wireless benchmark

Discussion in 'Cisco/Linksys Wireless Routers' started by _Shorty, Jul 27, 2005.

  1. _Shorty

    _Shorty Network Guru Member

    got bored yesterday, and frustrated with alchemy's wireless speed, and decided to benchmark the wireless speeds of a handful of firmware for my v1.0 WRT54GS, out of curiosity. I transferred a ripped DVD from my machine to a wireless machine that has a Linksys WMP54GS card in it with speedbooster enabled. Not like that matters, I don't think, since the speeds are nowhere near rated. Signal strength is excellent, connected at 54Mbps, but don't get anywhere near that with any of the firmware. CPU usage on the router with all (well, except maybe the Linksys firmware since I can't tell with that one, but I'm sure it must be too) the firmware is incredibly high, presumably a bug with the driver for the wireless...

    7 File(s) 4,338,293,152 bytes

    alchemy 1.0

    2779.95 seconds
    bah, 2780 is good enough
    2780 seconds


    linksys 4.70.6


    1233 seconds


    HyperWRT 2.0

    4:44:57.15 15:03+5:53=20:56

    1256 seconds


    DD-WRT v22 final r2

    5:43:43.62 16:17+22:14=38:31

    2311 seconds



    alchemy 1.0


    linksys 4.70.6


    HyperWRT 2.0


    DD-WRT v22 final r2

  2. howardp6

    howardp6 Network Guru Member

    The results for stock Linksys firmware are quite good. From a standard 802.11G connect you should get 22Mbs to 24Mbs. The speedbooster gives you about 20-22 percent improvement. HyoerWRT gives repectable results also. The 54Mbs is a connection speed not a measure of throughput. There is overhead, which further redudes your throughput. Witn a WRT54GX and a WPC54GX I was able to get about 41.8Mbs measured by IPERF. The connection speed is supposed to be 108Mbs. So percentage wise you are getting a higher percentage of the connection speed than I am. With a WRT54Gv1.0/WMP54G combination I was geeting a throughput of 23.7Mbs at the same location. Were you using the DVD drive as a source? You're results would be better if the data came from your hard drive since its' sustained tranfer rate is much higher than the DVD drive.
  3. _Shorty

    _Shorty Network Guru Member

    I know there's *some* overhead, but it can't be that high, can it? I mean, without WEP or anything else turned on, it should be capable of more than that. I thought I'd read somewhere here or wrt54g.com perhaps that the wireless driver in these things isn't using DMA properly or something, something that could be fixed to bring down the CPU usage in them anyways. Which was also the likely cause of the slow throughput. There really shouldn't be any reason for the CPUs in these things to be going to 70-90% while moving data that relatively slow over the wireless connection. It's not a powerhouse by any means, but still, a 200MHz CPU shouldn't have to work that hard to move that relatively small amount of data around.

    The computer was reading the files from a raid0 volume, so I don't think the HDs were presenting much of a bottleneck at all. The wireless end was writing to a single drive, but its HD light barely lit up, just blips here and there.


    Ah, just googled for 802.11g throughput, came up with several interesting articles, such as this one.


    Didn't realize they'd cut back on the throughput so much. Still can't believe how much CPU is being used to move data though, quite surprising. I guess there isn't much room for improvement after all as far as throughput goes, but I wouldn't be surprised if there were drivers bugs...

    edit - hmm, might my numbers just below 20Mbps be due to speedbooster/afterburner not being functional? Pretty sure I had it turned on in the config pages for those firmware, but perhaps not.
  4. howardp6

    howardp6 Network Guru Member

    The lower numbers may be a result of thr firmware. Some firmware (un-named) have been known to have lower throughput. The CTS protection should be disabled. The enabled setting was to enable 802.11B devices to operate on the router. I understand the the presence of a 802.11B client will cause the route to lower throughput. I do not know that for certain, but it has been mentioned is some forums. The article was interesting.
  5. _Shorty

    _Shorty Network Guru Member

    I did have it set for mixed B & G clients, although I have no B clients here. I only have it set that way for a few relatives who sometimes visit, whose machines happen to be old enough to have B adapters in them. I do believe it is supposed to be slightly faster if you set it to G-only, but I don't know how much faster. Got a busy out of town weekend ahead, but maybe I'll test that out when I get back. (Motley Crue concert! haha, ah, teenager in the 80's, woot :) )
  6. superjohnny

    superjohnny Network Guru Member

    Since nobody has asked the obvious I'll go ahead...

    The Alchemy and DD-WRT are supposed to be a hacked versions of the Linksys firmware. Why then are they so much slower?
  7. bergi

    bergi Guest

    Because they use the old linksys code (ver 3.xx) which had this poor performance as well.

    btw. DD-WRT V23 will be much faster than V22.

  8. none

    none Network Guru Member

    I am using the Alchemy 1.0 firmware on my two WRT54GS boxes for a WDS bridge and the actual throughput is aroung 8Mbps with a 54Mbps radio link.

    I enabled G-only mode, burst mode and WPA.

    Is there any firmware with WDS support based on the official Linksys firmware 4.xx?

    Alternative, is there any way to use WDS with the official firmware?

  9. _Shorty

    _Shorty Network Guru Member

    well, IIRC, you're supposed to get your bandwidth cut in half with WDS, so that looks like a pretty good result considering that.
  10. none

    none Network Guru Member

    I don't think I should get hald bandwidth since I am using the wireless connection only for the WDS bridge.

    The clients are connected to the WRT54GS boxes with a wire.

  11. SirDracula

    SirDracula Network Guru Member

    I have a WRT54GS with a Dell 1350 mPCI client card, both with Afterburner enabled. I'm using iperf to get the bandwidth numbers (http://dast.nlanr.net/Projects/Iperf/)

    I get 14-15 Mbits/sec. with Afterburner enabled.

    If I turn off Afterburner it doesn't make a difference, so I'm not sure it's really using Afterburner.

    This is using Alchemy 1.0 with the Afterburner option set to 'On'

    As others noticed the CPU usage on the router is very high. In fact I have mine overclocked to 216Mhz (you can find the details on how to do it in some other threads) otherwise my windows file sharing would not work (the CPU would be pegged at 99% on large transfers; with the overclock it still 90-95% but at least I can transfer files).

    Can't wait for the 802.11n wireless routers which should also have 1Gb switch also (yes, we'll need to upgrade all client cards). Then we can talk about networking, so far the numbers are laughable.
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice