:52:10 GMT