I like to request the enablement of "Broadcom 2.5G" and NOT to be confused with MultiGig via Aquantia or the likes.
Obviously, this is a niche request to support the very few SFP modules that sync at such speeds, but they do exist in the market, such as: https://www.sourcephotonics.com/product/sps-34-24t-hp-tdfo/ or any powered by an Intel® FALCTM ONT GPONSFP PEF98035 Chipset PEF98035ETV13 S LLC5.
Such modules do work when the interface is configured with: speed-duplex 1000-full as auto configuration fails but are not running at their default data throughput of 2.5Gbps.
As you're aware the ICX 7250 8 SFP+ ports are divided by 2 WarpCore's, 4 SerDes pairs per core, and us mere end users can observe as much by running the simple command: phy 1/2/1.
And from my past encounters with the WarpCore, it's feasible to run native SerDes 8b/10b 3.125Gbps to achieve a data throughput of 2.5Gbps and most definitely so when the right bits are set, eg.
https://community.broadcom.com/communities/community-home/digestviewer/viewthread?MID=698164
https://www.dslreports.com/forum/r32230041-Internet-Bypassing-the-HH3K-up-to-2-5Gbps-using-a-BCM57810S-NIC
https://www.dslreports.com/forum/r32393899-Internet-Bypassing-the-HH3K-up-to-2-5Gbps-using-a-BCM57711-NIC
Hopefully this request doesn't fall on deaf ears and speed-duplex 2500-full works in a future firmware release.
Thanks.
Obviously, this is a niche request to support the very few SFP modules that sync at such speeds, but they do exist in the market, such as: https://www.sourcephotonics.com/product/sps-34-24t-hp-tdfo/ or any powered by an Intel® FALCTM ONT GPONSFP PEF98035 Chipset PEF98035ETV13 S LLC5.
Such modules do work when the interface is configured with: speed-duplex 1000-full as auto configuration fails but are not running at their default data throughput of 2.5Gbps.
As you're aware the ICX 7250 8 SFP+ ports are divided by 2 WarpCore's, 4 SerDes pairs per core, and us mere end users can observe as much by running the simple command: phy 1/2/1.
And from my past encounters with the WarpCore, it's feasible to run native SerDes 8b/10b 3.125Gbps to achieve a data throughput of 2.5Gbps and most definitely so when the right bits are set, eg.
https://community.broadcom.com/communities/community-home/digestviewer/viewthread?MID=698164
https://www.dslreports.com/forum/r32230041-Internet-Bypassing-the-HH3K-up-to-2-5Gbps-using-a-BCM57810S-NIC
https://www.dslreports.com/forum/r32393899-Internet-Bypassing-the-HH3K-up-to-2-5Gbps-using-a-BCM57711-NIC
Hopefully this request doesn't fall on deaf ears and speed-duplex 2500-full works in a future firmware release.
Thanks.
up-n-atom
The modules are not much different from an Access Point. They're more than just a conduit like typical optics since they have an embedded SoC with OS and ethernet layer for triple play VLANS, but at the same time are limited, which is where an L3 switch can expand on. In the case of the linked module, it's running OpenWRT linux and provides all the conveniences. The capacity is tied end to end.
This is how the LAN/SFP end of the module can be configured.
and the result of SSH into the module and running: onu lan_port_cfg_get 0
As you can see the LAN/SFP end is set to Ten-bit Interface AUTO, speed is set to 2500, duplex is set to full, etc.
Ideally, if you were to SSH and run: onu lan_port_status_get 0, you'd like to see the following result
errorcode=0 pport=0 mode=15 enable=1 link_status=5 phy_duplex=1
rather than
errorcode=0 pport=0 mode=15 enable=1 link_status=4 phy_duplex=1
The 1st is achieved using a BCM57810S Adapter with a 2.5G driver patch (It uses a WarpCore as well) and the 2nd is from an ICX 7250. Where link_status is
Simon, Employee
Thanks for the suggestion, let us know if you have any other ideas.
up-n-atom
Thanks for the consideration and time.