WiFi speed experiment - Asus Eee Pad Transformer Prime

There have been quite a few threads on Prime's WiFi speed. However, I believe no-one has done a test, results of which I show below. I wanted to compare WiFi speed of the Prime and my Dell laptop. Speedtest.net measures internet speed, which is bound to fluctuate as server load and access time change. So, I decided to measure things locally. The methodology is as follows.
First, I created five text files with randomly generated ASCII characters. Each of these files was 50 Mb (52428800 bytes) in size. Then, I placed them on a local HTTP server. Since I don't have any program/app both on Windows/Android to precisely measure download time/speed, I used a simplified approach, consistent on both devices. I opened Chrome browser and measured how long it takes to download each file using a stopwatch. I did not ask Chrome to "save as" file, as it can start pre-loading the file before I confirm. I created a simple web page with links to all five files and clicked them one by one, waiting until entire file was open in the window. I simply watched the indicator in the tab, until it stopped spinning. By using five separate files I avoided problems with caching. In order to rule out any trend in our local server speed, I alternated the files, downloading the first file to the Dell and then to the Prime, then the second file and so on.
This is not the most accurate method, but sufficient enough for this purpose. I repeated the experiment five times, so I could estimate measurement errors. The results (download time, in seconds) are presented below:
Dell Latitude E6500 with Windows 7, 4 GB RAM
Transformer Prime, S/N C10KAS######, build .14
Dell Prime
----------
39.4 101.0
37.1 102.4
40.8 101.7
38.7 102.8
38.9 100.0
which converted into speeds (Mbytes / s) gives:
Dell Prime
----------
1.27 0.495
1.35 0.488
1.23 0.492
1.29 0.486
1.26 0.500
As you can see, there is not much variability in either sample, and the mean speed (with standard error) is 1.28(0.02) and 0.492(0.002) Mbytes per second, for Dell and Prime, respectively. Ergo, the Dell is about 2.6 times faster downloading (large) files than the Prime.
I leave the reason for this rather huge difference open to discussion. Is it the WiFi adapter in the Prime? Is it processor speed? Memory? Storage? I don't know. The fact is that download speeds on my Prime are much slower than on my Dell.
I haven't done experiment like this before .14 upgrade, so I have no idea how it looked like before.
Edit: In the follow-up to the discussion below I have done internet speed tests. This is all done on a fast university network, which shows speeds of up to 500 Mbps, when measured from a wired desktop. The result is interesting: both laptop and the Prime show similar speeds of about 2.3 Mbytes/s. This is much faster than local download speed. I can only guess that speedtest.net is streaming data and counting arriving packets, so it measures the actual WiFi speed. My download test involved opening the file in the browser, which requires lots of memory and some processing. Perhaps processing data in the browser takes a lot of time, hence the difference.

---

What about distance from the router?
What distances were these done at. What about varying the distance?
The Prime is known to have the Achilles heel of quickly dropping off throughput when you distance yourself from the router. I would be curious a) What distance these tests were done at, and b) to see what they are like at even further (or perhaps closer as well) distances.
No doubt about it though; the numbers you are getting from your Prime are not too swift. Question. Are you affected by any usability issues, such as buffering etc., when streaming video? How is overall performance?

---

SmartAs$Phone said:
What distances were these done at. What about varying the distance?
Click to expand...
Click to collapse
Both devices were sitting on the desk next to each other. The distance to the nearest router is about 3 meters. The router is on the ceiling with nothing in the line of site.

---

SmartAs$Phone said:
No doubt about it though; the numbers you are getting from your Prime are not too swift. Question. Are you affected by any usability issues, such as buffering etc., when streaming video? How is overall performance?
Click to expand...
Click to collapse
I don't have any particular usability issues. The tablet feels positively quick and responsive when using the interface, playing games and so on. However, web browsing is rather sluggish in compare with the Dell. Web pages that open almost instantaneously on the Dell can take quite a few seconds on the Prime. On the other hand, I can stream video clips, for example from BBC news, with no problems.

---

What kind of ping times to the router from each of the devices?

---

barryflanagan said:
What kind of ping times to the router from each of the devices?
Click to expand...
Click to collapse
Good question. I get about 2 ms on the Dell and about 8 ms from the Prime, to the local server. But these results are quite variable and I haven't done proper statistical processing.

---

You should specify the WNIC in laptop, rather than the laptop itself.
You're testing for the optimal case (close-range, no obstruction). A test of 10+ meters, ie one or two rooms away, with some obstructions, would be more representative. Rather than finding that the laptop performs better than the Prime, which is pretty much a given, it would be more informative to see how the Prime does with respect to distance & obstruction.
Since speed bottleneck typically involves video content (streaming) or large file transfer, it would be more accurate to larger non-compressible files for transfer, or streaming. A good test would be to stream 1080p, and gradually stretching the distance until drop-offs are noticed.

---

e.mote said:
You should specify the WNIC in laptop, rather than the laptop itself.
Click to expand...
Click to collapse
I'm not sure. Device manager says "Intel WiFi Link 5100 AGN".
e.mote said:
Since speed bottleneck typically involves video content (streaming) or large file transfer, it would be more accurate to larger non-compressible files for transfer, or streaming. A good test would be to stream 1080p, and gradually stretching the distance until drop-offs are noticed.
Click to expand...
Click to collapse
I'm not sure how to measure streaming speed (apart from spotting drop-offs, but that's subjective). I picked a simple method to measure file transfer speed on both machines under optimal conditions.

---

The 5100 is an older draft-N dual-band unit with a 1x2 (1 xmit, 2 rcv) arrangement, supposedly capable of 300Mb/s. Most laptop wifi are 1x2 or 2x2, with a few 3x3. In constrast, most mobiles are 1x1, with smaller antennas. So it's a given that laptops will have better wifi reception. You don't need to test for that. Test for something you don't already know about.

---

e.mote said:
The 5100 is an older draft-N dual-band unit with a 1x2 (1 xmit, 2 rcv) arrangement, supposedly capable of 300Mb/s. Most laptop wifi are 1x2 or 2x2, with a few 3x3. In constrast, most mobiles are 1x1, with smaller antennas. So it's a given that laptops will have better wifi reception. You don't need to test for that. Test for something you don't already know about.
Click to expand...
Click to collapse
I don't know what "1 xmit, 2rcv" is and how it affects transfer speed. If by "reception" you mean received signal strength, this is not an issue as I am sitting very close to the router and have a very strong signal both on the Prime and the laptop. And yet the tablet is much slower in file transfers.
Are you suggesting this is due to less capable WiFi component (antenna, adapter, whatever it is)?

---

1x2 = one antenna is used to transmit, and two to receive; it means max 150 Mb/s uplink and 300 Mb/s downlink (downlink speed matters more, as you'd expect). These are theoretical speeds, and real throughput is substantially less.
Antennas in mobile devices are smaller and less robust than those in laptops, and there usually is only one for wifi (hence 1x1), whereas in laptops you have two ants. Mobile devices also use SDIO interface that reportedly limits speed to 30-40 Mb/s.
I'm saying that wifi in mobile devices (phones, tablets) is inherently limited, and can't be expected to perform on par with laptops. Comparison against other mobiles, or against itself with varying distance, would be more fruitful.
In any case, it's an academic exercise, since you can't change out for a different antenna or wifi module. There was talk of the Prime being able to run Ethernet USB adapters, and I had asked for some to try out wifi USB adapters, but none responded. I take that to mean USB wifi isn't currently operable w/o a custom kernel.

---

I recently had 2 primes. They both showed equal signal strenght with wifi analyzer but speedtest.net shows 1 getting half the speed as the other. I returned that one even though it had a 90% working gps (was still off by 30+ ft and didnt track me well during turn by turn).
For sure its hardware (maybe pogo pins not making good contact or bad PCB) since I updated both to ICS .14 and it didn't affect speedtest results.

---

I have no server, but performed a very similar test.
I used WiFi File explorer PRO to transfer a large file from my computer. To do this I connected my laptop to the router with cable. I placed the Prime close to the router, something like half a meter. Result:
File size 726581248 bytes, time 298 seconds.
If I calculate well this means 19000 kbps or 2.325 MB/s. Not bad...

---

I was testing this exact method two nights ago. I'll do it again and post the results. I'll have three devices... evo4g, evo3d and my prime at a distance of about 30 feet. I'll even throw in laptop speed just for fun.
Sent from my Transformer Prime TF201 using Xparent Blue Tapatalk

---

I realize that you wanted to test your wireless speed, not your internet connection speed. The results you got on the Prime translate to roughly 4-5 Mbps.
Do you have a fast intenet connection (Cable Modem) that's capable of, say, 10+ Mbps?
What kind of speeds do you see on the Prime at speedtest.net? Assuming that your internet connection is not a bottleneck, I'd expect that you'll see something similar to your simple wireless tests. I'm particularly curious if you see significantly higher speeds at Speedtest. This wouldn't jive well with your wireless tests.

---

My speedtest.net results tops around 11000 kbps after running it several times. Unfortunately only in the same room with the router. This is OK since this is the maximum of my net.

---

ok here are my tests. I will start with my Asus G73jh laptop..
My Evo3D
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
My Evo4G
TFP C20KAS
Here is TFP right at the router, about 1.5 feet away
Like I said before, this is 30 feet away through a wall and a LCD tv. I have no problems with WiFi. I think I have a newer model, not sure.

---

e.mote said:
1x2 = one antenna is used to transmit, and two to receive; it means max 150 Mb/s uplink and 300 Mb/s downlink (downlink speed matters more, as you'd expect). These are theoretical speeds, and real throughput is substantially less.
Antennas in mobile devices are smaller and less robust than those in laptops, and there usually is only one for wifi (hence 1x1), whereas in laptops you have two ants. Mobile devices also use SDIO interface that reportedly limits speed to 30-40 Mb/s.
I'm saying that wifi in mobile devices (phones, tablets) is inherently limited, and can't be expected to perform on par with laptops. Comparison against other mobiles, or against itself with varying distance, would be more fruitful.
In any case, it's an academic exercise, since you can't change out for a different antenna or wifi module. There was talk of the Prime being able to run Ethernet USB adapters, and I had asked for some to try out wifi USB adapters, but none responded. I take that to mean USB wifi isn't currently operable w/o a custom kernel.
Click to expand...
Click to collapse
Well, I suppose it's more or less still messing with the kernel... But WiFi adapters have been made to work with the Prime.
http://forum.xda-developers.com/showthread.php?t=1484339

---

This is getting weird. As I mentioned in the original post, I intended to measure WiFi transfer speed, not internet speed. But since I have done it on a fast network at my work (university), I thought I might give an internet test a try.
Speedtest.net app shows speeds around 17-19 Mbps, which is about 2.3 Mbytes per second. This is over 4 times faster than my local download speed. This is rather unexpected, and unless speedtest.net is lying, there is an issue with our local transfers.
I believe speedtest.net streams down some data and measures it as it arrives. Perhaps streaming is done in a different way than downloading a big file? Or perhaps downloading requires further processing - storing in local storage, which might slow things down. I have just realized that I was actually opening a 50-Mb web page in the browser. This might be the real reason for much slower speeds of the tablet. I guess I would have to done the test differently, downloading a file directly into storage, though this would still be slowed down by the lack of processing power.
I have edited the original post to include these comments.
PS: At work, my wireless laptop shows a very stable 20 Mbps, and my desktop, which is wired, shows speeds varying widely from 50 to 500 Mbps.
PS2: I tried speedtest.net on the tablet again, standing directly under the WiFi repeater. This gives speeds almost 20 Mbps. I think our WiFi network is limited to 20 Mbps.

---

@ Felisek
You may try with WiFi File Explorer PRO. It seems there is a bottleneck somewhere in your test environment. This is just an idea.