Conference 7.286::fddi

re:             <<< Note 31.1 by MIPSBX::thomas "The Code Warrior" >>>
                        -< HPPI is different than FDDI >-


Ok...  I'll accept that.  I was under the impression that HPPI was a type
of adapter used for FDDI.  

Care to expound on the difference?  Is HPPI _RELATED_ to FDDI, is it 
something completely different, or what??  

Thanks....   --chas

    This is something that HPS and LDP Marketing are working on . Contact
    Harold Hager DUGGAN::HAGER for more info.
    
    Debbie
    

...but in one sentence: it's completely different.

	paul

>               <<< Note 31.4 by KONING::KONING "NI1D @FN42eq" >>>
>
>...but in one sentence: it's completely different.
>
>	paul

THANKYOU, Paul.

	--chas

    Having just heard a little about HPPI at Network University, it sounds
    as though HPPI is basically a parallel bus running either 800 or 1600
    MBPI, to be used primarily within a very local environment, such as a
    computer room.
    
    If anyone wishes to expand on this, please do so, or point us to
    another notes file where more on HPPI can be learned.
    
    Al
    

HPPI is a high speed channel that grew out of work at Los Alamos several years
ago.  Today many vendors are involved and there is an IEEE (ANSI?) standards
commitee working on a draft standard.  There are several ideas about the
physical layer, but at the moment the draft standard only specifies multiple
coax in two widths, I think it is 32 bits and 64 bits.  This means something
on the order of 40 and 70 pairs per cable.

As far as the wire protocol is concerned, there is nothing that really specifies
the frame format; in fact, the initial motivation was strickly to move images
from Crays to display stations in realtime.  Data is moved in `bursts' and part
of the burst includes information on the make-up of the burst.  A rough proposal
includes a way to specify a data frame from one or more bursts.  However, this
proposal is only minimally satisfactory: it's ok as far as specifying that
a frame is being transmitted, but it is extremely weak in ensuring that the
start and end of the frame is correctly detected.  Once that problem is worked
out, the next step will be to settle on a frame format and some sort of
error detection code or possibly error correction code.  (At the data rates
that this in intended to operate at, any errors in transmitting a datagram
are extremely expensive, whether they are detected or undetected errors.)
So, the state of the standard is that it is far from complete and as yet offers
absolutely no guidance in how it can be used for communication.

There are other product development efforts that are going on which will give
the impression that HPPI is real.  IBM for example has announced support, but
what they and others are doing is building hardware to the specs of Los Alamos
and others.  Meanwhile everyone is supporting the standards effort.

This reminds me of FDDI; FDDI began many years ago amid much hoppla, but as yet
there are no real FDDI products and some key parts of the standard are yet to
be approved (although it is finally rapidly drawing to a close).  I think that
many of the current participants in the HPPI effort are taking a lesson from
FDDI and are not waiting for the standard to start building products.  However,
HPPI won't have much impact except in the high end until the standards are
much better defined and they change the physical layer to fiber or similar,
(you aren't going to string much 100-150 conductor cable into individual offices).

So, in summary, HPPI is able to do things that FDDI can't (digital images) and
can't yet do things that FDDI can (datagrams).  The focus of the two are today
quite different.

    "...but as yet there are no real FDDI products".
    
    That's not a true statement anymore.
    
    	paul

    What happened to the contents of reply 31.1?
    
    Could someone point me to a draft copy of the X3T9.3 (I think that's
    right) spec for HPPI (formerly known as HSC <= name changed for obvious
    DEC copyright reasons).
    

    There was no contents.  Matt just likes to write one-liner answers and
    put them in the reply title rather than the text...
    
    Marty Halvorson might be able to help you find a draft copy.  
    
    	paul