Hello Xiang,
Great results. A 0.1dB change in CD penalty is about right for a ~10x change in BER at these high BER levels for moderate FECs. The waterfall curve spacing doesn’t change much. Looks like FEC B has a slightly shallower BER curve.
Coincidently, I was just exchanging emails with Mark, and we commiserated that the datacom industry has eaten it’s one-time free FEC lunch. We now have to do real work to put food on the table. This goes for CD penalty as well, where the
big win was going from BER of 1e-13 to 1e-5. We may have to ask Sudeep and Vasu to bring back BCH.
😊
Chris
From: Xiang Zhou <zhoux@xxxxxxxxxx>
Sent: Saturday, August 22, 2020 6:18 PM
To: Chris Cole <chris.cole@xxxxxxxxxxx>
Cc: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Definition of "Breakout"?
I compared 3 FECs with PAM4 modulation format
1) KP4 FEC with operating BER 2e-4
2) FEC B with operating BER 1.4e-3 and overhead ~11%
3) FEC C with operating BER 4e-3 and overhead ~ 12.9%
Under one simulation condition assuming identical fiber length and modulation chirp,
CD penalty for KP4 FEC, FEC B and FEC is observed to be about 1.4dB, 1.5dB and 1.3dB, respectively.
Hello Xiang
What is the difference in operating BER between the FECs that you simulated and the CD penalty for one of them?
Chris
From: Xiang Zhou <zhoux@xxxxxxxxxx>
Sent: Friday, August 21, 2020 12:08 PM
To: Chris Cole <chris.cole@xxxxxxxxxxx>
Cc: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Definition of "Breakout"?
I have run some simulations for several higher-gain FECs, what I see is that the improvement in terms of CD penalty is quite modest.
I could share some of these results during the study group meetings.
Hello Xiang,
FEC with higher coding gain has higher operating BER, which reduces CD penalty.
MLSE for datacom optics is a very, very bad idea. We are not going to get integer reductions in power consumption if we keep going into the weeds with ever more complex DSP.
To settle this, Ilya and I have cooked up a Datacom DSP brawl for OFC 2021 Rump Session. Please come and watch the mudslinging.
Chris
From: Xiang Zhou <zhoux@xxxxxxxxxx>
Sent: Thursday, August 20, 2020 7:22 PM
To: Chris Cole <chris.cole@xxxxxxxxxxx>
Cc: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Definition of "Breakout"?
From 100Gb/s per lane to 200Gb/s per lane, fundamentally we lost 1.5dB optical link loss budget if everything scales perfectly, so the margin for
additional fiber chromatic dispersion (CD) penalty will be smaller for 200Gb/s per lane. FEC with higher coding gain is effective in reducing the SNR requirement, but is less effective in reducing CD penalty since higher Baud rate signaling is needed (the
tolerable CD is inversely proportional to the square of the Baud rate).
Modulation chirp also has strong impacts on fiber CD penalty. Longer reach (>1km for 200G per lane) could be achievable by screening EML with lower
chirp, or using more powerful DSP such as the MLSE, but such designs will have significant impact on the cost and/or power. As you mentioned, it may not be a good idea to over-engineering (with higher cost/power) to support the tail of the reach distribution.
Hi Cedric,
The dispersion effects for 2km at 200G per CWDM4 lane are the same as dispersion effects for 8km at 100G per CWDM4 lane, all other variables being the same.
802.3 decided that there was not enough margin for 10km at 100G per CWDM4 lane, so created 6km 400GbE-LR4-6 standard. The 100G MSA reviewed recent results and decided that 10km
at 100G per CWDM4 lane works just fine.
8km is between these two decisions, so pick your poison.
4x200G/800GbE-FR4, when we get to it in the future, should be defined for 1km reach. Cost is everything so why eat into the margin with higher dispersion penalty to support tail
of the reach distribution.
200G/lane optical should wait until the Cu do electrical first. When 25G/lane optical went first, we lacked the imagination to include FEC. Industry had to wait for 802.3bj to give
us KR4 FEC before getting 25G/lane optical right.
Chris
From: Cedric Lam (林
峯) <clam@xxxxxxxxxx>
Sent: Thursday, August 20, 2020 10:16 AM
To: Chris Cole <chris.cole@xxxxxxxxxxx>
Cc: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Definition of "Breakout"?
Thank Chris and John. We are in the era of abundance. Abundance of bandwidths, standards and names, which sometimes can be confusing (maybe it is just me). Thank you guys for
the education. When it comes to 800G, my colleague Xiang told me that dispersion will kill CWDM4 in the FR reach. So we may need to think about how we can be less confusing.
Hi Cedric,
FR4 (2km) has always used CWDM4, at all rates.
LAN-WDM has been used for 10km or greater reaches, i.e. LRn and ERn
Chris
From: Cedric Lam (林
峯) <000011675c2a7243-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Thursday, August 20, 2020 10:07 AM
To: STDS-802-3-NGECDC@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGECDC] Definition of "Breakout"?
Also, I was told that DR4 in IEEE is kind of like PSM4 with parallel fibers, and FR4 uses LAN-WDM. So there is no good way of discerning between PSM4 or FR4 (a WDM interface) in
IEEE naming. Here D and F means the transmission distance, R is line coding and 4 means 4 lanes which could be 4 pairs of fibers or 4 wavelengths on a single pair of fibers. So we may need to think about how to fix that in IEEE naming.
John,
I disagree. The term “breakout” is used in a general sense whenever a module can be put into use as four independent interfaces supporting ingress signals on four independent time
domains. The choice of reach does not change this understanding.
The following modules are in the market or in progress to be in the market.
QSFP+ for 4 x 10GBASE-LR
QSFP28 for 4 x 25GBASE-LR
QSFP-DD for 4 x 100GBASE-LR1
and also
QSFP-DD for 4 x 100GBASE-FR1
The QSFP-DD for 4 x 100GBASE-FR1 is also known as 400G-DR4+. It can also be used to support 400GBASE-DR4 with 500 meters of parallel cabling.
The QSFP-DD for 4 x 100GBASE-LR1 is also known as 400G-DR4++. It can also be used to support 400GBASE-DR4 with 500 meters of parallel cabling but is a less likely use case.
The relevance of breakout to the IEEE 802.3 process is when the high-capacity module supporting breakout is thought to be the prevalent implementation of the PMD for broad-market
potential instead of a single implementation of the PMD in a given form factor. In companion with this is the definition of a multi-port MDI with demarcation of which Tx and which Rx go together. These definitions are usually left for form-factor MSAs to define.
The QSFP-DD MSA defines MPO and four separate SN/MDC duplex connectors to support quad-port breakout. The QSFP-DD for 4 x 100GBASE-FR1 AKA 400G-DR4+ using SN connector is an example.
Jeff
I think there are two different concepts that end up being colloquially referred to as breakout. The first is that case you detaiil below- a set of parallel media lanes that can
be grouped in various ways- either as a single unified PHY or as a number of slower PHYs. The other case is a module that happens to hold a number of PHYs that are completely independent- like a QSFP-DD/OSFP module that has 4 x 100GBASE-LR optics.
I’m Ok with declaring that “breakout” just covers the first case and your list is a good start at scoping the definition, but if we do that, I’d like us to figure out a name for
the other case. I find that when I talk to people about modules- it is important to clearly address both cases, since many folks have only one of the cases in mind and conversations can get confusing.
I am also Ok with defining “breakout” to cover both cases, but then we can make that explicit in the definition.
--
DaveO
All,
As I explore the scope for the Beyond 400 GbE effort, I have been having a number of conversations related to “breakout”
While we all discuss it – I have never seen some actual formal definition that is agreed upon within 802.3. So I would like to get some input.
I am going to start with breakout actually does and solicit input before proposing some definition to potentially use.
I see break out of the following –
-
Backplane
-
Twin-ax cabling based on multiple different pairs
-
SR optics based on parallel MMF
o
DR optics based on parallel SMF
FR / LR / ER optics – I don’t see as being part of breakout.
Thoughts?
John
To unsubscribe from the STDS-802-3-NGECDC list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-NGECDC&A=1
To unsubscribe from the STDS-802-3-NGECDC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-NGECDC&A=1
|