[Discuss for BPI-R4] Batch production version: 2x10G SFP or 1x10G SFP + 1x 2.5G ethreal port?

We now plan to mass production, which hardware version is your choose? Batch production version: 2x10G SFP or 1x10G SFP + 1x 2.5G ethreal port ???

2x10G SFP version:


1x10G SFP + 1x 2.5G ethreal port with PoE support version:


2.5Ghz seems more interesting to try more things on the same board.

1 Like

2x10G SFP, otherwise it will be useless for 10G wired internet

My vote is for:

I agree with graphine.

2 x 10g sfp version is better i think.

I think, but this also depends on the pricing. because you will need an extra sfp module

[edit]: this one should’t be a reply

Is any release date and price :slight_smile: ?

We are doing small batch production now , when we finished EVT , will public sale , maybe at October

Note: wifi 7 module not include , wifi 7 module all test ok , but public sale will at Q1 2024

1 Like

:+1:. I have one Question :
will be a serial number stored into the R4?

2x10G SFP is better.

Is phy chip for 2.5G installed on vertical PoE module? Is it PSE or PD?


SFP jack it is just connector and metal case. For PoE need add RJ45 with PoE transformer, power supply, PD/PSE controller and PHY chip. The BOM for PoE board cost more.

2x10G SFP and also fix the kernel bug :wink:

?? which kernelbug do you talk about?

of course the dual sfp version. but why it takes so long.:dizzy_face:

1 Like

No, the MT7988 SoC comes with a built-in 2.5G PHY so no additional PHY IC is needed. The vertical module is for supplying the board with power via PoE.

dual sfp will probably be more future proof – even though I plan to test it through different configuration (some old configuration that might not support sfp)… I can probably just add a rj45 port by converting one of the sfp port or from the usb port

The kernel crashes we already see on the R3 version, which is still causing unstability with OpenWRT: https://github.com/openwrt/openwrt/issues/12143

I bet the R4 will have the same issue.

This one is hard to nail down…before fixing we need to know why it crashes. A crash happing after some time is unable to trace. We need a way to reproduce the issue within minutes to exclude specific parts (e.g. hw acceleration),adding debugs etc.

A crash after days/weeks can also be caused by environment (power instability, temperature, external frequencies).

I havn’t seen this before,but atm my r3 is currently only running for max an hour and without real workload.

Why there is no RJ45 2.5 GB ports?

The vendor driver adds various watchdogs triggering hardware and driver resets, so the hardware can just “get stuck” or reset for external reasons. Afaik some of that is handled also in the driver in vanilla Linux, but it can be that something is still missing.