XPS 15 L521X WiFi Test Results

I spent a good 4 hour testing and compiling this data.

The Router used for testing was a Netgear WNDR4000. It is capable of up to 450Mbps theoretical wireless throughput. It is capable of up to 1Gbps theoretical wired throughput.

LAN Speed Test 1.1.7 was used to copy a file (100MB,250MB,500MB) to a networked PC that was attached via GigE cable to the WNDR4000. This effectively tests the wireless throughput and takes out extra variables such as internet and the bandwidth of speed test servers.

Net Meter 3.6 build 437 was used to provide graphs illustrating real-time upload/download throughput.

The network card configuration of each laptop was as follows:
XPS15: Intel Centrino Advanced-N 6235 (Max 300Mbps)
T420S: Intel Centrino Ultimate-N 6300 AGN (Max 450Mbps)

The environment was the following:
1. Office - 6ft - no walls
2. Dining Room - 40ft - two walls
3. Pool Patio - 60ft - Three/four walls

On-Battery Wireless Power Setting: Maximum Performance

No other WiFi networks were present ( > 80 RSSI).

Intel WiFi Drivers: 15.2.0.19

Laptops were never on at the same time when testing and other typical sources of wireless interference (wireless handsets, microwaves, large bodies of water, humans) were non-factors in this testing.

Here is a summary of results:

The Upload and Download speeds are the average throughput in Mbps over the course of the file transfer. LAN speeds are provided to prove that the speeds incurred via wireless testing were in fact limited by the wireless connection throughput.

The complete results are included in a WiFi.rar file here: http://sdrv.ms/MdcmkY

I have included the summary of results table as well as PNG screen captures of the throughput graphs and test results.

Please note that the T420S is equipped with a 3x3 card that is capable of up to 450Mbps and the XPS15 is equipped with a 2x2 card that is capable of up to 300Mbps. The T420S should outperform the XPS15 in all cases. The T420S is an ultra-portable business class laptop with a first class 3x3 card. I should have popped the 6235 out of the XPS15 and stuck it in the T420S and compare performance between the two machines that way, but I ran out of time. So this really gives you an example of what is some of the best WiFi throughput available today with a somewhat decent consumer router and a top of the line business laptop solution versus a top of the line consumer laptop solution.

I am pleased with the performance of the XPS15. It is not nearly as bad as people are complaining about. The trends in the data I saw were mostly what I expected. I used large file sizes to achieve average data points. I should have done multiple trials, but I only had one evening to work on this test. Either way, I think that these results are definitively more telling than ANYTHING ELSE I have seen on this forum in regards to the XPS15's wireless capabilities.

This tells me that the WiFi performance is not a deal-breaker and it is not a reason to not buy.

If you disagree, please first collect data at least as robust as what I have gathered and reference it in your post before claiming anything.

I could write up a couple pages worth of text analyzing the data, but I'd rather spend some time with my family that I'm visiting, so I'll let you nerds analyze it yourself with the skydrive link above.

The purpose of this post is to spread what I believe to be sound WiFi performance numbers that have nothing to do with anyone's shoddy internet or lack of knowledge regarding inter-networking technologies.

This thread shall not be used to bash the XPS15's WiFi performance. If you have data to back up a less positive picture, then please do so. SORRY, but SPEED TEST RESULTS are not valid (reliable) data in my book.

Please ask any questions you may have about my testing methodology, as I'm sure I missed some crucial information on here.

Thanks!

For Bill B.:
1. i7-3612QM
2. 8GB (2x4)
3. 640M 2GB (GK107)
4. System Disk: SSD (Intel 520 240GB) + Additional Storage mSATA 32GB that came with system
5. Tried card swap--non-negigible delta in performance. Currently uses the original 6235.

 

I appreciate your testing, but your conclusion that the wireless is not as bad only pertains to your case. There is an overwhelming majority that has issues, including myself. It is unforgivable that such a flaw can make it through testing. The fact that you have a Roll Royce router is masking the issues.

Video here, the L502X picks up 7 other wifi access points while the L521X only picks up 2. The RSSI is also considerably worse on the L521X and results in the speeds tanking. I have a 100Mpbs connection which is more than fast enough to stretch both laptops wireless capabilities.

Dell XPS L521X Wireless Test - YouTube

 

Not to be rude but these issues invalidate most of your conclusions:

1. Speedtest.net is being used. This is not a preferable way to test Wireless throughput or illustrate to the non-engineers on here. Speedtest.net bandwidth is spastic and unreliable at times.
2. The laptops are right next to each other and on at the same time. This day and age they shouldn't interfere with each other, but to be scientific about this, you would not want to do is going on in the video.
3. The larger numbers of WiFi networks listed by the L502X ARE UNUSABLE..they are like < -90 dB...they are useless. It's likely that the XPS15 card/drivers ignore anything under something like -80 something. However, this is my speculation only. But seriously, please do something useful with those WiFi networks that only show up on the L520X...I am very confident that you'd be lucky if you could even associate your laptop with the AP.
4. You run the speed tests at the same time. I assume that the speedtest leverages TCP connections (I could verify, but I'm not right now, so I'll assume). TCP connections should ideally use a "fair-share" each when competing for bandwidth, but this often doesn't work out in real-life testing.

So I'm not disputing that you wouldn't get similar speedtest results if you fixed all of the issues I listed above, but then again, speedtest is not the best way to test wireless throughput. Speedtests provide different results at different times. That is why in my OP I stated that speedtests are not valid data points in my book. Please provide test results that leverage non-WAN interface results in a more controllable setting with fewer variables that you cannot control.

 

I did state in the comments of the video it was not the most scientific. Speedtest is reliable so long as I choose the download centre that is in my town. What is important to me is that my downloads are fast. This is not the case with the L521X, they are terrible. Currently it is 2am in the morning, so I have no time for testing.

The fact remains the wireless signal and speeds are far more consistent with the L502X. It also has a far, far better RSSI resulting in much more consistent speeds. I appreciate your testing, but there is no way you are going to suggest to me the issues are down to my configuration. I also tried it with a Ubuntu live usb and it also resulted terrible speeds.

 

Have you even looked my test environment and my results. I did all of this work to specifically remove any "configuration" details having an impact. The absolute only thing that affects the throughput numbers in my testing is the laptop wireless solution, whereas in your speedtest results (and others on this forum) there are countless variables that affect your results. YOU CANNOT MAKE CONCLUSIONS based on your speedtest results. The only thing that you can conclude is that your setup won't work for you. We need more people to conduct more scientifically sound testing and less speedtest.

I can agree with you that the XPS15 compares less favorably to my T420S and your L502X, but you CANNOT make the conclusion that the lesser performance will provide a bad experience in general.

 

Your "I don't have a problem with my laptop, therefore yours doesn't have a problem" attitude is obnoxious. Stop telling people whether or not they should be satisfied with their laptop based on your individual results.

Using Speedtest.net is fine as long as you run the test a few times. If you consistently get worse results with one laptop in a certain spot, there's a problem, regardless of how many other variables there are.

In the owner's thread, there were plenty of people getting under 1 Mbps down just 30 ft from their router, while getting 10 or more with a different laptop. Others had significant trouble just downloading new wifi drivers in an attempt to fix the issue. There is a problem there. If there was truly no problem, then why does Dell have an entire team of engineers dedicated to finding a solution? Get your head out of the sand.

Clearly you're getting better results than other people. Just be happy with that.

I also have a problem with you telling people what they can and cannot post here, as if you were a mod.

 

The WiFi issue definitely deserved its own topic. Thanks for creating it, gfgrimm.

Risco, I understand downloads are very slow for you. What is basic internet browsing like for you? Going by gfgrimm's table, I assume basic web surfing works fine, even at a reasonable distance from the router. On the other hand, downloading/steaming high quality videos, which is what I'd like to be able to do, is probably painfully slow.

 

I have no WiFi concerns relative to my L502X and other computers (all over one year old) in my house. My son is frustrated with WiFi performance on his L521X in his apartment across town. The two times he visited me, bringing his brand new L521X, the situation did not improve for him.

I will not ride the high horse of advanced technology and scientific testing. I only say that much (and in lower case).

 

Nice job, gfgrimm.

I'm assuming all connections were 11n -- 11g can't go that fast on the 2.4 GHz band and there's no reason to test 11a on the 5 GHz.

The lower performance on the 5 Ghz at further distances is consistent with my expectations, but not the lower performance compared to 2.4 Ghz at 6 feet. I haven't run an actual data transfer test as you have, but in terms of theoretical max rate reported by my D-link router (based, I guess, on the signal strength), the 5 GHz band is 2 to 4 times faster after at near distances than the 2.4 Ghz band. Your results, however, show significantly better performance on the 2.4 GHz band at short distances.

 

Thanks for the info, it's nice to see such scientific testing. I believe people are noticing the slow speeds when browsing, like I did, and their first reaction is to use speedtest.net. I understand it's not the most accurate, but convenience is golden. That being said, I did a quick experiment using yours as a template. I have a ghetto router and not nearly as nice as internet as you, so I might fit the "general consumer" profile better. I had another laptop wired up to the router and used lan speed test 1.1.7 with a 100mb file size. I only did one trial because of a lack of time but will try more later. I also changed the on battery power setting to maximum performance.

1.5.1.1 Drivers:
5ft... On batt= 28.00 down, 39.24 up
Plugged= 18.55 down, 37.98 up

45ft (through a wall)... On batt= 12.03 down, 12.21 up
Plugged= 14.09 down, 17.02 up

1.5.2 Drivers:
45ft (through a wall)... On batt= 15.87 down, 17.75 up
Plugged= 14.81 down, 20.66 up

I understand that my results aren't completely reliable, but they are better than speedtest according to you. I will say that though it says 15ish Mb down and 20ish Mb up 45 ft away, videos will buffer when browsing at this distance. I don't know if these tests completely mute real-world experiences.