I flashed some WNDR3700's quite a while back now (maybe 10 years) and they've been working fine until recently one died due to overheating.
If I'm going to have to replace one unit I may as well replace the other three at the same time to only have to support one platform.
I'm looking for at least 1GB hardwired LAN (are there any 10GB units yet) and able to have two radios active at the same time (my current setup has both 5Ghz and 2.4Ghz radios active as I have some legacy nets that need the 2.4Ghz stuff).
I'm pretty relaxed about doing my own custom builds and although I haven't done it for a few years am pretty sure I can pick it up again without too many problems.
I don't want to spend a whack of money on these, so they don't need to be the latest greatest functionality, as long as the have concurrent 2.4Ghz and 5Ghz nets that's what I must have.
Understandable rationale. Maybe start by selecting a successor and replace the failed unit only first and see how it delivers, before going all in? In my limited experience the wndr3700 (v2 in my case) series was a pretty lucky design that worked above expectations....
I think 2.5 Gb ethernet is slowly creeping into the homerouter market, but 10 Gb seems still in the rare and expensive bin....
I note these are 880KHz MIPS based so similar in CPU performance the your old 3700s, so if the 3700 sufficied, the wsm20s likely will as well.
I agree in principle. In practice this is a bit more subtle though. If your expected "load" for the expected life time of the new units does not change much about the current load, it is pretty immaterial whether the CPU family is deeply EOLd already or still recent. The question boils down to will OpenWrt support that platform for the desired life time, a question that is somewhat hard to answer, but my gut feeling is, probably yes, but that might depend on the upstream kernel, if say MIPS is unceremoniously dropped from upstream, i wonder whether/how long OpenWrt would still carry it? A similar question poses itself for 32bit arm designs (but I assume their sunsetting days are likely further out than MIPS').
I think there's so much legacy implementation of the MIPS versions that even if upstream got EOL'd then all that really affects is new builds.
The units I have I literally custom built about 10 years back now (Kernel 3.10) and they've served me well, with that in mind I've even looked at perhaps picking up some legacy ones from E-Bay and just reinstall my original build on them (E-Bay looks like I can pick up units for circa £10-£15 each).
But, I do recognise that some of the newer features like WiFi6 and others would be nice to have, if not essential.
For me these are functional devices rather than development inquisitions, so while I might look to 'enhance' what the current units do if I have to spend the time doing a new clean build, the reality is that that's not a priority for me.
I do think if MIPS was dropped from upstream openwrt would carry on just fine, but perhaps with less 'core' development and more 'tinkering maintenance'.
It's dead in a sense that you have no new platforms utilizing it, upstream (vendors) have dropped MIPS a long time ago and performance is lagging behind pretty much everything. You can also claim that a Penitum 75 is "perfectly fine" because it boots but I wouldn't call it usable for everyday use today.