Well I have a 12V with 5A power supply but the connection is loose. Those high power supplies are really rare so my other 12V supplies are only up to 2A with which the device will not boot at all.
1 Like
That is odd. I am running it through USB PD on a powered USB Switch and the output is rated at 20W. I also have USB-C cables with built-in power meters and it shows 6-7W for the R4 just idling in U-Boot / OpenWrt with one NVMe SSD connected in the M.2 slot on the bottom and an SFP Fiber module plugged, but not connected to anything.
Dunno what that was but now it seems to be running stable since I switched to a non flaky power supply.
Another thing that's weird, the MAC address of the br-wan device has a new MAC address, every time I restart the device. Any idea why that may be?
Interesting & good to hear. Thanks for the datapoint.
Totally unrelated: sometimes it's about peak transient power requirement while the stable equilibrium requirement is much lower than that.
For example, my Huawei GPON on loan from ISP has a short, transient, peak requirement of 12W during boot up. Once enters steady state, at any given time, it's less than 5W, regardless of how heavy data traffic going through.
I use USB-C PD supply which is easy to get and not that easy to loose.
What watt rating did you find appropriate for the USB-C PD adapter & cable you are using?
Do you have a link for us?
I was using the Chromebook 45W USB-C power.
1 Like
Still no case listed on the official store yet.
But the price of R4 seems to have dropped by 8% (?).
I hope to see on the official store some combo options like: aluminium case + R4 + heatsink with fan or fanless. Plus screws to secure the R4 base board to the case. Such a kit will be awesome.
3 Likes
... if you are fine with a silicon vendor who doesn't give a cent to improve drivers in upstream Linux and will leave you with outdated SDK sources (at best). Despite having 10G ports the hardware is barely able to route 10G IP traffic, let alone PPPoE or QoS, with current upstream Linux drivers. I'm not even sure if Marvell still hires the engineers who designed the CN93xx series and if who ever works there now would have the capability to improve the situation in terms of knowing how the hardware actually works.
It's a classic result of "keeping things secret and under NDA as long as possible". At some point (apparently) the vital knowledge needed to maintain drivers with all offloading features needed to actually use the hardware to do routing task at wire speed is just lost, and we will have to engage in digital archeology to find out how it was ever intended to work.
5 Likes
Hello All Banana Pi BPi-R4 users,
Just wonder are you able to chip in queries from Crowd-sourcing the Wi-Fi Chip info
This is to verify whether the latest MediaTek filogic 880 chipsets fully support the wireless bufferbloat mitigation features.
Thanks
1 Like
BPI-R4 by default doesn't come with any WiFi
I'm in the process of writing a review about the R4. If you guys have any comments on that I would be happy to read them.
5 Likes
I believe some devs might have it for testing, but now everything still in early stage so who knows there will be change or not? Just wait for the final product to come out first.
1 Like
Regarding the benchmarks you are listing it would be good to know
- which SFP+ -modules you used
- the overall consumption (+ CPU-load), when transferring data over fibre
Having a list of sellers/shops shipping to Germany later on would also be a big plus (and entry point for new users).
1 Like
This post https://forum.openwrt.org/t/a-wireguard-comparison-db/ contains Wireguard benchmark results from a couple of BPI-R4 users which might be useful to add.
In addition this post https://forum.openwrt.org/t/please-run-iperf3-c-127-0-0-1-and-post-results-here-to-make-router-database/ contains iperf3 benchmark results
1 Like
Looks pretty good so far. I'd add core type to the comparison table (CPU cores 4xA53...), doesn't the R3 have A53 (LITTLE) vs R4 with A73 (big) cores?
1 Like
Good finding about one m.2 SSD adding another 1.5-2W extra power consumption at idle.
I think there is an issue. Perhaps check in both MediaTek v21.x and OpenWrt snapshot. Either one or both do not handle m.2 NVMe idle properly.
An m.2 NMVe at active idle should be less than 100mW these days. At sleep / deep sleep should be negligible or not measurable (unless with precision instrumentation).