Performance on wndr 4000

[ebec]

Am I missing something in the speed test section of the reviews? It seems that the maximum speed they get from the routers (not only the 4000) is way too low. After reading the tests of the latest routers I can’t believe the results.
I did a simple copy test from my desktop pc, just to see what speeds I could get. The desktop PC is connected on a gigabit LAN port. On both tests I have the laptop on the table at about 8 -10 feet away from the router. The router is behind a glass door in the TV bench.

Download:https://lh4.googleusercontent.com/-7zPjpn00g5Y/Tq7poiPzwfI/AAAAAAAACsk/Q9uw9mi8Ye0/s585/download.jpg

Upload:https://lh4.googleusercontent.com/-8AWIZV8N5og/Tq7pkksH6WI/AAAAAAAACsc/suomk3WwQVk/s663/upload.jpg

As you see the upload speed is much higher than the download, part of that is because before the upload test I “rotated” the laptop a few degrees to make sure I had a steady 450 mbit connection. The maximum download speed I see is about 200Mbit under optimal conditions.
Under less than optimal conditions 15 – 20 feet away around the corner of a brick wall (fireplace) I still get download speeds over 100 Mbit

The laptop is a Lenovo T420s with an I7 2620m CPU and Intel 6300 Wireless card connected on the 5GHz band

Anyone care to comment?

 

[overdrive31]

Sore way to test for speed, try again.

 

[ebec]

Sore way to test for speed, try again.

Yes probably, but how do you explain the SNB’s results. Do you agree with SNB that none of the new triple stream routers are capable of speeds over 100Mbit?
http://www.smallnetbuilder.com/lanwan/router-charts/bar/87-5-ghz-dn-40-3s

You cannot dismiss the fact that my actual wireless Download/upload speed from the laptop to the desktop PC is about 170/240Mbit.

Edit
Did a bit more search. This is more like what I would expect:

http://www.pcmag.com/article2/0,2817,2386682,00.asp#fbid=LURGaeFI7F_

 

[thiggins]

Thanks for your post.

The basic explanation is that our testing is focused on using consistent methods so that the data shown in the charts can be used for relative comparison. The method is described here.

Our test methods are not meant to show the maximum performance of any product. Manufacturers use any optimization they can get to come up with any performance data they publish. Optimizations include testing in RF chambers, running multiple streams, using selected test clients and using scripts with performance optimizations.

The high performance throughput IxChariot script uses a 1,000,000 byte file size, sets the send and receive TCP/IP buffer sizes to 64 KB and also uses Microsoft's Overlapped IO socket operation.

The standard throughput script we use, sets the file size to 100,000 bytes by default, uses "default" send / receive buffer sizes and does not use the MS overlapped IO method.

The script is edited only to adjust the file size so that the resolution of the plots shows short term variation. We usually start at 3,000,000 Bytes for high signal strength locations (A, C) and adjust it down in lower signal locations D and F).

The PC Mag article results seem very low for 2.4 GHz and somewhat high for 5 GHz. But the test setup descriptions don't specify bandwidth mode.

Your results seem somewhat high for uplink. But it could be due to the Microsoft file copy optimizations in Win 7. I'll run some similar tests and report back.

 

[ebec]

Thank you for the explanation

I guess people are looking for different characteristics when they decide on a product. Personally I bought the new router because I could not take advantage of my new 50/10 VDSL line. I also had problems when viewing RAW files (25 Mbyte) and streaming HD video from my DSLR over wireless.

Since the routers tested here are consumer products I think the tests should reflect actual use in a home environment like copy files, streaming video and surfing the web. I don’t know what kind of traffic your test software mimic, but it sure looks like its demanding. I guess it’s comparable to testing hard drives or SSD disks where you get a big difference when testing transfer speeds for sequential or random access.

 

[thiggins]

I agree that it would be nice to have more test data. But SmallNetBuilder already publishes more test data on products than any other publication. Adding more tests just isn't practical.

The one minute IxChariot tests we use are comparable to doing large file transfers or continuous streams. They send a continuous stream of large (1,000,000 Byte +) files for one minute as fast as the connection will support. Much different than web surfing, but more similar to large file transfers and HD streaming.

I'm glad that you're happy with your purchase and that it is providing you with the performance you need. You are fortunate because you are operating in a line-of-sight situation and can use the 5 GHz band. This provides the best chance for successful HD streaming.

In general, however, wireless, even 450 Mbps N, doesn't provide reliably steady high throughput for flawless HD streaming.