Pandora Wi-Fi On 350kilobytes/s


Status
Not open for further replies.
DroneB Dev said:
EvilDragon said:
Current speed Michael gets while downloading off the net:
200 - 300 KB/s on NORMAL image!
Not the minimal WiFi image :)

That's 2.4MBps :)

thats more than enough for my neighborhood, ahh, those old days that we only had 802.11 a

im more worried about range (tx/rx) performance, i do really hate chips so low TX power that even sitting under the router still have some issues.

That's typically the Download speed I get when downloading from websites and I'm connected directly to the router by gigabit Ethernet. This isn't really a true test of the Wifi IMO.. like others stated a large file downloaded from a Samba or Web server on a local LAN would be a truer test to see how fast it would go.
 
Last edited by a moderator:
Now THAT is what I call AWESOME NEWS!!! :D

Simple technical curiosity since I've been doing electronics as a hobby for the past 15 years.
-What was the value of the resistor before the mod and what is it now that it's modded?
-What caused the big difference in speed between the NORMAL image, and the minimal WiFi image? (cpu requirement? tcp/ip settings?)
 
EvilDragon said:
Current speed Michael gets while downloading off the net:
200 - 300 KB/s on NORMAL image!
Not the minimal WiFi image :)

That's 2.4MBps 1.6 - 2.4 Mbps :)
So 200-300 K(ilo)B(ytes) / s is a respectable but not fantastic 0.2-0.3 M(ega)B(ytes) / s, or 1.6-2.4 M(ega)b(its) / s.
 
Last edited by a moderator:
TrashyMG said:
DroneB Dev said:
EvilDragon said:
Current speed Michael gets while downloading off the net:
200 - 300 KB/s on NORMAL image!
Not the minimal WiFi image :)

That's 2.4MBps :)

thats more than enough for my neighborhood, ahh, those old days that we only had 802.11 a

im more worried about range (tx/rx) performance, i do really hate chips so low TX power that even sitting under the router still have some issues.

That's typically the Download speed I get when downloading from websites and I'm connected directly to the router by gigabit Ethernet. This isn't really a true test of the Wifi IMO.. like others stated a large file downloaded from a Samba or Web server on a local LAN would be a truer test to see how fast it would go.

i never said dont do the LAN testing, i just said people complain too much about it, just some years ago all we had was 802.11a protocol.
we also have to check how much can the processor deal with the data I/O from all sources (wifi 2 SD).

also its a big improvement against the less than 56k panda had some days ago.

and as a last statement, i doubt you would really need that much of network speed on the panda, besides media streaming,2.4MBps well we are close enough to a useable speed for video streaming, and if you want more, well, right now it might not be possible since the panda cant deal with heavy formats until they get hardware acceleration on it.
 
Last edited by a moderator:
MWeston said:
P.S. Yes, transferring a file from my PC would be smarter but I have to figure out how to do that still.  I normally turn anything like that off for security reasons.  I don't think I've done file sharing since the headache-y days of Win98!
Python makes it easy.

Code:
$ cd /home/somedir
$ python -m SimpleHTTPServer


It will start an HTTP server, serving files from the current directory, on port 8000. This should also work if you use Windows, as long as Python is installed.
 
Last edited by a moderator:
I have been learning and programming some simple python on and off. Didn't know about SimpleHTTPServer. It's real nice.

Right now I reading a book on drupal... I know it's totally off topic.

It's great to see the wifi working ok :)
 
Pickle said:
In my scp tests ive avg between 400-500 KB/s
WPA2 works

Awesome! It still amazes me that one tiny resistor can make such a huge difference. Tells you a lot about how much clock speeds matter.
 
Last edited by a moderator:
DroneB Dev said:
im more worried about range (tx/rx) performance, i do really hate chips so low TX power that even sitting under the router still have some issues.
This interests me too. I live in a concrete apartment building. I could walk the floors and stairwells with no interruption to my internet radio. I went for a walk and got four houses down the block before it wasn't able to keep the buffer loaded enough to play the stream.

It also played fine whether the lid was open or completely closed and with the unit in my hand (although performance does degrade on high speed stuff if you cover the antenna with your hand).

amf66 said:
Awesome! It still amazes me that one tiny resistor can make such a huge difference. Tells you a lot about how much clock speeds matter.
The resistor was the series resistance in the oscillator driver. I did change it once on a previous revision but since there were no complaints back in the day, I kept it larger for lower current consumption. The wifi clock requires a very strict clock profile so I had to decrease this resistance to drive the crystal harder (larger voltage swing), increase the bandwidth and put the crystal into a tighter range around resonance. It probably means an extra mA or two of constant current drain so not a big issue in that respect.
 
Last edited by a moderator:
I just installed Arora on the Pandora.
Pretty nice, fast browser, no rendering errors, no crashes :)

Maybe we should replace Midori with it, at least for first release, as Midori loves to crash (even on my desktop PC...)
 
EvilDragon said:
I just installed Arora on the Pandora.
Pretty nice, fast browser, no rendering errors, no crashes :)

Maybe we should replace Midori with it, at least for first release, as Midori loves to crash (even on my desktop PC...)
So I take it you're already including the whole of Qt in the standard image?
 
Last edited by a moderator:
EvilDragon said:
I just installed Arora on the Pandora.
Pretty nice, fast browser, no rendering errors, no crashes :)

Maybe we should replace Midori with it, at least for first release, as Midori loves to crash (even on my desktop PC...)
Same here, it pisses me off way too often.
 
Last edited by a moderator:
I dont understand why we cant get som reliable iperf numbers instead of downloading files from internet or via scp or something.

iperf is a simple commandline tool for measuring tcp throughput. It exists on multiple platforms, and should be easy enough to compile.

On server:
> iperf -s

On client:
> iperf -c hostcomputer


It is easier to optimize something when you got reliable numbers to optimize with.
 
macke said:
I dont understand why we cant get som reliable iperf numbers instead of downloading files from internet or via scp or something.

iperf is a simple commandline tool for measuring tcp throughput. It exists on multiple platforms, and should be easy enough to compile.
Consider the last word in the quote above. It starts with a 'c' :p
 
Last edited by a moderator:
macke said:
I dont understand why we cant get som reliable iperf numbers instead of downloading files from internet or via scp or something.

iperf is a simple commandline tool for measuring tcp throughput. It exists on multiple platforms, and should be easy enough to compile.

On server:
> iperf -s

On client:
> iperf -c hostcomputer


It is easier to optimize something when you got reliable numbers to optimize with.
iperf doesn't give reliable results with this chip. For example, if doing simple wget shows ~300kB/s, doing iperf says it's < 1Mbit/s. It seems starting the transfer is slightly delayed what confuses iperf.
 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top