Kernel Version vs Firmware Version

DD-WRT K2.6 vs K3.x vs Tomato Router Firmware Head to Head on Linksys E4200 5   Recently updated!


In March of last year, I published this article claiming that the latest (at the time) DD-WRT builds based on the Linux 2.6 kernel (K2.6) ran faster on a Linksys/Cisco E4200 router than their K3.x counterparts. Since then, I’ve heard apocryphal stories from some in the DD-WRT forums that the K3.x builds have now “caught up” to the K2.6 builds in terms of speed. But does the evidence actually show that?

The answer is, as is usually the case, “it depends.” On more modern router hardware (higher speed processor and more available RAM), I suppose it’s possible. But I wanted to see if it was true for the Linksys E4200 router. As of 2017, it’s a bit long in the tooth in hardware terms, but it’s still a very popular device for third-party firmware fans. My testing was limited only to the E4200 v1, so while it’s probably safe to extrapolate my results to other Linksys E-series router of approximately the same vintage, any firm conclusions beyond that will require your own testing.

Kernel Version is Different than Firmware Version

Before I get to the testing, I’ve seen some confusion regarding DD-WRT version numbers and the Linux kernel version on which they are based. If you look on your DD-WRT router’s “Status / Router” tab, you’ll see a “Kernel Version” and a “Firmware Version:”

Kernel Version vs Firmware Version

Kernel Version vs Firmware Version

Don’t confuse a K3.x build (meaning based on a Linux 3.x kernel) with the v3.0 version number. They are not related at all, and it’s just coincidence that they kind of match right now. All the DD-WRT builds I tested were DD-WRT v3.0.

Testing Method

I ran all these tests on January 1, 2017, while enjoying some vacation time at our cabin in East Wenatchee, Washington. Because it sits on the Columbia River, the location is ideal for speed tests; it’s connected to the high-speed fiber network installed by the public utility district to link the hydro-electric dams along the river. Localtel, my local ISP, hosts a test server on Speedtest.net, which I used for all tests.

My cabin sits in a sparsely populated area, and the only two neighbors close enough to create channel interference use WiFi channels 1 and 11, leaving channel 6 wide open for me on the 2.4Ghz band. Neither of my neighbors use the 5Ghz band.

Wired testing was done using a MacBook Air 7,2 (early 2015 vintage) and a Cable Matters SuperSpeed USB 3.0 to RJ45 Gigabit Ethernet Network Adapter with a 3′ Cat6 patch cable. WiFi testing was done using the same MacBook Air’s internal WiFi adapter, as well as an iPhone 6. For both types of tests, no other network clients were attached to the router.

Wireless speed testing was performed separately on the E4200’s 2.4Ghz and 5Ghz antennae by disabling one of the antennae at a time. A custom SSID was used to ensure no other wireless clients were connected besides the two test clients.

The router’s nvram was erased and the router rebooted before and after each firmware flash. All firmware default settings were used, with the exception of the following:

  • Router was overclocked to 533Mhz.
  • WiFi channels were set to 6 on the 2.4Ghz band and 36 on 5Ghz band, with no interference on either band. Only one band was active at a time.
  • 2.4Ghz band was set to “N-Only” and 5Ghz was set to “NA-Mixed.”
  • Wide HT40 (40Mhz) channel width was set on both bands.
  • WPA2 Personal encryption was enabled.
  • SSHd and UPnP were enabled (as both likely would be in a real-world setting).

I used a custom SSID of “E4200” and the password “testing123” for the WiFi tests to ensure no rogue clients could connect during testing.

After each flash, I ran a total of 15 speed tests:

  • Three wired speed tests.
  • Six 2.4Ghz band tests alternating between the MacBook Air and the iPhone until each had completed three tests.
  • Six 5Ghz band tests alternating between the MacBook Air and the iPhone until each had completed three tests.

After completing all 15 tests on all firmwares, I noted there was no statistically significant difference between the wireless speeds of the MacBook Air compared to those of the iPhone. In general, the iPhone’s WiFi tests were slower than the MacBook Air’s, but they always correlated in a comparison of the head to head tests, i.e. if the MacBook Air was faster on one firmware vs another, so was the iPhone. For this reason, and to make the reports below a bit easier, I’ll only include the MacBook Air’s WiFi results in my findings. But know that the iPhone’s speed tests did nothing to skew the results. They match up perfectly with the MacBook Air’s.

I’ve also included the highest download and upload speeds in each batch of results below, and will use each firmware’s highest value to compare. Taking an average of speeds isn’t that useful or statistically accurate, since I’m trying to test to maximum potential speed in a real-world setting.

It’s also interesting to note that based on the test results, my ISP is clearly attempting to cap download speeds, and doesn’t appear to do so on the upload speeds. So for the majority of tests, the upload speed was faster than the download speed.

E4200 DD-WRT and Tomato Firmwares Tested

I started with a fresh flash of the latest stock Linksys firmware, Version 1.0.06 (build 3): FW_E4200_1.0.06.003_US_20140520_code.bin.

Next, I tried testing with the latest-available build of DD-WRT, which was 30949, released on December 15, 2016. But it proved too unstable on the E4200 to test properly and would lock up the router requiring a power-cycle, so I used build 30880, which (as of the writing of this article) is the stable build I recommend in my E4200 Max Speed settings article.

I flashed and tested DD-WRT builds in this order:

  • K2.6 Mini (trailed build): dd-wrt.v24-30880_NEWD-2_K2.6_mini-e4200.bin
  • K2.6 Mega: dd-wrt.v24-30880_NEWD-2_K2.6_mega-nv60k.bin*
  • K3.x Mega (trailed build): dd-wrt.v24-30880_NEWD-2_K3.x_mega-e4200.bin
  • K2.6 Kong 22000++ build: kingkong-nv60k-broadcom.bin**

I included the K2.6 Kong 22000++ build in my tests because it’s been a solid and stable choice for the E4200 since 2014. During testing of the above builds, there was no measurable wired or wireless speed differences between any of the DD-WRT K2.6 builds, so only the DD-WRT K2.6 Mini results are included below.

I also decided to include two of the most popular Tomato K2.6 firmwares for the E4200: the Shibby build tomato-E4200USB-NVRAM60K-1.28.RT-N5x-MIPSR2-138-Mega-VPN.bin and the Tomato RAF build tomato-E4200USB-NVRAM60K-1.28.9014MIPSR2-RAF-v1.3g.bin.

Finally, just for fun, I tested the hardware I used to replace the E4200 router at my cabin: a Ubiquiti EdgeRouter Lite and UAP-PRO access point (though I now recommend the newer UAP-AC-PRO for anyone shopping for an access point). Those results are also included below.

All the above third-party firmwares for the Linksys E4200 (including the stock firmware) are available in my DD-WRT firmware repository at http://ddwrt.stevejenkins.com/.

E4200 Router Speed Test Results

Stock Linksys Router Firmware – Version 1.0.06 (Build 3)

Stock Wired Speeds



Stock 2.4Ghz WiFi Speeds



Stock 5Ghz WiFi Speeds


DD-WRT K2.6 Mini Results – Build 30880

DD-WRT K2.6 Mini Wired Speeds



DD-WRT K2.6 2.4Ghz WiFi Speeds



DD-WRT K2.6 5Ghz WiFi Speeds



DD-WRT K3.x Mega Results – Build 30880

DD-WRT K3.x Wired Speeds



DD-WRT K3.x 2.4Ghz WiFi Speeds



DD-WRT K3.x 5Ghz WiFi Speeds



Tomato RAF K2.6 Results – Ver 1.28.9014

Tomato RAF K2.6 Wired Speeds



Tomato RAF K2.6 2.4Ghz WiFi Speeds



Tomato RAF K2.6 5Ghz WiFi Speeds



Tomato by Shibby K2.6 Results – Ver 1.38

Tomato by Shibby K2.6 Wired Speeds



Tomato by Shibby K2.6 2.4Ghz WiFi Speeds



Tomato by Shibby K2.6 5Ghz WiFi Speeds



Ubiquiti (UBNT) EdgeRouter Lite Router + UAP-PRO Access Point

UBNT ERL + UAP-PRO Wired Speeds



UBNT ERL + UAP-PRO 2.4Ghz WiFi Speeds



UBNT ERL + UAP-PRO 5Ghz WiFi Speeds



Results Comparison

The following table shows the fastest tested download and upload results for each of the firmwares tested (again, there was no difference between any of the DD-WRT K2.6 build variants tested, so only the K2.6 Mini is shown). You can sort the table by clicking on the header or filter using the search box.

BuildConnectionDown (Mb/s)Up (Mb/s)
StockWired417.17 305.83
StockWiFi 2.4Ghz91.3797.27
StockWiFi 5Ghz172.20201.75
DD-WRT K2.6Wired146.94122.64
DD-WRT K2.6WiFi 2.4Ghz52.6869.03
DD-WRT K2.6WiFi 5Ghz93.1865.87
DD-WRT K3.xWired81.0472.57
DD-WRT K3.xWiFi 2.4Ghz43.3261.63
DD-WRT K3.xWiFi 5Ghz68.5763.90
Tomato RAFWired123.72115.43
Tomato RAFWiFi 2.4Ghz52.8756.72
Tomato RAFWiFi 5Ghz95.0883.88
Tomato by ShibbyWired139.33111.76
Tomato by ShibbyWiFi 2.4Ghz59.2067.44
Tomato by ShibbyWiFi 5Ghz125.3384.17
UBNT ERL + UAP-PROWired599.80889.08
UBNT ERL + UAP-PROWiFi 2.4Ghz72.4294.07
UBNT ERL + UAP-PROWiFi 5Ghz96.58161.68

Final Analysis

For hardwired connections on an E4200, the stock Linksys firmware can’t be beat at 417.17 / 305.83 Mb/s. The K2.6 kernel DD-WRT couldn’t even reach half those wired speeds with 52.68 / 69.03 Mb/s. Even worse, the K3.x build was almost half again as slow as the K2.6 with results of 81.04 / 72.57 Mb/s.

For 2.4Ghz WiFi, the stock firmware is again significantly faster than the K2.6 DD-WRT and Tomato builds. And while the K3.x DD-WRT speeds were only slightly slower than the K2.6 speeds, they were still slower: 52.68 vs 43.32 down and 69.03 vs 61.63 up.

For 5Ghz WiFi, the stock firmware finished with a clean sweep. At 172.20 / 201.75 Mb/s, it clobbered the K2.6 results of 93.18 / 65.87 Mb/s. The K3.x was, once again, the slowest of the bunch at 68.57 / 63.90 Mb/s.

Bottom Line: DD-WRT K2.6 vs K3.x

If you have a Linksys E4200 and care only about speed, keep the stock firmware. If you’re willing to give up some speed in return for DD-WRT’s features, always choose a K2.6 build over a K3.x. The Tomato results back this up, as they were similar to the DD-WRT K2.6 speeds and also faster than the DD-WRT K3.x speeds.

Of course, on newer hardware with a faster processor and/or more RAM, it’s possible a K3.x build could be just as fast (or maybe faster?) than the same K2.6 build. But if you have an older Linksys E-Series router, the bottom line is that a K2.6 build will always be faster.

I should also note that I no longer use the E4200 as my primary router. I’ve switched to UBNT routers like the UniFi USG (which I run at our main house) and the EdgeRouter (which I run at two of our secondary residences). Both require a separate access point such as the UAP-AC-PRO, but the performance and features made it an easy choice for me. I still keep old ones around as client bridges, and as back-up WiFi routers if I have an equipment failure.

Do you still run a Linksys E4200 or other E-Series router? Do you still want to run a K3.x build instead of a K2.6 one? Have you tested K2.6 vs K3.x on different hardware with different results? Tell me about it in the comments below!

  • ReD-BaRoN

    I have the UAP-AC-PRO and use the E4200 as the wired-router only, using DD-WRT essentially just for VPN. I’d go back to stock fw if it had VPN support.

    A while back, the UAP-AC-PRO was getting a lot of bad press for performance vs. other AC APs at the time. Have you determine these report to be true? Has recent firmware rectified the issue?

    I’d really be interested in articles you may have on tweaking UAP-AC-PRO performance.

    • There’s really not much tweaking TO do on the UAP-AC-PRO, except keep up with current firmware (I hang out in the UBNT forums to make sure I have the latest).

      I just make sure I have it set to channels that don’t have interference, and things are usually blazing for me. 🙂

  • Im gonna flash back my e4200 to original firmware… BIG THANX Steve!

    • If you want top speed, then that’s the right choice! 🙂

  • Thanks for sharing your findings, Oscar!