Eds Wifi Stick

[EvilDragon]

scolling through a page looks really bad sluggish and slow but i wonder if they have a proper video driver working yet as that could explain the tearing and slow speed when scrolling through the page.

I have a Touch Book (identical processor and SoC), and on it that happens only when when using finger/stylus to scroll. I think the drag-scrolling plugin for Midori is just a little inefficient. I assure you it's not a problem for Pandora as arrow buttons and page up/down are very responsive.

Same here. It's only the drag-scrolling.
You can see me use the buttons later in the video and that's kick-ass fast

 

[Jourdy288]

"[NAME REMOVED BY REQUEST] is an OMAP Field Application Engineer at Texas Instruments.In the Sales and Applications division, he works with customers to develop the latest smart phones based on TI's OMAP processors.He has 15 years experience focused on embedded software and architecture.Prior to joining TI in 2002, [NAME REMOVED BY REQUEST] was head of development at Web2U; a pioneer in SetTopBox technologies. "

from: [LINK REMOVED BY REQUEST]

EDIT: Removed pic, left link to Twitter account.

Is this the guy?

 

[Dead1nside]

Douche!

 

[mali]

^ That person is real so it might be better to remove his pic, just in case

 

[r00tw00t]

Booo!

edit: beer

 

[WizardStan]

Is the chip in the Pandora a 54Mbps chip? Also I thought someone said the chip uses the SD card BUS to do everything. So wouldn't this be a hardware issue or did I read that wrong. I will try and find the person and quote them but I am at work right now.

They don't know if it's a hardware or software issue. It appears that a large number of packets are being dropped. Off the top of my head, this could be either because there is interference which is causing corruption on the line (hardware) or possibly because the software isn't getting to the packets fast enough before they're dropped by the module (software). There are other scenarios as well, of course, and if I understand correctly, MWeston is wiring up a debug circuit so that we can figure out exactly where the packets are being dropped from.

 

[borgqueenx]

Mali, me and some other guys are gonna live forever. Our avatars where shown on "The" device.

 

[dyslexic]

Actuly i just thought about this and it doesn't add up to me ,right now 1000x wifi stick
are setting OP back £7 a stick and they are getting them at that price as they are buying in bulk
so the grand total is £7000 ,now that is quite alot of money for only 1000 customers
but couldnt that £7000 pay for a driver ? then all is fixed and even the first 1000 customers
will be more happy to get working wifi.

I mean where did this chip come from ,what other devices use the same chip on an arm OS ?
there must be a driver out there that should work ? i do also understand that while we wait
OP are testing the chip for a hardware issue and that in the mean time it would be pointless
to aquire such a driver at cost ifthe hardware is to fault ,but then again if the hardware
is to fault in some way ,then can't the chip just be reordered in a working fashion.

We already wasted thousands of pounds on a 'wifi expert' called [NAME REMOVED BY REQUEST], an ex-ti employee and we got nothing.

I remember that being mentioned.
The thing what i am amazed about is that you thought that the ex-ti fellow did not do a good job while now it may be hardware.
Didn't you think back then to check the hardware if it was bullet proof.
You should have don the test then on the hardware what you are going to do now.
I know it may still be software but the hardware should have had all the possible checks back then.
Yet again someone else got used as scapegoat while OP neglected to do all the needed test.
Now there are more delays.Just for ones it would be nice to hear you say "well we came short sorry"
And if the name you gave here is a real name than that is a show of bad manner.

 

[Alpha2]

Sometimes it's not possible to do a test for a problem unless you know there's a problem. Not everything is caused by neglect. Shit happens.

 

[WizardStan]

I remember that being mentioned.
The thing what i am amazed about is that you thought that the ex-ti fellow did not do a good job while now it may be hardware.
Didn't you think back then to check the hardware if it was bullet proof.
You should have don the test then on the hardware what you are going to do now.
I know it may still be software but the hardware should have had all the possible checks back then.
Yet again someone else got used as scapegoat while OP neglected to do all the needed test.
Now there are more delays.Just for ones it would be nice to hear you say "well we came short sorry"
And if the name you gave here is a real name than that is a show of bad manner.

You can design a car from scratch, build and test the engine and prove that it runs, drop it into your car, and still find out that the fuel line is insufficient to drive the engine with enough horses. They did test the wifi chip to the bare metal, as far as they could tell it was running well. It wasn't until everything else came in around it that they were able to see it wasn't performing as well as they had hoped.

 

[Lorkile]

Not to mention they are handling this pretty well, though I would have greatly preferred some sort of decision as to what's going to happen and a blog post for clarification.

 

[WizardStan]

Not to mention they are handling this pretty well, though I would have greatly preferred some sort of decision as to what's going to happen and a blog post for clarification.

ED said he wrote a blog post, but they don't want to upload it until MWeston has had his chance to establish whether it is hardware or not. They're just speculating what the problem is right now, and it does no one any good to post speculation on the official blog.

 

[Zotty]

You can design a car from scratch, build and test the engine and prove that it runs, drop it into your car, and still find out that the fuel line is insufficient to drive the engine with enough horses. They did test the wifi chip to the bare metal, as far as they could tell it was running well. It wasn't until everything else came in around it that they were able to see it wasn't performing as well as they had hoped.

So bare metal testing wifi performed at full speed? If so a software problem can be ruled out. If not, the tests are flawed. Given the fact that throughput is an essential specification for a network device, this should be one of the first requirements to test.

Reading MWeston is developing a debug PCB or something like that afterwards is another indication testing procedures failed. Now on itself this is not bad, we all learn new things on a daily basis. But the thing is that with a bunch of Pandora's already out there the question rises of why this hasn't been picked up earlier? Why this late in the process? It's called Open-Pandora, so be open and honest about this stuff.

 

[craigix]

Actuly i just thought about this and it doesn't add up to me ,right now 1000x wifi stick
are setting OP back £7 a stick and they are getting them at that price as they are buying in bulk
so the grand total is £7000 ,now that is quite alot of money for only 1000 customers
but couldnt that £7000 pay for a driver ? then all is fixed and even the first 1000 customers
will be more happy to get working wifi.

I mean where did this chip come from ,what other devices use the same chip on an arm OS ?
there must be a driver out there that should work ? i do also understand that while we wait
OP are testing the chip for a hardware issue and that in the mean time it would be pointless
to aquire such a driver at cost ifthe hardware is to fault ,but then again if the hardware
is to fault in some way ,then can't the chip just be reordered in a working fashion.

We already wasted thousands of pounds on a 'wifi expert' called [NAME REMOVED BY REQUEST], an ex-ti employee and we got nothing.

I remember that being mentioned.
The thing what i am amazed about is that you thought that the ex-ti fellow did not do a good job while now it may be hardware.
Didn't you think back then to check the hardware if it was bullet proof.
You should have don the test then on the hardware what you are going to do now.
I know it may still be software but the hardware should have had all the possible checks back then.
Yet again someone else got used as scapegoat while OP neglected to do all the needed test.
Now there are more delays.Just for ones it would be nice to hear you say "well we came short sorry"
And if the name you gave here is a real name than that is a show of bad manner.

Unfortunately you have no idea what you are talking about.

 

[kingoddball]

^^^^

Hahahahahaha!

 

[WizardStan]

So bare metal testing wifi performed at full speed?

I didn't say that, and I don't think MWeston did either. As far as I know, they tested it as much as they could without the OS, but there's only so much you can test with an engine up on a block. Put the engine in the car, and suddenly there's a whole lot of other things going on that are impossible to gauge until they're actually there.
I don't see where you're going with your "be open and honest about this stuff". They didn't, at any point in time, hide the fact that wifi wasn't working. We've known for a long time that wifi wasn't working. A few people offered up their assistance. Craigix posted about their problems with hiring a "professional" months ago. They have been entirely open and honest. Despite all the effort, they still don't know for certain what the problem is. What do you suppose they could have possibly said that would have helped the situation any?

 

[Zotty]

So bare metal testing wifi performed at full speed?

I didn't say that, and I don't think MWeston did either. As far as I know, they tested it as much as they could without the OS, but there's only so much you can test with an engine up on a block. Put the engine in the car, and suddenly there's a whole lot of other things going on that are impossible to gauge until they're actually there.

Very true, but I would assume the wifi chip was tested isolated before putting it on the final PCB. Just to see if the thing works without any chance of other peripherals interfering. That way we could have avoided the whole software/hardware question in the first place.

I don't see where you're going with your "be open and honest about this stuff". They didn't, at any point in time, hide the fact that wifi wasn't working. We've known for a long time that wifi wasn't working. A few people offered up their assistance. Craigix posted about their problems with hiring a "professional" months ago. They have been entirely open and honest. Despite all the effort, they still don't know for certain what the problem is. What do you suppose they could have possibly said that would have helped the situation any?

I must admit I've only recently returned to these forums, so I didn't know that. Apologies for the assumption, you are spot on as to what I meant to say.
Concerning the openness, had lot's of words to type, but your 'history lesson' kinda shot those out of the water, so I'll send those to my internal /dev/null

 

[Dead1nside]

There really wasn't any reason to assume Wifi wouldn't work. It was recommended to them by TI, I guess they saw that there was a semi-open driver, working on an in-production device. DJWillis and Notaz have seemingly been working on this for ages and they did hire an ex-TI guy to help them out, which didn't turn out well at all (shame on you). Now they're offering a win-win situation to us all, with regards to wifi stick now, fix later. That's pretty damn open and great of OP, so thank you.

 

[conso]

Zotty, what are you talking about? This isn't about TI having send out a batch of defect chips. Even if that was true, how would you test that "isolated"? If there is a hardware problem, it's about how the chip is connected.

 

[Zotty]

I used to do a lot with those small colour LCD displays you find in mobile phones for example. Datasheets weren't always helpfull, so before building the complete design I took a display and hooked it up to something that could provide data for it. Then I started sending stuff to it, testing various commands and how much data I could throw at it before the controller's limits were reached. Also tested differences in wiring and how it would affect performance. Next a driver was build and only then did I start finalising the design. It's not 100% proof, but at least I knew the hardware was performing as it should be and the drivercode was functional.
Now this is not completely comparable with the Pandora and wifi since that's much more complicated, but I hope it clarifies what I mean with isolated testing. Other factors contribute to the actual performance in the final product, no illusions there. But in case of the Pandora you would large know if the driver is the problem or not. If it doesn't perform anymore in the final design, you can probably skip looking at driver code such as setting registers, processing packets, etc and focus on other software/hardware interacting with wifi parts.