Mc3research

I'll give terra a try next time I'm working on the iPad... I'll have to see this >50MB crash for myself.

Before you get around to test, D/L Logviewer on the iPad. This will give you some degree of info on what happened. (It's also nice to have some minimal glympse inside the Walled Garden.)

:::.. Download Test Results ..:::

Download Connection:: 18824 Kbps or 18.4 Mbps

Download Test Size:: 50 MB or 51200 kB or 52428800 bytes

Download Speed:: 2353 kB/s or 2.3 MB/s

Tested At:: http://TestMy.net version:12.3

Validation Link:: https://testmy.net/db/Cl746XE

Test Time:: 2012-03-02 13:39:51 Local Time

1MB Download in 0.44 Seconds - 1GB Download in ~8 Minutes - 336X faster than 56K

This test of exactly 51200 kB took 21.76 seconds to complete

Running at % of hosts average (Suddenlink Communications)

User Agent:: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:10.0.2) Gecko/20100101 Firefox/10.0.2 [!]

So here's the real kicker: With all other network config the same, the iPad is consistently getting 50% BETTER throughput than my MacBook Pro.

And a sidenote TestMy.Net bug report:

If your logs show (and I think they do) a 257Mbps result, it's mine and it's an error. It was inadvertently produced by PageBack over the DL Test Url. Delete the result so it doesn't skew the integrity of the test result history.

Who is the ISP?

If I recall correctly, broadbandreports.com has some speed monitoring tools that might help you document and trend line the slowdown.

I don't have as much experience with DSL modems as cable modems, but most likely the modem has an error log that would be interesting to look through.

... But I was cheating because it wasn't really on an iPad. Rather my iMac running xcode's iPad preview. It really feels like an iPad though... how it loads. Best I've seen on my network with a real iPad was about 30Mbps... actually just under 30Mbps like you had. Maybe that's the limitations of the real device? I haven't tested them enough myself to be sure... maybe you can provide more iPad insight.

-D

I'm just guessing that the I/O speed of a low power consumption 802.11 chipset may be the limiting factor. I also suspect that if the file was being written to disk, the throughput would have been much lower. It is, however, fast enough to demonstrate that with the right server and network config I can deliver 20 channels of SD and 10 channels of HD to multiple 802.11n devices in range. (more if I use 5Ghz and 2.4Ghz simultaneously.). But this is pushing the limit of staying on topic for this forum....

Mike

Across several tests, I've achieved 25 - 29 Mbps in the following Topology:

Browser = Terra 1.3.6

OS = IOS 5.0.1

Device = iPad2 CDMA, 64GB, dual core 1Ghz Apple A5

802.11n = 5Ghz 48(P) 44(S) @ < 10m, indoor, NLOS

Access Point. = Netgear 3700 v1.0.7.98NA (dual band with Atheros 680Mhz RISC NWP)

LAN = 1Gbps, CAT6 (1m)

CPE = Motorola Surfboard 6120 v1.0.1.8-SCM02-NOSH

ISP. = Suddenlink.net, 50Mbps/2Mbps nominal and verified

avg latency. = 59.7ms (52B null packets at 10 pings/s)

Traceroute to mytest.net, 64 hops max, 52 byte packets:

 hop(4) 36 bytes  cdm-66-76-30-37.tylrtx.suddenlink.net (66.76.30.37)

 hop(5) 36 bytes  64.211.195.141

 hop(6) 36 bytes  ae2.csr2.dal2.gblx.net (67.16.151.82)

 hop(7) 36 bytes  po9.ar5.dal2.gblx.net (67.16.151.133)

 hop( 36 bytes  verizon-2.ar5.dal2.gblx.net (64.208.27.1)

 hop(9) 36 bytes  0.ae1.xl3.dfw7.alter.net (152.63.96.46)

 hop(10) 36 bytes  0.ge-7-3-0.xl3.mia4.alter.net (152.63.1.110)

 hop(11) 36 bytes  tengige0-6-4-0.gw11.mia4.alter.net (152.63.84.130)

 hop(12) 36 bytes  internet-gw.customer.alter.net (152.179.26.218)

 hop(13) ********************  Request timeout

 hop(14) 36 bytes  24.244.158.2

 hop(15) 64 bytes  208.87.33.151

Best results are obtained with large file sizes (>= 25MB)

Observations:

1. Terra appears to be significantly faster than Safari (which is interesting since they share a lot of the same programming in xCode)?

2. Terra appears to have a bug that causes a crash on large test files (> 50MB): around 53-58MB into the test, Terra will crash)

3. The tailhop from Potwin to Ashburn 152.179.26.218 to Nassau, Bahamas 208.87.33.151is a little odd, and I'm assuming DNS is pointing http traffic to a different server.