Hi, I just installed Gargoyle on a WRT54GL v1.1 router and it works great. Actually wireless "Client + AP" does not work with the AP using WPA2 but I can live without that.
I know this router is low on RAM (16MB), and my main attraction to it are the features with graphing bandwidth usage (per month per LAN client), and also URL usage and blocking (I have 3 young kids).
So my question is, if I disable the wireless function entirely in Gargolye (I can use a second AP in bridged mode) then does this free up significant CPU and memory resources, so it may perform the other routing functions more reliably?
In other words, less CPU and RAM usage with 5 ethernet "clients" compared to 2 eth and 3 wifi? Or the radio chip is sufficiently capable that it makes no difference?
Thanks Eric! I look forward to getting Gargoyle on more powerful hardware in the future (ideally preloaded and for sale from you).
-Emmett
WRT54GL faster with wireless off?
Moderator: Moderators
Re: WRT54GL faster with wireless off?
There is definitely some overhead associated with wireless, but I'm not sure exactly how great it is. The overhead will naturally be greater if you're using encryption.
You'll probably get at least slightly better performance, provided you're using the same total amount of bandwidth if you have 4 wired (there are only 4 LAN ports on the WRT54GL, not 5 -- one ethernet port is for the WAN), instead of 2 wired and 2 wireless. But that caveat is a big one -- more bandwidth is available on an ethernet port than via wireless, so you may end up increasing bandwidth if you do that, and therefore system resources required.
Sorry I can't tell you exactly how big the trade-off is: the best strategy may be to just try it and see.
You'll probably get at least slightly better performance, provided you're using the same total amount of bandwidth if you have 4 wired (there are only 4 LAN ports on the WRT54GL, not 5 -- one ethernet port is for the WAN), instead of 2 wired and 2 wireless. But that caveat is a big one -- more bandwidth is available on an ethernet port than via wireless, so you may end up increasing bandwidth if you do that, and therefore system resources required.
Sorry I can't tell you exactly how big the trade-off is: the best strategy may be to just try it and see.
Re: WRT54GL faster with wireless off?
Thanks a lot Eric. That mostly answers my question.
I be clearer, I think I should have asked how does CPU+RAM usage compare with a sustained 20Mb/s from the WAN port to 1 eth port, versus 20Mb/s from the WAN port to 4 wireless clients using WPA2 on the built in broadcom wireless AP.
Like you say, just try it and I will.
I do have a separate 8 port eth switch, and a separate AP. I want the Gargoyle router to put all its CPU and RAM resources toward bandwidth logging and URL checking and perhaps QoS.
Is the upper limit purely performance, or can things get seriously ugly when RAM is insufficient?
I will get better hardware when I have time to work on it. Maybe I just buy a WRT54G-TM from the USA. I am in Aus. Or perhaps I try a firmware build myself for an economical but nicely speced MIPS+Atheros based router like the TP-Link TL-WR941ND. But I've not looked into doing my own builds yet.
I be clearer, I think I should have asked how does CPU+RAM usage compare with a sustained 20Mb/s from the WAN port to 1 eth port, versus 20Mb/s from the WAN port to 4 wireless clients using WPA2 on the built in broadcom wireless AP.
Like you say, just try it and I will.
I do have a separate 8 port eth switch, and a separate AP. I want the Gargoyle router to put all its CPU and RAM resources toward bandwidth logging and URL checking and perhaps QoS.
Is the upper limit purely performance, or can things get seriously ugly when RAM is insufficient?
I will get better hardware when I have time to work on it. Maybe I just buy a WRT54G-TM from the USA. I am in Aus. Or perhaps I try a firmware build myself for an economical but nicely speced MIPS+Atheros based router like the TP-Link TL-WR941ND. But I've not looked into doing my own builds yet.