Posted: Mon Oct 19, 2020 10:58 Post subject: WRT300N v1.1 poor wired & wireless performance
Hi all!
I'm having performance issues with my Linksys WRT300N v1.1 and hoping to find troubleshooting support since my efforts were unsuccessful so far...
My network setup is very basic. I have a decent 200/20 Mbps cable connection and a Cisco EPC3825 cable modem/router/AP from my ISP. The intention was to have the setup upgraded in such a way that:
ISP's device would act as a modem only (wireless and any filtering disabled)
WRT300N would serve as the primary device both for wireless and wired network access
So I bought the Linksys WRT300N online - already with the dd-wrt mega onboard. Upon receiving the package I quickly verified that it boots, did a hard reset (it asked about the password change after that) and started connecting everything with a good quality cat. 6 cable (connected the modem's LAN port to WRT300N's WAN port). This being not my first dd-wrt setup - everything went smooth as expected.
Then I saw the speed testing results with the outside world:
WRT300N maxes out consistently at 30/10 Mbps - both on wired and wireless (~15% downlink, ~50% uplink)
I tried doing many changes to the setup trying to troubleshoot it. Tried different Ethernet cables, power adapters, wifi channels, settings for TX power, overclocking.
I even re-flashed the device from the pre-installed mega build to v3.0-r40559 std (08/06/19) - tried the double NAT and AP-only approaches with the same outcome. I'm running out of ideas now but it does somehow point me to a hardware issue with the Linksys WRT300N itself.
Probably important to note that I am able to get the full 200/20 Mbps when connecting to the Cisco device via cable (has Gigabit LAN). The wifi speeds are slower at around 70/20 Mbps (there is a lot of surrounding networks). I'm testing the speeds mainly using the speedtest-cli command line tool - got similar results from the web version or other services.
Would be grateful for any tips how to approach this in a logical manner. Are there any tools or techniques I could follow?
This router have 300Mhz CPU. No way it can handle 200M/sek WAN-LAN. Get a decent router.
Thanks for the response.
This is something I tried to verify and exclude/confirm as an important factor. I underclocked and overclocked the router and the results remained the same.
Perhaps it's just me feeling stupid for buying a device not matching my requirements/expectations... but this is somehow surprising anyway and I'm just trying to understand why is the WAN-LAN routing so CPU heavy and why disabling NAT doesn't change the results at all.
For one: isn't it so that those kind of speeds were achievable (via wire of course) even in previous gen devices?
Also: is it only a matter of the horsepower? My understanding recently were that you need about 1GHz to route a Gigabit connection (if your device supports one of course) so this one having 300MHz should handle the full advertised 10/100 connection just fine.
Perhaps it's just me feeling stupid for buying a device not matching my requirements/expectations... but this is somehow surprising anyway and I'm just trying to understand why is the WAN-LAN routing so CPU heavy and why disabling NAT doesn't change the results at all.
For one: isn't it so that those kind of speeds were achievable (via wire of course) even in previous gen devices?
Also: is it only a matter of the horsepower? My understanding recently were that you need about 1GHz to route a Gigabit connection (if your device supports one of course) so this one having 300MHz should handle the full advertised 10/100 connection just fine.
It should be able to reach 85-95 Mb/s on wired, esp as an AP. My old WRT54G's can reach this and it runs at 250 MHz. Btw, 40559 isn't a good build; try the latest after reading the relevant build thread(s). Then reset and manually set up.
Stick to the older K2.4 builds on this ancient hardware because it WILL run faster than any k3 or k3.1 kernels...
Didn't expect this but that surely improved things a lot! Now I'm able to get 85/20 almost consistently - wired or not. Good stuff.
In the end I'm going to setup this router at my parents house where they have a 60/10 Mbps broadband connection - sounds much more suitable there. Will then look for something newer (and dual band) for personal my setup.
Stick to the older K2.4 builds on this ancient hardware because it WILL run faster than any k3 or k3.1 kernels...
Not if used as a gateway (or non-bridge client/repeater) with SFE enabled. Plus k4.4 will have a lot more security fixes than k2.4, which I'd say is all the more important on a gateway.
But in general I agree @uciekaj should test w/ the broadcom/ build first. I'd have to go thru my gmail to find which models BS said should work on the mips32r1 build, but I seem to recall it some of the WRT300/600-series. _________________ #NAT/SFE/CTF: limited speed w/ DD#Repeater issues#DD-WRT info: FAQ, Builds, Types, Modes, Changes, Demo#
OPNsense x64 5050e ITX|DD: DIR-810L, 2*EA6900@1GHz, R6300v1, RT-N66U@663, WNDR4000@533, E1500@353,
WRT54G{Lv1.1,Sv6}@250|FreshTomato: F7D8302@532|OpenWRT: F9K1119v1, RT-ACRH13, R6220, WNDR3700v4
Joined: 21 Jan 2017 Posts: 1783 Location: Illinois Moderator
Posted: Tue Oct 20, 2020 14:22 Post subject:
jwh7 wrote:
msoengineer wrote:
Stick to the older K2.4 builds on this ancient hardware because it WILL run faster than any k3 or k3.1 kernels...
Not if used as a gateway (or non-bridge client/repeater) with SFE enabled. Plus k4.4 will have a lot more security fixes than k2.4, which I'd say is all the more important on a gateway.
But in general I agree @uciekaj should test w/ the broadcom/ build first. I'd have to go thru my gmail to find which models BS said should work on the mips32r1 build, but I seem to recall it some of the WRT300/600-series.
K4.4 is not available for these ancient routers...
K3.1x is...that said, I could not get the new K3.1/mips32r1 to work on a similar WRT600Nv1.1 a few months back.
The OP is running 12 year old hardware...any hopes for up-to-date security should be a foregone conclusion at this point. And, patches for most BIG/Important security issues should be backported by BS, afaik.... but K2.4 will have its limits for overall updates since it's not actively being maintained by the linux community so that part is true....
I think you are scare mongering here... unless someone can point the us readers to specific security holes in BS's K2.4 implementation... I would not know where to corroborate your security issue claims.
I find that if you're going to run ancient hardware, it's best to sticking to the Kernels that were around at the time...at some point the 8mb in flash and 32mb of ram will not be able to handle the newer kernels without using nearly all the ram.... _________________ FORUM RULES
K4.4 is not available for these ancient routers...
K3.1x is...that said, I could not get the new K3.1/mips32r1 to work on a similar WRT600Nv1.1 a few months back.
When did you try it? When BS ported k3x from 3.10 to 4.4, he included the code for mips_r1:
https://svn.dd-wrt.com/browser/src/linux/universal/linux-4.4/brcm/mipsel_r1
Perhaps he didn't update the makefile to use it though; I'll try to get an update from him on it.
Same idea still applies though. Checking my gmail, BS did add the WRT300Nv1.1 and tested it, at least back in May:
BS wrote:
many of the 2.4 routers can use the k3.10 mipsel_r1 builds like the wrt600n and wrt610n and the wrt300n1.1 (i tested these models already). it just depends if the firmware fits. but its rarely tested on all models. so its not unlikely that it will work on a wrt54gs[*]. running on this old devices was never a issue. only mipsel_r1 support is required and a lot of such old devices have hard memory and flash restrictions. so a k3.x kernel or even 2.6 will not work good on 16 mb ram. in the mipsel32_r1 build i removed all chipset support for wifi chipsets not seen on these older devices to reduce size.
[*] NOTE: last time it was tested (by mrjcd), the WRT54*'s didn't work with it yet.
msoengineer wrote:
The OP is running 12 year old hardware...any hopes for up-to-date security should be a foregone conclusion at this point. And, patches for most BIG/Important security issues should be backported by BS, afaik.... but K2.4 will have its limits for overall updates since it's not actively being maintained by the linux community so that part is true....
That's exactly the point, not all CVE's/etc get backported, and some that do have to be reverted due to driver incompatibility (which can include k3.x backports). Naturally this is much less of an issue with non-Broadcom. For example, the last CVE that was merged:
https://svn.dd-wrt.com/changeset/44324
msoengineer wrote:
I think you are scare mongering here... unless someone can point the us readers to specific security holes in BS's K2.4 implementation... I would not know where to corroborate your security issue claims.
Lol; see the above example. And I'm confident that stating "more security fixes" is not quite "omg you're gonna be hacked!"
msoengineer wrote:
I find that if you're going to run ancient hardware, it's best to sticking to the Kernels that were around at the time
That is simply a limitation of proprietary drivers. I find it's best to at least test the latest kernels that are compatible. Luckily BS has access to them and the desire to port them to k3x, now k4.4 (at least for normal 'k3x' build), and he said he 'eventually' plans to move them to k4.9. And if one tests them and find it doesn't work as well for their application, they can revert to the better build.