Why do you bother with these overpriced routers?

Slightly exaggerated perhaps but seriously for a Celeron even at full gigabit routing it's probably got one core fully clocked and 4 cores idle. So I still call that basically idle. And 1 full hour a year like that would be .9998 of the time fully idle and .0002 of the time mostly idle.

I do have a kill-a-watt so maybe I'll measure one of my older celerons and shove it into the power thread

Much appreciated! Again, I am not wanting to doubt your sentiment, just that seeing hard data makes it much easier to compare things.

True, but isn't that the case for all X86 routers? I'd roll an external AP anyway.

I just think it would be a nifty target to experiment on.

Yes, but for your specific use case of being a travel router, semi-decent on-device wifi would be useful (as that would be the primary uplink in most cases and also required to feed phone/ notebook as AP concurrently). A weaker CPU is easily offset by good/ reliable wlan (which neither rtl8723bs, nor brcmfmac, nor mwifiex, would be).

Still nice to experiment with.

I find it interesting that the discussion goes from x86 to energy prices, but nobody ask what the OP means with "overpriced routers". I recently bought a TP-Link Archer C2600 (dual core ARM IPQ8064 and all the NSS offload cores as discussed in other threads and very good ATH10 radios) for "only" 20 euros. Technically secondhand, but with all the original plastic foil still attached to the device and antennas, original box etc. I don't consider that "overpriced" at all and for my 350/35 cable connection it is more than sufficient.

1 Like

Good point. For a long time you could buy the RT3200 at £50. What's the issue there, and how would I beat that with Heath-Robinson-esque RPi4 or PC chassis with wires sticking out and five separate parts, etc.? Nah, everything in one nice-looking box that just works thank you. Job's a good-un (for me since I only have a sub 100Mbit/s connection).

Also, forget Greta and energy bills, what about WAF? That's super important too. And then not to mention the GF 'geek factor'. I mean do we want our homes to look like the lab in a Back to the Future set or what?

But @slh's summary above seems well framed. With a fast enough connection you may need a custom solution. There are nice looking and small devices with e.g. 4x LAN ports that I have seen @frollic recommend based on eBay listings and what I really like about that is that it cuts down on electronic waste, which I think is also worth putting into our mix since it seems pretty horrific what gets thrown away and how that is handled.

1 Like

I actually totally want a wall sized cabinet of blinking lights, switches, and dials and cranks...


So I posted some measurements of my slightly older Celeron box: Power consumption of routers and associated networking equipment [collection] - #18 by dlakelan

Short story is, it uses 4-5 watts sitting idle and 7 watts running a speedtest in firefox in a gnome desktop environment with a monitor plugged in (not including monitors power itself.)

It depends on the mainboard, PSU and additional (including onboard ones!) components (e.g. dedicated graphics chips would immediately cause a +(30-70) watts delta, to the idle figures). A very minimal mainboard can even leave out the traditional southbridge these days (at the expense of PCIe lanes and other I/O to distribute) and treat the x86_64 CPU as SOC.

Just to provide some data for similar systems, both baytrail-d j1900 Atoms:

  • Rohde & Schwarz 'gateprotect' FW-7543B-GP1
    • j1900
    • 4 GB DDR3 RAM
    • 30 GB SATA SSD
    • 4* I211 (igb)
    • external 12V/ 3A PSU
    • 11 watts (semi-)idle (normal background operations), while being used as router
  • ASRock Q1900DC-ITX (this consumer board offers more fluff than the made-for-purpose gateprotect, e.g. an additional onboard SATA controller)
    • j1900
    • 2*4 GB or DDR3L/ SO-DIMM
    • 500 GB Samsung HM500JI 2.5" HDD ("spinning rust")
    • 1* RTL8168g/8111g (XID 4c0, r8169)
    • external 19V/ 3A notebook PSU
    • 6-7 watts (semi-)idle (normal background operations), while being used as LAN server with several services running

Both measured with a semi-decent (consumer) power meter (not a simple smartplug), taking the power factor into account and measured at the 230 VAC/ 50 Hz plug for the total system (excluding the monitor power usage, although a monitor and USB keyboard/ mouse are connected to the ASRock). I have measured both systems for a day/ a week each, average power requirements approach the measured idle figures for longer periods of time. Artificially inflated system usage (concurrent multi-threaded kernel build while also doing an ffmpeg/ x264 encode) can bump the power requirements to ~22 watts (gateprotect) to 28-29 watts (ASRock), but the idle figure remain a good approximation for real-world usage scenarios over time.

(and yes, the gateprotect had to do fast routing with sqm/cake at 400/200 MBit/s during that time, something my ipq8065 based router chugging 15 watts idle can just barely cope with while doing plain routing without sqm).

1 Like

Many WIFI routers now come with 2.5Gbit ports which is $$$ on x86 platform and additional 3-6 watts.

15W of power consumption of WIFI router is likely with WIFI on, maybe even multiple Ethernet ports active (average power consumption is 1W per gigabit). Also ipq8065 is rather old, made on 28nm process node. Newer ipq807x are made on 14nm and their power consumption is likely >25% less.

Many WIFI routers including ipq806x and ipq807x could be even more efficient if hardware acceleration was supported under OpenWRT.

Personally I run a Proxmox VM on a 45W @idle i5-6500 Kabby Lake CPU, single 10Gbit nic, with 6VMs, including OPNsense and OpenWRT. I am considering on moving routing, home assistant and DNS to ipq8072A based router so I can add GPU for machine learning and gaming to VM server w/o worrying about idle power consumption.

Correct, but without any real traffic happening, apart from idle background chatter.

Correct, albeit in both cases - 4+1 ports in case of the nbg6817/ ipq8065, 3+1 ports in case of the gateprotect, 1 port in case of the ASRock, everything the devices have to offer (link up, idle background chatter, but not trying to push the device).

So is baytrail-d, launched Q4'13 based on a 22 nm process, the comparision isn't that far off (ipq8064 was certified by the FCC in mid 2014, as a revised spin-off from the APQ8064 smartphone SOC, which was released in mid 2012).

Not if they have 5GBASE-T or 10GBASE-T ports, far from it - the simple ipq8071a based xiaomi ax3600, yes (~6.5 watts, but that bails out at routing ~600 MBit/s, without SQM - while the j1900 can easily route 1 GBit/s at wirespeed without waking up and with sqm/cake up to ~830 MBit/s).

EDIT: the Q2'08 vintage Atom N270 (1* 1.60 GHz + HT) on a 45 nm process also manages to route ~600 MBit/s (without sqm), at ~28-29 watts idle (Intel D945GSEJT, 12V/ 5A Notebook PSU, 1* r8168 NIC, 2 GB DDR2, 500 GB 2.5" spinning rust).

Maybe, but that would only be relevant at full load - not for my idle- or long term measurements, the ARM cores (2* KRAIT300/ cortex A15 @1.7 GHz in case of ipq8065, 4* cortex a53 in case of ipq807x (ipq8071a: 4* factory overclocked to 1.4 GHz, ipq8072a+: 4* 2.2 GHz; 4* 2.0 GHz (2.42 GHz burst) in case of the j1900) clock down to the minimum for those tests anyways (and remain there for most of the day).

Not really "$$$" anymore. 2.5Gb NICs are very cheap, even pretty good ones. Like $10-20 cheap.

Even more, there are compact devices available with multiple 2.5Gb NICs for a price lower than those of routers with 2.5Gb ports.

Power consumption is another matter though...

  • i225v3 1.9 w TDP
  • i226 1.3 w TDP

power comsumption vid [about 10:10 mark] of N5105 w I225 nics w OpenWrt, so the I226 expected to be less. Not overwhelming number given the horsepower.


WNDR3700 still suits my needs

  • 20mb/s on wifi or wired
  • very very stable
  • low power consumption ~7w
  • no weird issues like with some less well supported routers like wifi calling maybe not working

the biggest reason i didn't set up an x86 office pc for openwrt is that at the end of the day, router (not expensive or fancy....) does the job very well, adding a pc means i still need something to handle the wifi connection, so instead i consume ~30-40w total to use wndr3700 as an AP...

maybe if i needed a high performance wired only router, and not use wifi at all (i considered it), i could do x86 pc with some gigabit NIC cards

You most definitely can operate an x86_64 router in that power range.

Two examples in this TDP range, both with 4 Gbe ports:

That's however only the CPU, and TDP isn't a good way of measuring power usage.


Maybe so, but it uses a lot less power than a desktop motherboard and SKU. Just dissing it because TDP isn't a good measure fails to appreciate that it's an embedded board and CPU designed for low power consumption networking devices and is used by numerous firewall appliances.

I use an 8-core C3758 variant of one of these boards. I used to have it plugged into a smart plug and its power usage is on a par with a high-end consumer router. I'll get around to reconnecting it to the smart plug again and measure it's real world power consumption so I can post here. May take a day or two...The point was to demonstrate to the OP that there ARE in fact low power consumption x86_64 platforms

I have been disappointed to date with OpenWRT's ease of setup. I remember the DD-WRT and Tomato days of 2005-2013 and it was a huge learning curve going to OpenWRT. I want to unlock the possibilities of my router but otherwise think about it as little as possible. Basically my dream is the webUI of an off the shelf Netgear or whatever but with extra little checkboxes for CAKE and guest network isolation.

In theory, a popular all in one router well-liked on these forums should be easier to troubleshoot. (Because everyone has already had the same problems.) I think OP&friends are making a compelling argument for me to go x86 next time though.

I have never really understood this sentiment, yes, OpenWrt doesn't give you mandatory firstboot guided configuration assistants, but the configuration pages for the basics (wan/ lan) are straight forward and imho on par with commercial- and other 3rd party firmwares (in a sense more so, as it's always the same, for any supported device). While I do share some criticism about the wlan configuration (not all important settings (regdom, channel, essid, encryption) on one visual page, but scattered through 3+ sub-tabs), it's not really difficult either, once you've gone through the pages.

At least to me, it would be quite interesting what exactly looks so strange and difficult to you.

The advanced features are another topic (and to some extent this also applies to sqm), but that's a direct consequence of having the possibilities and flexibilities of these features in the first place, while staying within the confinements of small flash sizes and supporting a wide variety of different devices and device classes (routers, access points, NAS, switches, etc. - from mips over arm, to x86_64, including architectures no one really knows) use cases, features and environments. Of course a walled garden without any installable addons would be easier to streamline, but you lose access to any configuration options the designer didn't think of originally that way.

Disclaimer: I am seeing the firmware interfaces of varying OEM firmwares quite regularly (ideally only shortly, before installing OpenWrt), but I find those often more confusing in their attempt to make things 'easier' and to hide more advanced -but common- settings, be it:

  • unskippable firstboot assistants, which won't work anyways (VLAN on WAN, havving to spoof the WAN MAC address)
  • hiding VLAN settings or not offering them at all (at least not for multiple VLANs on the LAN side)
  • disabling IPv6 by default and only providing a very limited set of options there
  • not always providing access to the channel settings ('mesh' uplinks), while also doing a bad job of avoiding interference/ congested channels
  • only very rudimentary settings for static DHCP leases, often with no local DNS resolution
  • let's not even start about VPN options...
  • very different GUIs, even for devices from the same manufacturer (and the same time frame)
  • a strong push from vendors to short-lived smartphone config apps (we've been there in the past already, with windows-xp-only configuration apps for ISDN pbx systems and routers/ switches, or flash/ java based GUIs...) and proprietary (sometimes paid) cloud integration. Networking devices may easily outlive XP, the android version of the day, flash or browser-side java applets - and the lifetime of any proprietary cloud has been shorter than mine, so far.
  • let's not even talk about the chinese-language-only webinterfaces from Xiaomi

[this might be better split off into another thread, but I am genuinely interested what is considered to be difficult about the basic OpenWrt configuration].