Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: Implementation vs. Standard

Or in the immortal words of 802.3 (clause 1.2.1 and other places):

The models presented by state diagrams are intended as the primary
specifications of the functions to be provided. It is important to
distinguish, however, between a model and a real implementation. The models
are optimized for simplicity and clarity of presentation, while any
realistic implementation may place heavier emphasis on efficiency and
suitability to a particular implementation technology. It is the functional
behavior of any unit that must match the standard, not its internal
structure. The internal details of the model are
useful only to the extent that they specify the external behavior clearly
and precisely.

I don't know of any implementations that put the MAC state machine in real
Pascal code.


-----Original Message-----
From: Booth, Bradley [mailto:bradley.booth@xxxxxxxxx]
Sent: Tuesday, March 27, 2001 10:02 AM
To: 802.3ae
Subject: RE: Implementation vs. Standard


If the circuit you design complies with the standard, how is anyone to know
how you implemented it?  What matters is what it seen on the MDI.  If that
looks correct, then you can implement it in FORTRAN for all I care.


		-----Original Message-----
		From:	Justin Gaither [mailto:jgaither@xxxxxxxxxxxxxxx]
		Sent:	Tuesday, March 27, 2001 10:59 AM
		To:	Taborek, Rich; 802.3ae
		Subject:	Implementation vs. Standard

		 << File: Card for Justin Gaither >> Rich,
			a while back I sent a message about PUDI(/INVALID/)
and decoder
		placement.  In your reply you said:
		Rich Taborek wrote:
		> Justin,
		> The Sync state machine is written to operate in the 10B
domain as is the
		> Deskew state machine.
		> The 10GBASE-X PCS does not mandate the location and number
of decoders
		> in an implementation. The PCS Sync and Receive state
machine structure
		> is virtually identical to that of the 1000BASE-X PHY. The
		> function can be located elsewhere, but I don't view such a
change as
		> being more or less "correct" than it's current location.
		> As to your second point, Invalid represent a superset of
		> code-groups and running disparity errors. Therefore,
running disparity
		> error is not a suitable replacement for Invalid.
		> Best Regards,
		> Rich

		I am new to the standards based design, so I dont quite
understand where
		the line between standard and implementation is drawn.

		It was my understanding from the spec. that the "PCS shall
implement its
		Transmit process as depicted in Figure 48..., including
compliance with
		the associated state variables as specified in 48..."
means, I have to
		put the decoder where specified, and not before the deskew
and sync
		state machines.  Now I know I could design a circuit that
behaves the
		same as the standard but place the decoder before the deskew
and sync,
		but is this still compliant?  Can I still select Yes in the
		meeting the requirements of the State Machines?


		Justin Gaither                       Phone: 512-306-7292
		RocketChips a Division of Xilinx     Fax:   512-306-7293
		500 N. Capital of TX Hwy.
		Bldg 3                         email:
		Austin, TX 78746               WWW: