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

Re: [802.3_100GCU] Objective changes for the Back Plane Initiative



Title: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

Hi Brad,

That wasn’t quite my point.

 

I was trying to illustrate that anything we fill into the equation does constrain the solution space, and we have already filled in one number (4 lanes) – why is it more of a sin to fill in the baud rate than the lane count in terms of constraining the solution set that could satisfy the objective?

 

Also, the objective should align with what everyone agrees is technically feasible, so if the maximum baud rate needs to be specified in order that people agree that a solution is technically feasible over the indicated material, that would seem to be a tradeoff the task force could make in the process of keeping alignment between the objectives and 5C responses.

Regards,

Steve

 

From: Brad Booth [mailto:Brad_Booth@xxxxxxxx]
Sent: Wednesday, February 15, 2012 7:23 PM
To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

 

Steve,

Interesting point.

I was referencing the point Adam made. If I interpret it correctly, you like the idea of objectives constraining the potential solution set.

I agree with Adam's point about creating two applications/channels without constraining the solution sets. There are two potential channels: one based on 802.3ap and another that is created by 802.3bj.

Cheers,
Brad



-----Original Message-----
From: Trowbridge, Stephen J (Steve) [steve.trowbridge@xxxxxxxxxxxxxxxxxx]
Sent: Wednesday, February 15, 2012 06:43 PM Central Standard Time
To: Booth, Brad; STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_100GCU] Objective changes for the Back Plane Initiative

Hi Brad,

I think you have an equation:

Per-lane signaling rate = (103.125 Gb/s)  / (# of lanes) / (bits per symbol)) * (1+FEC overhead %)

 

Any number you fill into the equation is going to reduce the set of solutions available to the task force. We have already taken the step of specifying the number of lanes (a step that HSSG did NOT take), which constrains the bit-rate per lane but not the baud rate per lane. If you don’t constrain the baud-rate per lane (and hence force more bits per symbol), do you have consensus that it is technically feasible to signal at a 26 Gb/s rate over any 802.3ap backplane?

 

You could leave out the lane count and consider the merits of 8-lane NRZ vs 4-lane PAM-4 if you want to open up another degree of freedom in the solutions that may be considered by the task force.

Regards,

Steve

 

From: Brad Booth [mailto:Brad_Booth@xxxxxxxx]
Sent: Wednesday, February 15, 2012 3:12 PM
To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

 

Not sure that would address Adam's point:
"Specifying the signaling rate implies constraints on the encoding and framing."

Do we really need the signaling rate?

Thanks,
Brad



-----Original Message-----
From: Joel Goergen [jgoergen@xxxxxxxxx]
Sent: Wednesday, February 15, 2012 03:52 PM Central Standard Time
To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

Dan
I support this format ... It certainly addresses all my concerns.

Joel


On 2/15/12 1:48 PM, "Daniel Dove" <ddove@xxxxxxx> wrote:

  Hi Adam,
 
 Good point. My intention was not to restrict the possible coding options, but rather to provide an objective that did not force a technical decision (including channel loss, coding alternatives, etc) while providing clear direction on what we are trying to achieve (two channel specs and coding approaches).
 
 I would be happy to amend the suggestion to include all proposals currently under consideration, or even add a reasonable amount of margin on the baud-rate definition to address future enhancements.
 

1.      Define  a 4 lane 100G PHY for baud rates up to 26Gbd per lane over copper traces defined by the IEEE P802.3bj Task Force with lengths up to at least 1m.

2.      Define  a 4 lane 100G PHY for baud rates up to 14Gbd  per lane over copper traces defined by IEEE  Std. 802.3ap with lengths up to at least 1m.


 
Dan
 
On 2/15/12 1:30 PM, Healey, Adam B (Adam) wrote:

   Re: [802.3_100GCU] Objective changes for the Back Plane Initiative  
 

Colleagues,
 
 
 
Specifying the signaling rate implies constraints on the encoding and framing. Such degrees of freedom should be left to the Task Force to determine the solution for the application(s) to be served.
 
 
 
The Brown/Bhoja proposal for PAM-4 encoding and framing has a signaling rate of 13.75 Gbaud. Would this proposal not be allowed by the set of objectives worded in the form of option B? It exceeds 12.89 Gbaud.
 
 
 
We need to find a means to describe the properties of the two applications/channels that make them distinct without unduly constraining the solutions that would serve them.
 
 
 
--Adam
 
 
 
 
 
 
 

From: Brad Booth [mailto:Brad_Booth@xxxxxxxx]
 Sent: Wednesday, February 15, 2012 4:05 PM
 To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

 
 


I prefer option B.
 
 
 
It is simpler, and IMHO it represents more accurately what the end users would like to see.
 
 
 
Thanks,
 Brad
 
 
 
 
 
 
 

From: Tom Palkert [mailto:tpalkert@xxxxxxxx]
 Sent: Wednesday, February 15, 2012 2:56 PM
 To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

 
 


All,
 
I read my own email and realized that I should clarify that I was just offering a suggestion to keep things simple. If everyone is onboard for option A I am just fine with it.
 
 
 
Tom
 
 
 
 
 
 
 

From: Tom Palkert [mailto:tpalkert@xxxxxxxx]
 Sent: Wednesday, February 15, 2012 1:02 PM
 To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

 
 


All,
 
I would like to support Joel’s B proposal. We have not put loss numbers into the objective in the past and the 1m distance shows consistency with the previous (ap) project.
 
Lets keep it simple.  
 
 
 
Tom
 
 
 
 
 
 
 

From: Joel Goergen [mailto:jgoergen@xxxxxxxxx]
 Sent: Wednesday, February 15, 2012 11:26 AM
 To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative

 
 


Scott / Dan
 I am fine with either of these, though admit Dan’s looks more interesting then the edited version of mine.  Dan ... I changed back plane channels to copper traces.  I see the two choices as:
 
 A)
 

1.      Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz over links consistent with copper traces on materials defined by the Task Force with lengths up to at least 1m.

2.      Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz over links consistent with copper traces on materials defined by the Task Force with lengths up to at least 1m.



 B) (my favorite)
 

1.      Define a 4 lane 100G PHY for baud rates up to 25.8Gbd per lane over copper traces defined by the IEEE P802.3bj Task Force with lengths up to at least 1m.

2.      Define a 4 lane 100G PHY for baud rates up to 12.9Gbd  per lane over copper traces defined by IEEE  Std. 802.3ap with lengths up to at least 1m.



 Joel
 
 On 2/15/12 8:48 AM, "Scott Kipp" <skipp@xxxxxxxxxxx> wrote:
 
Joel,
  
 I was not trying to specify a single source and copied the style of the text that you had proposed.  My objective gives one example where I used the material that Beth had referenced.  
  
 I could agree to your proposed objective if it were:
  
 1) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz over links consistent with copper traces on materials defined by the Task Force with lengths up to at least 1m.
 2) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz over links consistent with copper traces on materials defined by the Task Force with lengths up to at least 1m.
 
 
 I am fine with Dan’s proposed objectives as well.
  
 Kind regards,
 Scott
  
 
From: Joel Goergen [mailto:jgoergen@xxxxxxxxx]
 Sent: Tuesday, February 14, 2012 7:55 PM
 To: Scott Kipp; STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative
 
Scott
 
 You can not specify a single source in a standard ... At least it isnt a good idea.  We don’t specify manufacturers, we define specifications.
 
 I can create a meg6 stack detail that will not meet the objective.  I can define a meg6 stack detail that has as much loss as an fr-4 material.
 
 I started this thread and I think it is going in an off direction.  My point was that we need to specify approximate length, signal rate, and basic material floor.  We could be better then that, but you would not want to include less then that as part of the solution space we specify to.  The follow statements define that and meet all the concerns of what I have read or heard.
 
 1) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz over links consistent with copper traces on “improved FR-4” (materials definition to be defined by the Task Force) with lengths up to at least 1m.
 2) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz over links consistent with copper traces on “improved FR-4” ( materials definition to be defined by the Task Force) with lengths up to at least 1m.
 
 We could remove the parenthesis and text within them.  In doing so, we specify a rate, a length range to a maximum, and that we are using some kind of laminated resin with copper traces that are better then fr-4, which is the perceived minimum.  It is my opinion that the length should be specified as a max, which it is, and the material be specified as a min, which it is.
 
 With these as the objectives, we will be able to establish a transmitter, a reciever, and a suitable channel with which we will have specifications to design, debug, and ship the combination of the three..
 
 Joel
 
 
 On 2/14/12 4:47 PM, "Scott Kipp" <skipp@xxxxxxxxxxx> wrote:
 Dave,
  
 I agree with your point and propose that this could help us distinguish the two markets.  For NRZ, there doesn’t appear to be an improved FR-4 material that meeting this requirement, but there is a Meg6 that meets the requirements of 0.89 dB/inch (35dB/39.4”) at 12.9GHz.  For PAM-4, there are some FR-4 materials that meet the requirements of 0.84dB/inch (33dB/39.4”) at 7.0GHz according to Kochuparambil_01a_0112.  
  
 This shows that there is a big difference for what we are proposing for the two PHYs.  While I don’t think that most NRZ implementations will be based on Meg6, the implementer can use the material of their choice and go for a shorter distance.
  
 How about an objective like:
 
1)      Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz over links consistent with copper traces on Meg6 with lengths up to at least 1m.
 
 2)      Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz over links consistent with copper traces on “improved FR-4” with lengths up to at least 1m.
 
 
 We could add in the dB/inch for each material in the channel definition.
  
 Kind regards,
 Scott
  
  
 
From: Chalupsky, David [mailto:david.chalupsky@xxxxxxxxx]
 Sent: Tuesday, February 14, 2012 11:58 AM
 To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: Re: [802.3_100GCU] Objective changes for the Back Plane Initiative
 
Hi Joel,
 Is there a way to differentiate the wording regarding materials for the two objectives?   I think your intent is that “improved FR-4” has a different definition for each objective, but with the same terminology on materials it is not clear why the two objectives are distinct.
  
 I am in favor of two objectives, just anticipating that a good deal of word-smithing will ensue.
 Would a compromise be to keep the loss and reach, but drop the material?
 Thx,
 Dave
  
  
 
 
From: Joel Goergen [mailto:jgoergen@xxxxxxxxx]
 Sent: Tuesday, February 14, 2012 11:12 AM
 To: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx
 Subject: [802.3_100GCU] Objective changes for the Back Plane Initiative
 
All
 There have been some discussions around changing the following objective:
 Define a 4-lane 100 Gb/s backplane PHY for operation over links consistent with copper traces on “improved FR-4” (as defined by IEEE P802.3ap or better materials to be defined by the Task Force) with lengths up to at least 1m.
 
 To something like:
 •Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz
 •Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz
 
 I don’t support this because it leaves to much variation, along with additional specifications, in the definition of the transmitter and receiver.  Systems implementation of this type of specification will see 1) an increased design cost attributed to tools, modeling, and re-spins, 2) incomplete and inconsistent models providing false positives to a successful implementation, and 3) a complex set of additional metrics that will make the finished standard too complicated to follow.
 
 The following is a much better set of objectives that allow systems vendors a path forward using current design processes and implementations.
 1) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz over links consistent with copper traces on “improved FR-4” (as defined by IEEE P802.3ap or better materials to be defined by the Task Force) with lengths up to at least 1m.
 2) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz over links consistent with copper traces on “improved FR-4” (as defined by IEEE P802.3ap or better materials to be defined by the Task Force) with lengths up to at least 1m.
 
 I heard reference that by defining loss, material, and length, we can not meet that.  I disagree.  I think we are defining a loss up to 1m on improved fr-4.  If it makes people feel better, the following is also acceptable from a system designers point of view”
 1) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 35 dB at 12.9 GHz over links consistent with copper traces on “improved FR-4” (materials definition to be defined by the Task Force) with lengths up to at least 1m.
 2) Define a 4 lane PHY for operation over backplanes with a insertion loss of <= 33 dB at 7.0 GHz over links consistent with copper traces on “improved FR-4” ( materials definition to be defined by the Task Force) with lengths up to at least 1m.
 
 I do not support a loss only model.  It is not economically feasible.
 
 Take care
 Joel Goergen

 




No virus found in this message.
 Checked by AVG - www.avg.com <http://www.avg.com>
 Version: 2012.0.1913 / Virus Database: 2112/4809 - Release Date: 02/14/12

 




No virus found in this message.
 Checked by AVG - www.avg.com <http://www.avg.com>
 Version: 2012.0.1913 / Virus Database: 2112/4809 - Release Date: 02/14/12