From 2 years ago, apparently ships with OpenVPN.
Presumably working code for this SoC is here:
GitHub - GainStrongService/openwrt at 2102/Oolite-MT7981
You will have to translate from Chinese, though.
When I built my PC-based router years ago, it died after some time due to the fan in the power supply has stopped (fortunately without fire). After this, my motto is: no more moving parts in 24/365 DIY devices.
Wow...you pay only that much?
Here in the Netherlands I paid €0,40 per kWh (December 2023)
There is nothing about Mediatek IPs for ARM SecWorld mode (like AES accels/DDR encryption/etc), bootsec activation (if it supported) and fuses like JTAG debug deactivation. We are talking about secure router. As an example, secure boot is a part of well designed device, at least it protects against booting from a hacked/dead flash. And I would prefer to have it, if, as another example, my garage door/video surveillance cam will be connected to the router.
Also, if we will developing mass production device (hopefully 100s of thousands, isn't it?) than collecting pieces of probably incorrect information from different parts of Chinese forums will not allow to create something really stable. We are needed official docos, with official support (like hardware design review). Sorry, but a router that will work from time to time is not the best demonstration of the quality of development.
2 Likes
In my latest bill, kwh is already €0,23 (almost double in the last four years). France. I'm starting to dislike anything green related.
Linux has software to monitor fan RPM. lm-sensors so your motto is based on your failing to use all tools at your disposal to properly instrument your router, it's not based on an invalid use of a PC as a router.
And plenty of other stuff dies. Such as the ubiquitous "wall wart" 12 DC adapter that powers so much of the SoC crap out there on the market. I've had plenty of Cisco switches die due to power supply failures as well. Electric power cleanliness is an uncontrolled variable and a lightning strike in the vicinity can take out any piece of network gear.
Fan was In the POWER SUPPLY. PSUs with fan control/monitor output are top models from the server's world.
We get a significant amount of power from renewables. 15% in fact comes from wind power and that is continuing to grow. A wind farm costs nothing other than maintenance costs.
I believe in the EU there's a giant community of do-gooders worried about bird strikes and aesthetic views and other such rot to make effective use of wind power so you are pretty dependent on fossil fuel to generate power. Here, we just make gigantic windmills where the blades turn so slow that bird strikes are not a problem (or maybe we just have smarter birds) and when the Nannies bitch about the view being spoiled by a big-ass windmill we tell them to go pound sand
In the state I live in electric utilities are considered monopolies and under the law, monopolies are required to submit to regulation. That regulation permits the utility to make a modest profit but they have no fiscal interest in social engineering with prices.
Interestingly, the price-per-mile for an Electric Vehicle in the US is 1/4 (25%) of the price-per-mile for a gasoline powered vehicle. Yet people still choose dino-juice vehicles over EV's which proves that even in situations where the government is NOT manipulating pricing and the cheaper price just naturally is for the "greener" solution, that manipulating prices does NOT necessarily accomplish the desired goal.
As a "car guy" I would LOVE to see Big Oil removed from transportation power in the US.
Look nobody here is really that interested in datasheets, if that's something you need you can simply find it on google, I didn't say anything in anyway that's sounded mean or aggressive, I've been doing kernel stuff for years, I do android development and openwrt development so please do not challenge me and regard me as stupid
From:
Power Supply Fan RPM - Power Supplies - Linus Tech Tips
"...I changed the fan on my Thermaltake PSU with an Arctic P14.
I routed the RPM signal cable to a fan header on my ASUS motherboard.
This was quite easy in my micro ATX setup.
Now I can read the PSU fan speed on my ASUS tool...."
You really got to come into the 21st century. PC gear has come a long long way from 1987
Arg, the entire OpenWRT project has been built on collecting pieces of information from different bits here and there to suss out manufacturers like Netgear's approaches for their various devices, so OpenWRT would run on them. You are trying to take a project that is built on this and turn it into something it's not. Something "really stable" with secure boot and the rest of it is OPNSense/PFSense which gets right back into PC Land.
Why do I get the sense what you are really arguing here is you want the pluses of a PC-based router (security, flexibility, power) and the plusses of a SoC solution (cost, power consumption, KISS) rolled into one?
Many of those plusses are mutually exclusive. You can't have power without fast CPU clock speeds and you can't have fact CPU clock speeds without generating heat and you can't effectively dissapate heat without a fan or two, and on and on it goes. And the more stuff you throw in the more expensive the hardware gets
1 Like
I'm not so old :). And it was one of my first experiments with DIY routers creation, so I just didn't thought about such thing like overheated PSU with unpredictable results for the motherboard :).
OpenWRT is a software project that has some (small) support from various enterprises, such as some French telecomms. What is good for software is not always true for hardware. Especially for mass prod. That's exactly what I'm trying to tell. Choosing an undocumented SoC with an unpredictable lifespan for the first flagship hardware project. I don't know, IMHO, it doesn't looks very well.
Most don't think about off-voltage wall-wart AC adapters with unpredictable results for the SoC-based devices. 
I've rescued MANY a "broken wifi router" by binning the wall wart and replacing it with a new one.
I am actually arguing the same as you - that a "reference router" is a bad idea but I'm not looking at it from the silicon point of view but the business point of view. If the OpenWRT devs ever do get someone to build them a purpose-built device based on MediaTek or whatever, I just feel it will serve as a giant distraction, there's no point in building the project for 1 hardware device and then claiming every other router with a problem is not worth paying attention to since it does not measure up to the reference device.
I think it's also important to examine the example of focusing on a single device has done to many router projects in the past. Tomato by Shibby for example - defunct when it's reference device became obsolete. Several forks of that went dead as well. Merlin has become almost useless in fact it spawned a "Merlin Legacy" fork because it hyper-focused on only the newest ASUS devices. Gargoyle, Chilifire, Advanced Tomato, - all of them ended up niche projects.
It's a shame that the OpenWRT devs get so het up and excited about vaporware like this proposed device yet show such little interest in fixing some of the most basic and obvious and easy issues with the regular devices. For example the Netgear R6230 - for crying out loud, how difficult is it to change 1 byte - the label - in the OpenWRT R6220 firmware and create an official R6230 build? Yes yes we can read the directions and load R6220 firmware on the R6230 we know we know. But it's sloppy.
6 Likes
I'd rather travel long distances in something more flexible than an electric only car. One day I was stuck in a 90km traffic jam on a one-way road (unfortunately such jams are common in France during holidays) when the temperature was around -5C (20F). I can imagine the scale of the disaster if all cars were electric and some of them would run out of charge (winter, cold, lithium battery, too narrow road for the service truck).
Hi @avolkov-1221 @tmittelstaedt - I would kindly like to ask you to stay on topic with your replies. Neither the discussion about the properties of electric vehicles, nor experiences with self built PC routers belong into here.
11 Likes
Sorry for the offtop.
1 Like
No, just dumber people. (You might want to learn what the actually velocity is at the tip of each blade.)
Take a walk through a wind farm anywhere along the migration flights spring and fall. A thousand birds strikes a day isn’t out of the ordinary even with a moderate sized farm.
Ever lived for a month in the middle of a wind farm? Then tell me how you feel about your health after a month of putting up with he constant low resonances emitted by these “monsters”.
Sorry @jow. Too close to home.
6 Likes
My non-expert opinion is that a standard model of router designed to run OpenWrt from the beginning is a good idea.
It will certainly have flaws. I hope that none are fatal.
But just like open-source builds on other people's efforts to improve software, I would hope that we, as a community, can learn from the first standard model to build later ones that address any flaws found.
The important thing is to have a functional device that meets the minimum requirements.
There are many "nice to have's", but I think they can wait. Things like:
A more powerful processor/processors (for throughput and firewall)
Taking full advantage of big/little processors to minimise quiescent power usage
More switch ports.
PoE
Modular, upgradeable WiFi
Interfaces to allow use as a NAS
Dual flash for easy rollback from failed upgrades, and to allow easy remote upgrades
Sufficient memory and cpu-power to run docker/snap/application images or full virtualisation.
I/O for home automation/projects
ZigBee/LoRaWAN/Bluetooth and or other local and wide-area wireless
GPON
ADSL/VDSL
10G LAN
Secure Boot
None of this is what a first effort should or 'must' include. A lot of it is common with 'developer SOCs'. People can make cases for most of it. But we need a minimum viable OpenWrt product which the community can use as a good foundation to learn from, while providing standard, supported hardware not dependant on a third-party manufacturer's foibles.
I hope this project succeeds, preferably beyond anyone's wildest dreams.
Thank you for 20 years of making networking equipment better with your awesome firmware!
Now, to the OpenWRT One device:
- SOC: MediaTek MT7981B
Results in poor IO.
- Wi-Fi: MediaTek MT7976C (2x2 2.4 GHz + 3x3/2x2 + zero-wait DFS 5Ghz)
We have 802.11be already here, expensive 802.11ax device is just too late to the party unless it provides decent upgradability - and OpenWRT One doesn't.
- Ethernet: 2x RJ45 (2.5 GbE + 1 GbE)
Too little & too slow.
- USB (host): USB 2.0 (Type-A port)
Just bad compared to any similar device.
- Storage: M.2 2042 for NVMe SSD (PCIe gen 2 x1)
You'd better make it miniPCIe, and absolutely make sure it can provide at least 10W - NVME 2242 SSD will fit just fine via a dirty chear adapter, but also some AP-grade WLAN card or PCIe modem.
- Power: USB-PD 12-25V on USB-C port
Better use dumb 12V ~ 24V instead of PD crap.
- Price: aiming for below 100$
BPi R4 has overwhelmingly better IO, OpenWRT One compares to it very unfavourably with nothing but built-in WLAN going for it. Then again, BPi R3 also has built-in WLAN, and OpenWRT One looks inferior hardware-wise even to that. Unless it is MUCH cheaper than BPI R4 & R3, OpenWRT One will be very bad value.
With all due respect, improving support for already available and by far superior hardware-wise devices would likely be preferential. OpenWRT One doesn't look enticing for enthusiasts, with its mediocre specs and old technology - and I highly doubt it will be able to compete with similarly-specced consumer routers for those who mostly want their devices cheap, but would prefer OpenWRT over stock firmware. And prefential treatment of OpenWRT One by developers will set a bad precedent, disappointing many if not most users.
5 Likes