1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Deminished BW using WAP v. Wired through a WRT54GC

Discussion in 'Cisco/Linksys Wireless Routers' started by mphare, Dec 30, 2005.

  1. mphare

    mphare Network Guru Member

    Since I got Verizon FIOS, I've been testing my BW using http://whisper.cs.utk.edu:7123

    When I am wired connected through my WRT54GC I get really good BW, about 12.5 to 13 Mbps.

    However, when I come into the WAP side of the 54GC I get congestion and re-transmits and much less BW (see the 10s inbound test below, only about 5.5 Mbps)


    TCP/Web100 bandwidth test v4.12
    click START to begin
    running 10s outbound test... 1845 Kbs outbound
    running 10s inbound test... 5761 Kbs inbound

    Web100 Inbound Summary:
    Packets to your host: 5772 of size 1260 bytes
    average RTT: 40.0 ms minimum RTT: 30.0 ms
    Maximum Expected Bandwidth: 17 Mbs
    Inbound limits:
    congestion 85% server: 0% your host: 14%
    Bummer!
    packets retransmitted: 11 congestion events: 4
    Your host's receive buffer: 65520 bytes
    Server's send buffer: 82574100 bytes
    receive buffer: 65535 bytes
    Buffer sizes are very important in determining
    network performance. Larger buffer sizes can
    help increase thruput. If your buffer is smaller
    than the remote host, you should investigate
    increasing your socket buffer sizes.

    Web100 reports Tweakable Settings as:
    RFC 2018 Selective Acknowledgment: ON
    RFC 1323 Time Stamping: OFF
    RFC 1323 Window Scaling: OFF

    click START to re-test


    Any ideas why I'm losing BW on the WAP side?
     
  2. 4Access

    4Access Network Guru Member

    Is the wireless nic in the client that connects to the WRT54GC a 802.11g nic?

    What connection speed is the wireless nic reporting? It's possible the wireless client is simply too far away from the router or is experiencing a lot of wireless interference and is operating at a slower speed.

    I'd suggest using one of the following utilties to test the wireless throughput between a wireless client & a client connected to a LAN port. If you only get about 5.5Mbps in that test then you need to start looking for the cause of the wireless slow down.

    NetCPS
    Qcheck
    Iperf
     
  3. mphare

    mphare Network Guru Member

    I'm about 10 feet from the WAP and my NIC reports 48 Mbps (yes, it's 802.11g).

    I'll try the utils you referenced and see what they say.

    Thanks!
     
  4. mphare

    mphare Network Guru Member

    OK, I ran NetCPS (Thanks for the link, nice program)
    and here are my results.

    The first part symbolically shows my network configuration.
    FIOS to a router (theirs), to another router (mine) to a 10/100 switch to the WRT54GC and to a 10/100/1000 switch
    The 10/100/1000 switch goes to two computers (a P3 and a P4), the 54GC goes to two laptops (Celeron and a Pentium M)

    To compare, I tested the Pentium M both through the WiFi and through the wired connection on the WRT54GC.

    I'm assuming CPS means Characters (or bytes) per second, so I multiply these number by 8 to get the bits per second numbers. Other wise I have lousy throughput across my entire network.
    I'm a little worried about the numbers across the Giga-Bit switch. My assumption is this is a limitation of the NetCPS program, not my network.

    Any ideas why the WiFi is so much slower than the wired connection through the WRT54GC?

    Both computers are within 10 feet of the 54GC and I have about a -35 dB signal with -85 dB of noise.
     
  5. mphare

    mphare Network Guru Member

    So what are some possible causes for WiFi Slow-down?

    I've tried several channels and none seem to make any difference. My computer has sniffed out about three other WiFi's in my neighborhood, but I don't broadcast my SSID and I set hard profiles to only use my SSID on my laptops.

    My phones are 2.4 GHz, but they are not in use most of the time. My microwave is turned off and I don't have any fluorescent lighting.
     
  6. Private-Cowboy

    Private-Cowboy Network Guru Member

    I think the WRT54GC is the problem. As good as it looks and as good as the reviews (like the one from linksysinfo) are - lets tell it like it is - this thing is crap.

    I tested mine extensively the last weeks and while it's easy to use and very good looking and performs well for surfing / streaming audio and images / voip it just does not deliver a good bandwith.

    I tested it with netcps too wireing my laptop to the router and leaving my PC wireless:

    - In client mode my PC gets an avg of 1.52MPS (thats basically 12.2mbit) and transfers the 100MB in 65.88secs.
    - In server mode my PC gets 1.45MPS (11.6mbit) and transfers the 100MB in 68.94secs.
    - Both directions are pretty similar and wireing the PC and leaving the Laptop wireless doesn't change much.

    I also tested it with the media center extender manager (the tune app) and my Xbox 360. Wire PC and Xbox and I get max bandwith, wire only one and it gets ugly - the signal level barely reaches the "sdtv mark" so its something around 6-8mbit. Having both wireless works sometimes at a level of 6-8mbit (depending on weather, sun or god almmighty - who knows why the one day its good and stable and the other it's fluctuation as hell) and I'm able to stream wmb-hd of 6mbit without lag. But most of the time this is not possible.

    The signal on the PC side is very good, between 90-100% and on the Xbox 360 side its 3/4 bars.

    Initially I thought I could do a little bit more with wifi than just surfing, gameing and voiping but that seems impossible - at least with the WRT54GC. I read so often that wifi-g networks normally reach only half of that (around 20mbit so to speak) but I'd be hapy to even see half of that half, even with descent signals.

    So for me wifi is a failure. I should have bought some cables instead. I spent like $250 on my wifi stuff (router, pc-usb-adapter and xbox-360-adapter) and that not worth it.
     
  7. Private-Cowboy

    Private-Cowboy Network Guru Member

    Ok, I can now confirm what I already thought.

    I bought a WRT54G v3.1 in the local electronics store (they got plenty of 3.1s left) and flashed the newest official firmware (4.20...), no 3rd party experiments yet. And I've just set up the network, no tweaking (enable frame burst etc) done yet. And even in that unoptimized state the WRT54G outperformes the WRT54GC (I had an external 5dbi antenne on my GC):

    - The signal level on my PC went from 53dbi (today in bad bad weather) to 40dbi. That is a lot better in my eyes.
    - The Netcps tests (unfortunately done with unoptimized antenna position and a signal of 54di, so lower than with the GC) went from 1.5MPS (like 12mbit) to 2MPS (like 16mbit). I should do them now with 40dbi and optimized antenna position and I bet it would be even faster.
    - The Windows Media Center extender bandwitch nearly doubled and now is without the deep drops my GC suffers these bad-weather-days. Streaming of wmv-hd content is now possible without a hitch, even 8mbit ones.
    - In better weather and with optimized settings (and 3rd party firmware) the G could easily be twice as fast as the GC is. In the totally unoptimized state my ne G is running now it already is 50% faster than my optimized GC.

    So sad to say that but the GC is not made for high performance.
     

Share This Page