RE>Ultra SCSI-2

Skip Jones sk_jones at
Tue Oct 3 16:27:24 PDT 1995


As Tak mentioned, it is my understanding that DIFFSENS rules if the bus is SE
or LVDS.  Since I am mostly a marketing-type, I do not have much input as to
the technical merits of this assumption.  However....

The primary marketing advantage to the open market of supplying a 'new' product
during the cusp of a transitioning or new technology which supports each side
of the cusp is that the user does not have to change everything immediately to
the new technology in order to purchase the 'new' product.  It allows a new
infrastructure to be phased-in at a rate comfortable to the user.

For instance, say a user needs to upgrade his SCSI disk drive to more capacity.
 The user can buy a new drive today with the necessary capacity which happens
to be capable of Fast-40/LVDS data xfers.  The user can use the new drive with
the old SE Fast-10/20 host adapter until such time the user is prepared to
upgrade the Host Adapter to FAST-40/LVDS.

A secondary advantage for the Vendor's of such products with regards to
Operations is that it reduces part numbers, prevents inventory segregation, and
simplifies material planning. 

I agree with Tak's point regarding a disk drive's buffer bandwidth.  Any
interface (Fast-40, Fast-80, SSA, FC-AL, etc.) which boast speeds above 40-60
MB/s tend to catapult disk drives into a new realm of additional cost due to
the increased buffer bandwidth requirements. Looking out a year or two, if
drives sport a disk rate of 15-20 MB/s, then these higher speed interfaces are
going to require expensive 32-bit buffer memory configurations to maintain
sustained throughput greater than 70 MB/s.  Assuming memory prices continue to
maintain fairly consistent price-points and drive vendors do not become eager
to absorb less margin percentages, such a reality could pose quite a speedbump
to the general market acceptance of any of these new whiz-bang interfaces.

Skip Jones,
Marketing Manager, QLogic
From: Tak Asami on Tue, Oct 3, 1995 10:41 AM
Subject: Re: Ultra SCSI-2 (This title is too seriously funny: let's fix it..)
To: scsi at

Lee Cleveland of IBM wrote:
> Is there any more documentation that describes a POWER up sequence where all
> devices assume singled ended and at some point switch to differential
> based on the system letting everybody know?

No, not really, as far as I know...

My understanding is that terminators define the diffsens line voltage level
to inform all devices on the bus that it is a LVDS bus.  If not, the terminator
has to change itself from "ideal diode" circuit to the "active" termination
circuit for single ended.  Therefore, you immediately know if you are on 
LVDS bus or not when you turn on.

It is not clear to me if multiple devices on the bus can drive diffsens line
to force it to differential or single ended.  If it cannot (I don't think 
it can), the terminator decide what type of bus it is, so a single
ended bus stays SE and LVDS bus stays LVDS as long as they live.

Well, someone correct me if I'm wrong, but if I am right, then why is it
so important to have that "backward compatibility?"  Originally, I thought
it was a good idea, but more I think about it, it really doesn't save you
very little.

Only potential is *IF* chip vendors put out the chips with do-either-one
drivers *at the same price* as the single-ended only chip (which cannot
exist by definition), then drive vendors can put out the device that can
hang on either SE or LVDS, but they are not expected of changing the 

Well, think again.  For Fast-40 wide bus, your burst rate is 80MB/sec.
Your on-board buffer must be build to allow that kind of bandwidth, but
it won't be utilized for single ended case, since it is not likely that
you can drive Fast-40 single ended (or will we?).  That's a major cost
compared to a mere protocol chip.  For a "lower cost" SE only drive, that's
the first thing to go.

I guess I am not complaining, but want to understand exactly what are we
driving to....  I'm sorry if I am boiling over the subject that were already
discussed in the past meetings (which I must have missed).  But I do 
appreciate if someone can explain it to me....
                           Tak Asami
_|_|_|                     asami at
_|/ _| WESTERN DIGITAL     (714) 932-7621 : Voice
_|_|_|                     (714) 932-6496 : Fax

More information about the T10 mailing list