PandaBoard!?

[gadgetoid]

WTF!?


http://pandaboard.org/


I know the project isn't officially tied to the "Panda" moniker, but it's been a term of endearment to our elusive (for most of us anyway) handheld for (almost) years. This hits disturbingly close. And as soon as I heard the name I'd thought the OP team had something to do with it. Clearly not, though.

 

[EvilDragon]

Yeah, we heard about this a while ago


Nice one, eh?

 

[Aurekana]

Too bad the OMAP4430 SoC isn't available for general ordering, otherwise someone would probably find a way to mod a Pandora to make use of it... <_<


Referring in part to this thread: http://www.gp32x.de/board/index.php?/topic/56579-cpu-for-pandora-2/page__view__findpost__p__917150


In which several people are evaluating the performance difference between the OMAP3530 (SoC the Pandora is shipping with) and the AM3730 (in the beagleboard-xM). These two are apparently mostly pin compatible, so transitioning would mostly be challenged by desoldering the BGA chip.


I somewhat doubt the OMAP4430 will be anywhere remotely close to pin compatible with the OMAP3530, though it is fun to dream...

 

[Bosbeetle]

I just imagine mweston will always be soldering something, somewhere!

 

[lulzfish]

I'm loving the "entire website laid out as a single bitmap image" thing going on there, reminds me of that one Chinese / Korean website. (Can't remember which)


I don't really want to buy a dev board, though. Especially not from someone who can't make a website.

 

[neko]

I'm loving the "entire website laid out as a single bitmap image" thing going on there, reminds me of that one Chinese / Korean website. (Can't remember which)

And it's a JPEG, making the text look fuzzy.

I don't really want to buy a dev board, though. Especially not from someone who can't make a website.

It's from some of the same people who did the BeagleBoard, right? Anyway, the electrical engineers probably didn't make the website.

 

[EvilDragon]

It's from some of the same people who did the BeagleBoard, right? Anyway, the electrical engineers probably didn't make the website.

I don't think so - but that would explain the name as the company behind the beagleboard is the same that produces the Pandora boards

 

[craigix]

It isn't the people who made the Beagle board. I think it is TIs wifi division who have made that board. When it will actually be available is anyones guess.


That chip is crazy complex.

 

[Chairman_Now]

It's from some of the same people who did the BeagleBoard, right? Anyway, the electrical engineers probably didn't make the website.

I had a chat with a Mr. Gerald Coley (he is one of the primaries of the Beagleboard project), who was quite adamant that there would be no OMAP4 Beagleboard

 

[neko]

I had a chat with a Mr. Gerald Coley (he is one of the primaries of the Beagleboard project), who was quite adamant that there would be no OMAP4 Beagleboard

Ah.. so TI went ahead and designed their own beagleboard-like thing for the OMAP4.

 

[Chairman_Now]

Ah.. so TI went ahead and designed their own beagleboard-like thing for the OMAP4.

"Like" is the operative word. The Beagleboard is truly Open Hardware. The Panda will be partially under NDA.


I also met the CEO of Circuitco. I'll keep that conversation to myself

 

[neko]

I think it is TIs wifi division who have made that board.

Maybe they will get the wifi to actually work... I'm not thrilled with the 5% packet loss on my pandora.


What's different between the WL1251 and the WL1271 anyway?

 

[WizardStan]

I'm not thrilled with the 5% packet loss on my pandora.

5% is extremely high. I don't see any packet loss. How did you test that?

 

[neko]

5% is extremely high. I don't see any packet loss. How did you test that?

Ran ping with the ip address of my router. With the default packet size, the loss is around 1%, but with -s 1500 it's up to 5%.


The problem of course is that loading web pages or downloading files uses 1500-byte packets, and these are much more likely to get lost. Lowering the mtu helps, but it still drops packets and causes the connections to freeze.

 

[WizardStan]

That's really weird, because I've previously asked others with such slow connections to do the same test and they've reported no packet loss: it was just really slow. That's what led to the belief that it was definitely something software related, not hardware.


Your slowness may be because of excessive packet loss, not because of any potential wifi/router compatibility, and that packet loss may be the result of the modded resistor not functioning correctly. Heavy packet loss was the symptom experienced before the new resistor went on, maybe the tolerance isn't as narrow as originally thought and some (like yours) needed an even higher resistance.


Or maybe it is a driver issue manifesting in a slightly different way, it's hard to say. Just to be on the safe side, you may want to bookmark this thread, in case you want to refer to it if it becomes necessary to RMA the unit.


Maybe MWeston can shed some light on it. He did the resistor fix so should know whether this could be it and how we might test for that.

 

[Exophase]

I had pretty severe packet loss due to wifi interference. It went away when I changed wifi channel.


If this is anything of a general problem - and I fear it will be - applications that are not bandwidth intensive but require latency as low as possible (like games) are going to have to use UDP and some other form of redundancy to deal with the problem. When using TCP a dropped packet causes a huge latency spike.

 

[neko]

I had pretty severe packet loss due to wifi interference. It went away when I changed wifi channel.


If this is anything of a general problem - and I fear it will be - applications that are not bandwidth intensive but require latency as low as possible (like games) are going to have to use UDP and some other form of redundancy to deal with the problem. When using TCP a dropped packet causes a huge latency spike.

ping -s 1500 -c 1000 192.168.1.1


From my laptop (Atheros AR5007EG):

1000 packets transmitted, 998 packets received, 0% packet loss

rtt min/avg/max/mdev = 1.962/4.169/36.932/3.609 ms

From my pandora:

1000 packets transmitted, 972 packets received, 2% packet loss

rtt min/avg/max/mdev = 23.529/58.088/312.714/14.761 ms

Both were within 10 feet of the router. Packet loss is much higher on the Pandora. Interference might be a factor, but the packet loss is way too high considering how close it was to the router.

 

[WizardStan]

turn off power saving?


sudo iwconfig wlan0 power off

 

[neko]

turn off power saving?


sudo iwconfig wlan0 power off

1000 packets transmitted, 995 packets received, 0% packet loss

rtt min/avg/max/mdev = 7.965/13.915/96.039/7.209 ms

Much better... The powersave more should be off by default, it makes the wifi nearly useless.

 

[WizardStan]

You may also want to ask ED about the new beta wifi driver. It made even power save about as stable as no power save in my testing. Pings were still fairly high, but I think that's expected behavior with the power saving feature.