Re: [802.3_ISAAC] Invitation to the 802.3dm comparison study

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

I'd like to clarify a few points.

First, regarding the duplexing method, you might be thinking of the technology known as GMSL1, which does exhibit limited frequency overlap between high-speed and low-speed directions. In contrast, the technology known as GMSL3/2 does not have the limited overlap between the low speed and high-speed directions that GMSL1 has.

As illustrated in the attached plot (see notes below), the technology known as GMSL2 operating at 3Gbps demonstrates has, in fact, complete overlap between high-speed and low-speed directions. The 9b10b encoding of GMSL2 does provide low frequency shaping of the PSD but there is still complete overlap. So, both GMSL2 and 64B65B encoded technology known as ACT+GMSLE utilize the same duplexing method although they have different scrambling and line coding schemes.

Although it does not preclude other approaches to handle the high speed and low speed overlap in ACT+GMSLE, the same echo subtraction approach utilized in GMSL3/2 may be used. While the 9b10b encoding of GMSL2 allows a high-pass filter to be applied at the high-speed receiver to suppress low frequency echo residue from the low-speed transmitter without introducing baseline wander, there are several techniques which could be employed with ACT+GMSLE including high-pass filtering with a baseline wander compensation circuit.

Secondly, concerning reuse between ACT+GMSLE and GMSL3/2, I have thoroughly reviewed the ACT+GMSLE proposal for IEEE 802.3dm and discussed it with designers affiliated with my company. We have determined that there is significant IP reuse between parts of the PMA for GMSL3/2 and ACT+GMSLE. In addition to the same echo subtractor being applicable to both ACT+GMSLE as well as GMSL3/2, the Baud rate in the high-speed direction is very similar, and the same line drivers and high-speed mixed signal receiver (CTLE+DFE) can be utilized as-is. A CDR can support both rates. While other receiver approaches may be employed, there is sufficient commonality between ACT+GMSLE and GMSL3/2 to enable the use of a single transceiver IP for both with minimal additional complexity. The primary difference in a common high-speed receiver is, in my opinion, the baseline wander compensation circuit which may be required if echo subtraction duplexing for ACT+GMSLE mentioned earlier is used. The ACT+GMSLE 117.1875Mbps DME upstream differs from the 187.5Mbps NRZ low speed in GMSL3/2, however as I demonstrated in my contribution, the ACT+GMSLE low speed receiver can be of very low complexity and exceed the immunity levels presented in Neven and Mehmet's contribution.

The PCS digital logic and the MAC interfaces differ between 802.3dm and incumbent FDD solutions. This is not a surprise as one is Ethernet, and one is not. However, the intent, particularly with the reduced complexity of the combined and simplified ACT+GMSLE, is that many mixed signal blocks of the PMA and PMD can be reused between GMSL3/2 and ACT+GMSLE.

Rather than focusing on backward compatibility, my intention is to help define the standard to support "forward compatibility," enabling the same IP to be used with both an FDD CSI-2 based SerDes today and an ACT+GMSLE Ethernet MAC interface PHY in the future.

Regarding your statement, "If someone wants to implement a dual port ACT/GMSL PHY, it would be two PHYs next to each other." Yes, of course, a dual port PHY would require two PHYs next to each other. However, if instead you were referring to a dual-mode PHY, as I explained above, it is possible to develop IP blocks in the PMA which would support both approaches or even build a monolithic PHY IP PMA and PMD supporting both if someone wanted to.

I and my co-author contributed GMSLE to 802.3dm with the hope of converging the two separate paths of the Task Force towards a common objective. I and my GMSLE co-author have worked with individuals affiliated with ACT to incorporate several ideas from GMSLE into the merged proposal which reduces the implementation complexity. In my opinion, the term 'ACT+GMSLE' is justified.

I welcome comments on the reflector or at Ad-Hocs, IEEE Interims, and Plenaries regarding my contributions. For instance, several individuals have provided constructive technical feedback on my DME presentations, leading to improved analysis, simulation, and meaurement, which I intend to present in a future meeting. I am also prepared to review others' contributions, comparisons, and comments on the reflector and at IEEE meetings.

Cordially,
Joseph L. "Jay" Cordaro
Affiliation: Analog Devices

the attached picture shows in log frequency scale the output PSD of both sides of the link with the technology known as GMSL2 in the forward, high speed, 3GBaud NRZ and reverse, low speed, 187.5Mbps NRZ directions overlaid on the same graph. The PHY was operating in single-ended coax mode. The PHY is transmitting random data and is not in a test mode. I have overlaid the one-sided single-ended PSD of the proposed technology known as ACT+GMSLE in the downstream and upstream directions with the transmit levels, 64B65B encoding ,and Baud rates proposed. Random, scrambled data was generated.

You end your email with “so thank you for clarifying that you are open to changing your mind and support ACT/GMSLE if presented with good arguments”.

Again, if you are convinced that ACT withstands the direct comparison to the TDD proposal, support the comparison effort with constructive input.

I never ever responded to Amir’s question with “No, I would not change my mind whatsoever”, because that is against my beliefs. I might have said “No, this is not the point (meaning it is the wrong question), the comparison is not about me.” It is your misinterpretation that that means, I would not change my mind. I would kindly asked you to stop spreading such personal allegations on the reflector. I find that disrespectful.

Hi Kirsten, First, I look forward to getting answers to the questions in my email from Thursday. I would also like to encourage you and Gumersindo to update your presentation for New Orleans to match what was presented at the meeting. Gumersindo

First, I look forward to getting answers to the questions in my email from Thursday. I would also like to encourage you and Gumersindo to update your presentation for New Orleans to match what was presented at the meeting. Gumersindo made important updates to the presentation to clarify that the intent with this work was not to bypass the 802.3dm Task Force.

Regarding your question about what I meant when stating “During the discussion on May 1st at least one supporter of this document stated clearly that the outcome of this document would not change their mind about TDD vs ACT”, I was referring to your answer to Amir’s question if you would be willing to change your mind about TDD and ACT/GMSLE based on the proposed effort. Your answer to this question was “NO” and you followed up with the question to Amir if he would change his mind. At this point, the chair stopped the discussion, but I asked Amir later and he told me that he is absolutely open to changing his mind if presented with convincing arguments (and that he also communicated this to you in a private email). I was not the only one taken back by your clear “NO” answer to Amir’s question, so thank you for clarifying that you are open to changing your mind and support ACT/GMSLE if presented with good arguments.

Dear Ragnar, in your email you are saying “During the discussion on May 1st at least one supporter of this document stated clearly that the outcome of this document would not change their mind about TDD vs ACT”. Who do you mean? There are only

in your email you are saying “During the discussion on May 1^st at least one supporter of this document stated clearly that the outcome of this document would not change their mind about TDD vs ACT”. Who do you mean? There are only two supporters/authors on the document. Both Gumersindo and I can change our minds in the presence of good arguments. Any other interpretation of what has been said is a misunderstanding/misinterpretation.

Furthermore you write “I also heard at least one supporter of this document questioning if this document would change the mind of other individuals in the Task Force.” We sincerely apologize if more listeners had the same mishearing as you did. Part of what we believe makes a good engineer is the capability to change his/her mind in the presence of good arguments and we believe that the dm group consists of many very capable engineers. What would be the sense of striving for a comparison document otherwise?

Hi Gumersindo, Do I understand correctly that you are giving people 2-3 working days to send in “criteria deemed important”? Is this a hard cut-off date? I see that you talk about discussing “criteria fulfillment”. Can you please elaborate on

Do I understand correctly that you are giving people 2-3 working days to send in “criteria deemed important”? Is this a hard cut-off date?

I see that you talk about discussing “criteria fulfillment”. Can you please elaborate on what this discussion will output?

During the discussion in New Orleans, you clarified that this work would result in input into the Task Force and was not an attempt to circumvent the Task Force. However, I see that your presentation on the New Orleans meeting page has not been updated to reflect what you presented: https://www.ieee802.org/3/dm/public/0525/index.html. Can you please clarify who will be the target audience of this document? If it is the Task Force, do you believe that the Task Force lacks clarity on what it is trying to achieve?

How will disagreement be handled in in the discussion of this document? What kind of majority will be needed to put something into the document? Who makes the final call on what is in the output document?

In your list on slide 7 of the New Orleans presentation I see things like “Bi-directional use of ports”. In my mind it is obvious that we will have traffic in both directions on the link. Are you suggesting there that there should be a link that only goes in one direction, without any data flow in the other direction? Would you agree that this would be out of scope for the project?

Talking about things that are out of scope, in the Study Group there was majority support for including data rates above 10Gbps, but not the necessary 75% support for it. Would considerations about extending the data rates above 10Gbps be in scope for your document?

Like I said in the New Orleans meeting, I am not sure what the value of this document will be for the Task Force. I worry that this will be a distraction for the Task Force without delivering any tangible benefits. During the discussion on May 1^st at least one supporter of this document stated clearly that the outcome of this document would not change their mind about TDD vs ACT. I also heard at least one supporter of this document questioning if this document would change the mind of other individuals in the Task Force. If even the proponents of this document are saying that this document is not likely to change their mind, would it not make more sense for the Task Force to focus on its objectives?

Dear 802. 3dm participants, As announced on both the May 1st and May 16th 802. 3dm interim meetings, here are the details for the next steps on our joint comparison study. In the first step, we would like to agree on a list of criteria to address

As announced on both the May 1^st and May 16^th 802.3dm interim meetings, here are the details for the next steps on our joint comparison study.

If you would like to participate in this meeting, please reply to either me or Kirsten (Kirsten.matheus@xxxxxx) and you will receive a dedicated meeting invite (Teams-Link) or just join with the following link at the dedicated time:

You are welcome to send your input on the comparison items also if you are not able to join the meeting.

Systemfunktionen, Halbleiter, Vernetzungstechnologien
Max-Diamandstr. 5
80937 München

----------------------------------------------------------------
Bayerische Motoren Werke Aktiengesellschaft
Vorstand/Board of Management: Oliver Zipse (Vorsitzender/Chairman),
Jochen Goller, Ilka Horstmeier, Walter Mertl, Milan Nedeljković, Joachim Post, Frank Weber
Vorsitzender des Aufsichtsrats/Chairman of the Supervisory Board: Norbert Reithofer
Sitz und Registergericht/Domicile and Court of Registry: München HRB 42243
----------------------------------------------------------------

To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1