Cisco Linksys E4200V2 Maximum Performance Dual-Band N N900 Router

[njweb]

CNET How we test: Access points and wireless routers

• The biggest variable of all, router positioning, is not addressed or optimised - but then this goes for a lot of review sites. Half an inch can make all the difference, let alone which perpendicular alignment was chosen. Especially with a single distant test location this really could skew results.
• The office environment can't be controlled (doors open/closed, obstructions, changing furniture, movement).
• And numbers in a lot of their written reviews don't match the entries on the charts.

All information is useful information though! When are we gonna see your Amped vs E4200 results?

Thanks. I actually meant controlled relative to the test they posted for the V1 earlier in the year, to allow for accurate relative comparisons.
I posted a PS a few minutes ago about the CNet results.

I will run the tests today to compare my Amped R1000 against my Cisco E4200 with RE1000 range extender (and without).
Then I can decide which one to keep...

 

[rhombus]

(Anandtech actually says the new unit's amplifier and antenna are supposed to have been optimized for better, on paper at least, wireless performance).
QUOTE: "There's [sic] also been some optimizations to the antenna and amplifier configuration that should improve range and reduce dead spots."

It's always amusing to see techno journos rehash a press release badly.
Here is said Anandtech article.

If you click on the source listed underneath, you will get to the Cisco press release, which is at least accurate.

 

[rhombus]

Thanks. I actually meant controlled relative to the test they posted for the V1 earlier in the year, to allow for accurate relative comparisons.

You're welcome. What is controlled and not controlled in their testing, applies to V1 vs V2, no?

Looking forward to the results! Thanks.

 

[Shikami]

The RT-N66U could perhaps be the Broadcom router that some are looking for due to the disappointment with v2.

http://vr-zone.com/articles/asus-rt...ter-preview-final-retail-revision-/14242.html

Edit: this link does not work anymore. However this does: http://www.dd-wrt.com/phpBB2/viewtopic.php?p=642825

 

[Unregistered]

The E4200v2 is now showing up in some of the router charts on this site! It's performance seems terrible when compared to the original E4200...

 

[rhombus]

The E4200v2 is now showing up in some of the router charts on this site! It's performance seems terrible when compared to the original E4200...

Not true. The Smallnetbuilder 2-stream test client has changed since the v1 review. So the only truly comparable results are the 5Ghz 3-stream tests, where it looks about even.

I think there might have also been a change of laptop housing/antenna for the old 2-stream client somewhere along the line. Perhaps color coding or something to distinguish these in the charts would be useful.

 

[thiggins]

Not true. The Smallnetbuilder 2-stream test client has changed since the v1 review. So the only truly comparable results are the 5Ghz 3-stream tests, where it looks about even.

I reran some 2 stream tests with the new Intel 6200 client with a V1 E4200 and results are pretty close. This is noted in the review.

I think there might have also been a change of laptop housing/antenna for the old 2-stream client somewhere along the line. Perhaps color coding or something to distinguish these in the charts would be useful.

Two stream test client remains an Acer 1810T. Just the card was swapped. Charts are color-coded to connote different test procedures. Latest method shows green bars.

 

[rhombus]

I reran some 2 stream tests with the new Intel 6200 client with a V1 E4200 and results are pretty close.

Thanks Tim - brilliant stuff. Seems you spot-checked some locations in 2.4Ghz and 40Mhz mode for the v1 - do you have any numbers or plots to share? These would be really useful.

A direct re-test of the v1 with the Intel 6200 for 2.4Ghz 20Mhz Up and Down would be the ultimate comparison though - of great interest for those debating between v1 and v2 I think...! That's just a couple of plots... it is Christmas remember... !

Two stream test client remains an Acer 1810T. Just the card was swapped.

The old Intel 5300 was moved from a Dell laptop to the current Acer though some while back, right? Apologies if I can't keep exact track - I have gone over the procedures as carefully as I can... What colors denote these 2 different card/laptop combinations?

Charts are color-coded to connote different test procedures. Latest method shows green bars.

Thanks, I never even spotted that. Where is the key for the color coding? Sorry if I just can't find it.

 

[RogerSC]

Great review of the V2, thanks for that...I missed why you have blanks at a couple of the locations that you used to test in the wireless throughput tests. I'm assuming, because of the consistency, that you're no longer testing those locations (B and E), is that correct?

Thanks.

 

[thiggins]

Thanks Tim - brilliant stuff. Seems you spot-checked some locations in 2.4Ghz and 40Mhz mode for the v1 - do you have any numbers or plots to share? These would be really useful.

Links to plots of retest data have been added to the review.

The old Intel 5300 was moved from a Dell laptop to the current Acer though some while back, right? Apologies if I can't keep exact track - I have gone over the procedures as carefully as I can... What colors denote these 2 different card/laptop combinations?

Thanks, I never even spotted that. Where is the key for the color coding? Sorry if I just can't find it.

Go to Router Charts and select a wireless benchmark, like this. Click the Benchmark Info link right under the Router Charts image.

 

[thiggins]

Great review of the V2, thanks for that...I missed why you have blanks at a couple of the locations that you used to test in the wireless throughput tests. I'm assuming, because of the consistency, that you're no longer testing those locations (B and E), is that correct?

Blanks mean there is no test data.

Reason for reducing test locations is noted here.

 

[Shikami]

Such horrible performance, and limited connections. Such a disappointment this router is. I would not be surprised if there is some technical issue with it, or they are working on fixes to resolves the performance of this router. To pay that amount of money for such limitations is nothing but asinine.

 

[thiggins]

What is "horrible" about the performance?

 

[Shikami]

What is "horrible" about the performance?

Without going into too much detail, why would anyone pay for the inferior version 2? Compared to version 1 it is not worth the price premium. You cannot easily be supported with 3rd party firmware. Version 1 has around 3x the routing performance, and at least and around 4x the available connections for NAT. Going to version 2 should be an upgrade or equal in performance. The only trade off is 4x the USB performance. To me that is horrible. Is that not logical enough?

Something has to be wrong with it to be that limited.

 

[thiggins]

I suppose it depends on what your criteria are for a router. The primary reason for the V2 is to add three stream N support to the 2.4 GHz band. I think the USB performance improvement came as a bonus with the processor switch.

As I've said many times, throughput far in excess of what your ISP connection provides does not provide any benefit. While I'll agree the drop in simultaneous connections might be more likely to affect users, 8K connections should be fine for most cases.

There is nothing wrong with the router. It is simply designed to different performance criteria than you have high value for. I do think Cisco should restore the outbound port filtering option, however. This can be done easily in new firmware.

 

[Shikami]

As I've said many times, throughput far in excess of what your ISP connection provides does not provide any benefit.

Not to create contention, but I will always disagree with this. If this was true, in the logical sense. This would mean that nothing more than a WRT54G would suffice for most users. But you should know, and I know that there is more to PPS. To say that a faster router would not benefit a network, I think is illogical. Even if the allocated bandwidth is not exceeding the computation power of the router.

 

[thiggins]

Not to create contention, but I will always disagree with this. If this was true, in the logical sense. This would mean that nothing more than a WRT54G would suffice for most users. But you should know, and I know that there is more to PPS. To say that a faster router would not benefit a network, I think is illogical. Even if the allocated bandwidth is not exceeding the computation power of the router.

I would love to see any data you have to support your assertion.

 

[Shikami]

I would love to see any data you have to support your assertion.

This is networking 101 and it was part of my education. Because of such, I will always use it to grade even a SOHO IGD:

http://www.cisco.com/web/about/security/intelligence/network_performance_metrics.html

http://en.wikipedia.org/wiki/Throughput

Hardware architecture and (OS) software handling of the data did and still do not go hand in hand. Changes have been made to both to CPU handling, DMA memory loads, cache handling, transmit paths, zero-copies, packet scheduling, etc to allow networking performance to not only increase, but so as to not be as affected as much by the limitations of the CPU, memory, and operating system. http://www.nanogrids.org/jaidev/papers/ispass03.pdf

Networking is actually very hard on the system's hardware, matter of fact the smaller the packet is the harder it is on the hardware (if you did not read the paper). However, these additions and changes to hardware do not resolve the issues of networking, and because of such newer features are created to handle the problem of a problem. One example is the HW-NAT accelerator of the RT-N56U which does packet aggregation, TCP/UDP offloading, NAT translation offloading, and a couple of other features. Note the difference with acceleration on and then disabled when the 74K MIPS core (not a MIPS networking core) handles most of the data.

Without HW-NAT

Client connecting to 10.10.1.1, TCP port 5001
TCP window size: 87.5 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.1.10 port 44672 connected with 10.10.1.1 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 361 MBytes 303 Mbits/sec

With HW-NAT

Client connecting to 10.10.1.1, TCP port 5001
TCP window size: 87.5 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.1.10 port 44674 connected with 10.10.1.1 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 1.10 GBytes 943 Mbits/sec

Your tests are not showing the over all picture. You do not take into account the cost of functions that the router performs such as packet sequence inspection (SPI), multiple concurrent user usage, etc. You are taking a flow of data and sending as fast as possible to one simple point of the networking fabric. I am taking your data and formulating it to seem what it would be like for my networking infrastructure. This to me makes the version 2 router inefficient and horrible for the cost. It would be asinine to purchase this router until properly fixed. If a user wants 3 streams for wireless, it would be better to get another router when considering the price premium to performance ratio. Is this not the very reason for your website? So it surprises me that you are trying to refute this.

 

[thiggins]

Thanks for the references. Of course the throughput of any packet handling device depends on the factors you reference. And yes, my router throughput tests are only four simple views of a device's performance.

But my point still holds that when an ISP provisions X amount of bandwidth, there is nothing that a router can do to get more bandwidth from that service.

 

[Unregistered]

Thanks for the references. Of course the throughput of any packet handling device depends on the factors you reference. And yes, my router throughput tests are only four simple views of a device's performance.

But my point still holds that when an ISP provisions X amount of bandwidth, there is nothing that a router can do to get more bandwidth from that service.

Hi
One thing I notice no one taking into account or talking about, internal traffic. Am I wrong to assume this router would also be slower at internal traffic ?

Wouldn't it speed barrier slow multiple conection transfers on the internall side?

And last I am really eager to upgrade after trying the netgear, having wired speeds on wireless is so much easer than routing cables all the time.

So my last qwestion is can its speed be fixed threw firmware?

Thanks