IPG (Inter Packet Gap) in Full duplex mode for MII interface

IPG (Inter Packet Gap) in Full duplex mode for MII interface

Post by pran » Wed, 18 Jun 2003 18:36:10



Hi all,
   Just a simple question and that too to confirm. I think whether
Half duplex or Full duplex mode the IPG in case of MII interface is 96
bit time. Am i right on this or is it the case that in Full duplex
mode there is no need for min ipg of 96 bit time?
 thanks in advance,
pranav
 
 
 

IPG (Inter Packet Gap) in Full duplex mode for MII interface

Post by Rich Seifer » Thu, 19 Jun 2003 00:32:40




> Hi all,
>    Just a simple question and that too to confirm. I think whether
> Half duplex or Full duplex mode the IPG in case of MII interface is 96
> bit time. Am i right on this or is it the case that in Full duplex
> mode there is no need for min ipg of 96 bit time?
>  thanks in advance,
> pranav

The interframe gap is 96 bit times, regardless of duplex mode. (The
need, or lack of need, for an interframe gap is unrelated to duplex
mode.)

--
Rich Seifert              Networks and Communications Consulting

(408) 395-5700            Los Gatos, CA 95033
(408) 395-1966 FAX

 
 
 

IPG (Inter Packet Gap) in Full duplex mode for MII interface

Post by Glen Herrmannsfeld » Thu, 19 Jun 2003 01:32:36





> > Hi all,
> >    Just a simple question and that too to confirm. I think whether
> > Half duplex or Full duplex mode the IPG in case of MII interface is 96
> > bit time. Am i right on this or is it the case that in Full duplex
> > mode there is no need for min ipg of 96 bit time?
> >  thanks in advance,
> > pranav

> The interframe gap is 96 bit times, regardless of duplex mode. (The
> need, or lack of need, for an interframe gap is unrelated to duplex
> mode.)

I suppose I agree that it should stay 96.

But since part of the IFG is to allow for IFG loss in repeaters, and full
duplex doesn't go through repeaters, it would seem that not all the need is
there in the full duplex case.

On the other hand, a bridge/switch should be able to take frames from a full
duplex link and forward them to a half duplex link, at full speed.  This
would be hard if the full duplex link had a consistently smaller IFG.

-- glen

 
 
 

IPG (Inter Packet Gap) in Full duplex mode for MII interface

Post by Rich Seifer » Thu, 19 Jun 2003 03:24:23






> > The interframe gap is 96 bit times, regardless of duplex mode. (The
> > need, or lack of need, for an interframe gap is unrelated to duplex
> > mode.)

> I suppose I agree that it should stay 96.

> But since part of the IFG is to allow for IFG loss in repeaters, and full
> duplex doesn't go through repeaters, it would seem that not all the need is
> there in the full duplex case.

> On the other hand, a bridge/switch should be able to take frames from a full
> duplex link and forward them to a half duplex link, at full speed.  This
> would be hard if the full duplex link had a consistently smaller IFG.

The original purpose of the IFG was (and is) to provide some "breathing
space" between back-to-back received frames. Especially in the "early
days", it was very difficult for an Ethernet controller to be able to
keep up with the incoming frame flow. There was a certain amount of
"housekeeping" that needed to be done on a per-frame basis, e.g.
updating management statistics, releasing and re-establishing DMA
pointers, etc. The IFG provided this minimum spacing for the receiver
designer.

Yes, it is an artifact of (half-duplex) repeaters that they can "shrink"
the interframe gap at a given instant (as seen by some particular
station), but they cannot reduce the *average* (i.e., the sustained)
interframe gap for a stream of back-to-back frames. That is, a repeater
can make a specific sequence of frames from an arbitrary set of stations
appear at some point on the cable to have an *instantaneous* (i.e., for
this specific pair of frames) reduction of the IFG. To reduce the
steady-state IFG, the repeater would somehow have to "shrink time"; they
simply cannot do this (except to the extent that their clocks may be
slightly faster than the clocks in the sending stations, but receivers
must already accommodate this difference).

--
Rich Seifert              Networks and Communications Consulting

(408) 395-5700            Los Gatos, CA 95033
(408) 395-1966 FAX

 
 
 

1. IPG requirements in FULL DUplex networks

Is there any effort by the 802.3u committee to propose an IPG of less than
96 bit times to allow for faster throughput?

(This could be possibly be in the mii management reg. space to decide on
IPG/3 or IPG/2 etc.  )

--
 --
Thanks,

 Kumar

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Kumar Bhattaram                       LSI Logic Corporation
  Core Systems Engineering,             1525 McCarthy Blvd., MS G-813
                                        Milpitas, CA 95035
  Fax: (408) 954-4874

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

2. log-on for specific exclusive users

3. Inter packet Gap

4. RPC under MS-DOS

5. 9.6 uS Inter-Packet Gap

6. Hardware image (video) decompression for PC or AMIGA

7. do I need full-duplex modem, full-duplex soundcard, or both...

8. Maxtor 160GB on Win2k only showing 128GB

9. 100baseFX full duplex interframe gap?

10. manually set Cisco switch ports to 100 full duplex mode in some cases slower

11. monomod fiber optic in full duplex mode

12. Full Duplex Mode How to configure