OpenWRT Best Hardware Today

I have been running OpenWRT in my router since 2012 and after 12 years of great home service I am thinking in retire it to a testing environment.

Searching for the ideal hardware these days found that it's very hard to make up your mind using the TOH. Old and new hardware is mixed (yes, they have a release date), most of the specs to filter, but not their problems or recommendations. Its like choosing a tomato can in Google, infinite choices, when I would better like to go to Costco (a decent quality place I thrust) and chose between the 2 options... :slight_smile:

Of course we have many requirements out here, like cheapest, best for $100, wifi coverage, media and gamming, wired routing performance, etc. but why not keeping a Yearly or Quarterly decent recommendation (maybe 2 or 3 options) on those segments pinned on this category to make it easier?

I'll keep looking for the best option, just wanted to open the discussion on the method here :slight_smile:

1 Like

OpenWrt One, a likely candidate.

2 Likes
  1. when looking for an all-in-one router that covers many use cases look for GLiNet GL-MT6000 (on sale in store-eu.gl-inet.com for eu99)
  2. when looking for a tailored advice follow https://forum.openwrt.org/t/about-the-hardware-questions-and-recommendations-category/
  3. creating- and maintaining recommended options for a specific usecase/category is a time consuming task - see i.e. https://forum.openwrt.org/t/best-newcomer-routers-2024/
  4. defining usecases/categories for a best-buy guide in itself would need to include lots of variables like
  • bleeding edge or stable
  • price category
  • home layout / floorplan
  • additional needed services (vpn, wireguard, sqm, adblock)
  • all in one or (expandable) modular setup
  • formfactor
  • userbase
  • upstream support
  • one-step easy install or multi-step expert install
  • speed
  • cpu / memory

That having said: I did make a guide with some Buyer Tips myself which I update from time to time - do note though that this is based on personal preference
https://forum.openwrt.org/t/an-openwrt-beginners-guide/

4 Likes

Honestly I wouldn't necessarily look at hardware for OpenWRT specifically. Instead find options that you like and then filter based on OpenWRT support.

Here are some questions to help guide you:

  • What is your price point?
  • What Wireless standard do you want? You probably should look at either Wifi 5 or 6.
  • Do you want new or used?
  • How much ram do you need? (If you are running extra services)
  • What throughput are you looking for?

Also, I personally would suggest looking at a base router, switch and some access points instead of just one device. If you do not want to hard wire you can look at mesh. Also make sure to avoid Broadcom devices like the plague.

1 Like

Thank you very much for this. I just wish I had come across the linked post a couple of months ago when I started my OpenWRT journey.
Does OpenWrt have "pinned topics" at all as this would have helped me a lot as a newbee but there is only so much reading that one can do......?

They seem to run a a bit of a shady shop, it sums up to at least 119€ for Germenay, once you reach the checkout step. They are advertising prices without sales tax. Still cheap, but not ridiculous cheap anymore.

Just use THANKS10 as discount code to reduce the price by 10 Euro

That having said: I did make a guide with some Buyer Tips myself which I update from time to time - do note though that this is based on personal preference
https://forum.openwrt.org/t/an-openwrt-beginners-guide/

I don't know how I missed this guide, but ed8, that was exactly what I meant... We should have your post in a pinned thread, for those of us who need it every few years, it's just perfect!

1 Like

Thanks Darin755, couldn't agree more, that is exactly my reasoning.

The reason for this post is for those of us who every other year are looking for a router for home/work/family/friend (with different goals), it's not simple to find a fairly updated unbiased recommendation.

But ed8 post "https://forum.openwrt.org/t/an-openwrt-beginners-guide" from last Oct/2023 it's exactly that. But it's not only for beginners, it's an all times excellent reference to start from.

Mhhhhh...

That board (MT7981B) was selected last Dec/2023, and it was a good board at that time...

Unfortunately minutes later, in Apr/2024, the BPi-4 (MT7988A) with a processor slightly better, faster connection and better upgradeable, was out there, making the selected alternative less desirable...

I don't want to discourage the OpenWRT One team, just seems that these days a hardware initiative needs a dedicated team leading instead of a community. Maybe a more extensive partnership with BPi would work... I miss those times of the WRT54G :slight_smile:

1 Like

You can't think in this way.
At the time of Dec 2023, MT7981B was officially supported by OpenWrt (at least we have on market router with this chip being supported), there were better MT7986A router started to get porting to OpenWrt, so I believe these 2 were the best candidate during the time of selection.

In fact MT7988A (BPI-R4) was out at the same time (definitely not Apr 2024, since I got my first batch of BPI-R4 in Nov 2023) but without any Linux mainline kernel support, and until now we are only seeing snapshot built with many issues/features pending to work on, which is far from the status of being fully supported, so it's not that possible to have it in OpenWrt One, if OpenWrt One being delayed by a year the situation would be pretty much the same (but I guess we can choose MT7986A?)

This is non-commercial project, so it's hard to have dedicated team on it.

I really like the Nanopi range. I started on the R2S then upgraded to the R2S plus and am pondering a R5S. Years of service, oodles of storage and RAM with a processor that does everything i want (router and bandwidth monitoring but no SQM). I did a survey of cheap but good routers recently and it came out well.

My advice is look at my profile and find the post then see the comments. Then wait till black friday and buy one at a discount (although check on 11.11 too as China does good discounts then too - its called singles day and big in Asia i believe).

R2S is good, I also have it, on top of it I also bought R4S & R6S. I skipped R5 series because of SoC is in fact weaker than R4S, a direct jump to R6S can give a lot more horse power.

FWIW, I think if you use it as wired router, R4 is perfectly fine on snapshot. I know mine is (and without any competition in it's segment, either way). Is it totally perfect? Well, I am less than happy about the netifd upgrade in August nuking out of the box support for my fiber optic but eventually found a perfectly stable work around for that, too. I know some people complain about of sequence UDP but it has zero impact on me...

I have not tried the BE14 (and given the layout of my apartment, I probably won't) so can't comment on wifi.

In terms of performance the "best" hardware is going to be a modern x86 box running OpenWRT coupled with a dedicated business-grade WAP such as Ubiquiti UniFi or TP-Link Omada. At the expense of slightly higher power consumption, this is going to blow the doors off any all-in-one router.

3 Likes

I'm not sure if this is snapshot issue or not, but my SFP+ to RJ45 modules not being recognized so I can't use it on my network, and I saw that the hardware acceleration is not there yet? It would be a problem if you need to run at 10G

But I'm looking forward to more recent builds.

Hardware NAT acceleration works just fine at 10G.

As for the SFP+ RJ45 module, what does ethtool show when you plug it in?

looks like a very good hardware

I tried telling the team repeatedly that no one will pay for a lesser device just because of OpenWRT branding, especially the users here that are willing to reflash stock hardware from China if it saves a few bucks.

After asking the community for feedback on the One, it was all resoundly rejected because the plans had already been drawn up and there was no budging. It left a sour taste in my mouth and here we are, a year later with no One and an even more outdated hardware "plan".

What OpenWRT needs is a REFERENCE 802.11be device. KISS and rebrand a BPI-4 with whatever WIFI module the team deems worthy. I don't understand the disconnect here.

1 Like

Those who,

  • came up with the idea
  • did the negotiations with the ODM
  • did the hardware design
  • pushed the software development
  • and footed the bills

made the decision what they were going to spend their time and money on, surprising, isn't it?

Your only decision now, is if you are convinced and want to spend your money on it - or not and pass along. It's as simple as that - and no one will be mad if you do choose something else (quietly). Complaints that your ideas and the visions of those who actually did the work do not align, won't get you any further.

A few suggestions by the community were taken into account, at a point in time when the basic goal posts were already long set and a preliminary PCB design already available, changing the SOC never was on the table - and one of the stated project goals was keeping it roughly in the 100 USD bracket from the beginning (to keep it affordable <-- and we have seen previous examples(pl.), elsewhere, that did break the bank, for decent hardware, which still didn't age that well - as well as total failures that never delivered anything).

Now let's take a look at the BPi-r4[*] for a second, pricing for a kit (incl. case & PSU) including wifi7 card (BPi-r4-nic-be14, and these have only become available over the last few weeks) seems to be somewhere around 480 EUR (that is considering the ODM's own product, without the markup they'd need for a custom/ branded low volume design), that's pretty much five times as much...! Are you sure there'd be that many OpenWrt enhusiasts able and willing to afford that much? Where would you draw the line, 100 EUR, 150 EUR, 200 EUR, 300 EUR…? Beyond that, mt7995av support in mt76 is not quite where it would need to be for a successful product launch, yet - so going that route you would delay the product launch on bet on the future, on others doing the driver development for you, to make your product viable. How great that strategy tends to run out has been shown in quite a few other crowdsourced 'opensource' projects in recent years.

Let's look at the facts here, this project delivered the first batch of 150 devices (which sold out quickly, without real advertising). The next (larger) batch is in production and apparently expected to ship sometime around late October/ early November. No one was scammed out of their money with promises for castles in the air - insofar I would call this a success. If user demand allows a steady supply of further batches for the next 1-2 years (before the hardware becomes boring) remains to be seen - and the question if it might get a successor in the future is unclear as well.

Disclaimer: I am not involved in the OpenWrt One in any way, shape or form, nor do I have any information or insight into it beyond the public information gathered here and on the public mailing lists - nor do I own this device (or any filogic devices so far), nor do I expect buying it either, as it doesn't meet my personal desires either. But that's besides the point, it's not my place to complain about it either, only to decide what I do with my own money. I have no idea if the hardware works reliably, according to the stated goals and considering the capabilities of the underlying SOC, but the early feedback doesn't suggest otherwise (and even those who did brick theirs, while experimenting with flashing things incorrectly, apparently succeeded to debrick them again, so the last resort failsafe guards appear to work).
The opinions stated in this post are mine alone and not representing the project or anyone else.

--
[*] and the presence of the OpenWrt One does not affect the availability of the BPi-r4, nor its support by OpenWrt in any way (surprise, the BPi-r4 already is supported by OpenWrt, mt7995av support also seems to be progressing). There's space for 500 EUR high-end devices and ~100 EUR entry level devices on the market.

5 Likes