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

Linksys E2500 v3: What is the expected throughput?

Discussion in 'Cisco/Linksys Wireless Routers' started by WhiZZ, Oct 3, 2016.

  1. WhiZZ

    WhiZZ Addicted to LI Member

    Hi guys, looking for answers from E2500 owner.

    I bought Linksys E2500v3 since I was looking for mid-range Broadcom-based router to run Tomato and Linksys hardware seems to be decent-enough quality reflecting from past experience.

    I am running latest Advanced Tomato 3.3-138 (basically Shibby with different web UI).

    However I am a bit disappointed with the speed as I can only push the wireless link into 50 Mbps up/down on my symmetric 100 Mbps internet connection. That's for 2,4 Ghz btw. The 5Ghz performance is even worse, for some reason it was capped to 30-40 Mbps. The speed numbers are similar with stock firmware.

    Typical 2.4 GHz speed:
    [​IMG]
    and.. 5 GHz speed:
    [​IMG]

    I am talking about speed with the router in 3-4 meters reach radius (it's for a single room) with Intel 2x2 7260 Laptop (link speed 300 mbps). I have fine tuned the wireless to occupy most less busy channels, tries to toggle 20/40 MHz options, toggling preamble options, use AES cipher and even tried different MIMO capable devices

    By comparison, my backup router D-Link DIR412 N150 + DD -WRT achieved 40 - 50 Mbps (on 150 Mbps link) just fine with the similar config (channel, wireless mode, etc) and distance, so I think something is off with my E2500, I was under impression I should be able to push 60-70 MBps on 5 Ghz with E2500. No USB devices to share the bus with 5Ghz either.

    A bit Google seems to confirm my suspicion:
    1. PCMAG review states the throughput on stock FW is around 60-mbps
    2. Different setup (WDS bridge), but says E2500 5Ghz is capped on 30-40mbps [also this]
    To those who can get better speed, can you share your settings and firmware version, please?

    Cheers,

    Whizz
     
  2. xips_

    xips_ Serious Server Member

    Running within a crowded & noisy 2.4GHz neighborhood.

    R7000 running as gateway on v1.38 with 50/5 Mbps link. Inside same room @ 25 ft.

    2.4GHz is Channel 9@40Mhz width with WPA2+AES
    5GHz is Channel 153@80Mhz width with WPA2+AES


    E2500v1 running as access point behind gateway on Shibby v1.28 for e2500v1. Had problems with multi versioned Shibby FW. Normally only run with 5GHz band enabled & 2.4 GHz disabled due to noisy neighborhood, I have never tuned the 2.4 band on E2500 and only enabled the band for this quick test. readings through 1 brick wall.

    2.4GHz - Channel 11@20Mhz width with WPA2+AES
    5GHz - Channel 158@40Mhz width with WPA2+AES

    All upload speeds match providers' bandwidth limit.

    I found v1 of the e2500 sometimes temperamental with current FW versions giving random reboots. I had a similar problem to you and found that my former ISP carrier - at&t throttled all my wireless devices. Understand this is not your problem due to your other router test.

    5G-e2500v1.jpg 2.4-e2500v1.jpg
    5G-r7000.j7hdnpoe.jpg 2.4g-r7000.Co34PxWuD.jpg
     
  3. WhiZZ

    WhiZZ Addicted to LI Member

    xips__ Thanks for replying! I see that the speed goes pass 50 Mbps just fine on E2500 5 GHz band.
    What build number of Shibby do you use?

    Could you perhaps also share you advanced wireless settings? also is this N-only or Mixed AN-mode?

    I'm also interested on WLAN throughput test, though. So maybe if you feel like to post http://totusoft.com/lanspeed result or something else you like would be awesome.

    Eh how is this even possible? do they check the MAC or what? but everything supposed to be gated by our router, right?

    Cheers,

    whizz
     
  4. xips_

    xips_ Serious Server Member

    Tomato Firmware 1.28.0000 MIPSR2-128 K26 USB Max
    Advanced wireless setting are all default. Wireless mode set to auto.

    Never had any issues with LAN thruput. I only use the E2500 for a laptop with spotty R7000 coverage.

    Throttling is probably done via MAC. Two laptops & 5 different Androids experienced 50-75% throughput reduction with at&t. Switched carriers and now all throughput is equal. at&t pull so much crap that made me so furious that I'm probably on a terrorist watch list somewhre. They throttled netflix, published my fax line in the phone book as my voice line, allowed a third party to charge my line for voice mail services without my authorization and much much more. imo, at&t is criminal and should burn eternally. I get the same services from Cable One now for half the price at&t charged. Everything works and is faster.
     
  5. WhiZZ

    WhiZZ Addicted to LI Member

    Found the root cause: it's due to CPU bound

    I used top command during speed test and noticed the CPU is maxed by PPTP server, once I disabled PPTP server, 2.4 GHz speed went back to normal (60-70 mbps with 3-4 meter range).

    However 5GHz speed is still capped to 30-40 Mbps, with high CPU usage generated by process " [events/0]". Apparently this process handles pending events in the kernel generated by drivers. So after further digging I found out this bug actually has been discussed quite a long time ago:
    http://www.linksysinfo.org/index.php?threads/shibby-v104-bug-e2500-throughput-on-5-ghz.62951/

    This could be the explanation: https://forums.anandtech.com/threads/e2500-5ghz-wds, but -w-tomato-limited-to-30mbit-throughput-with-130-link-rate.2293102/#post-35938873

    Some users report the 5 GHz throughput is okay with stock FW, perhaps Tomato implementation of the 5GHz driver is a bit different, thus the speed cap.

    TL;DR E2500 5GHz radio sits on USB-bus instead of PCI-bus, Network I/O is bounded by CPU and Tomato 5GHz driver is not as optimized as stock driver.

    Final note: beside disabling PPTP, I activated the bcm_nat (seems to help lower the CPU usage) and now the 5GHz is capped at 50Mbps-ish.
     
  6. WhiZZ

    WhiZZ Addicted to LI Member

    Updates: Finally getting full speed (60-70mbps) at 5 GHz by overclocking the CPU to 500 mhz. I Also found out 5 GHz is buggy with build 132 and 138
    So I am staying at build 128.

    This is semi-destructive hack as overclocking generally shorten the device lifespan..
     
  7. Incidentflux

    Incidentflux LI Guru Member

    Trying to figure out which is a stable, fully functional release for E2500 version one (v1) hardware? Any users here?
     
  8. Incidentflux

    Incidentflux LI Guru Member

Share This Page