OMAP4440 Dual Core


But I also read his comment as using the DM3730 for the Pandora2 not just an incremental update for the current Pandora. I don't know how cutting edge the OP team has to be to make sales though as it is the only device that is open, has a keyboard. and has gaming controls.
I remember Craig wanted to used an iMXnn chipset (one with ARM9 or 11) for Pandora before OMAP3 with Cortex-A8 was suggested instead. A lot of competent people dissuaded him from doing such a thing because it was a very poor way to be competitive against the other handhelds. Using a DM3730 for Pandora2 would be exactly the same "error". Now that ARM is making a move toward multi-core, there is no point to make a second generation (not just an incremental update) with one mono-core which will be considered as obsolete at the release time.


I'm kinda fed up with this long wait for having this Pandora I bought, which is getting obsolete now considering the power hardware you can have on actual or future tablets (supposedly this two-month is not turning into two-years again...). The only feature which tells me Pandora is different from the others is about its gaming controls (I'm not even considering its keyboard as it is not a requirement for me), but I don't think it will be enough for a second generation to keep me interested with.
 
Last edited by a moderator:
I believe he's suggesting that right now, for future batches of the current Pandora, they could migrate to the DM3730 and save money. It's more or less the same as the existing OMAP3, I think; to the end user it would be an invisible switch, a kind of Pandora v1.5.

Except for its normal frequency : from 600 MHz to 800/1000 MHz. I'm not even speaking about overclocking potentiality. Developers may attempt to use this extra frequency and had the owners of old Pandoras pissed off because more and more applications will be slow because of it.
 
Last edited by a moderator:
Except for its normal frequency : from 600 MHz to 800/1000 MHz. I'm not even speaking about overclocking potentiality. Developers may attempt to use this extra frequency and had the owners of old Pandoras pissed off because most applications will be slow because of it.
Yeah, that would definitely be the best way to piss off the customers that actually made the Pandora project possible. Aside from this, if they publicly announce such a move there might be just enough cancellations for the team to go bankrupt.
 
heh? do you mean if there's a next generation of Pandora, OP would probably migrate on DM3730 instead of a OMAP4 ? if so, OP would really suck. The situation would be similar to customers who buy an old iPhone and are pissed off because now they need an iPhone 3GS or 4G to run all the hungry applications or games. I started regretting buying Pandora (Cortex-A8 being just a mono-core with minor bugs, all the stuff with SGX not optimized, what about the complicated TI DSP stuff ? etc.). OMAP4 is at least a dual-core (which is a minimal requirement for me and has a more refined NEON instructions set).


So, if a next generation should appear, it must be a complete different beast so it makes a sense to buy it even if we have the first generation to be sure the development on the first gen shall not be jeopardized by the second gen in a similar way it is for iPhone generations.

No, I'm saying that OP should be using DM3730 ASAP without any other design changes. The second batch should be DM3730. They shouldn't keep selling OMAP3530 Pandoras. It doesn't make sense. Customers being pissed off is just something they should swallow; it's the customers who have waited over 2 years for their Pandoras that should be pissed off. It sucks for the remainder of batch 1, but that's OP's fault for selling far more than they could supply in a reasonable timeframe.


You guys talk about developers making things that "require" the faster Pandoras, and that strikes me as nonsense. You don't make something that needs a clock speed to run, and I doubt anyone is going to slack off on something they wouldn't have otherwise because they have more MHz now. No, almost all of Pandora's software will consist of ports without a lot of optimization, and most of it will be fine at 500MHz. These faster Pandoras would allow some people to run a fringe selection of software better - high end emulators would be less slow and intensive games would run at a higher framerate. It's really not an awful lot different than some people being able to overclock higher. It's not a matter of compatibility.


But yes, they shouldn't actually consider this until all their orders are satisfied because otherwise a lot of people might cancel to get the better one. Yeah they might be pissed off, but I don't understand how improving the hardware on newer iterations (something companies do all the time) is unacceptable but taking years to fulfill an order is fine. It's a bad situation and OP has to make some compromises.


In return OP would actually be saving money because the 800MHz DM3730 chip is quite a bit cheaper than even the 600MHz OMAP3530. Obviously they wouldn't go in for the 1GHz one.


I want dual core like you, but let's be realistic, right now there isn't a single dual core Cortex-A9 device on the market in anywhere near Pandora's form factor. Tegra 2 is also not really an option, and I doubt OMAP4 is really ready for even Pandora-like volumes just yet. But I absolutely think that there should be a Pandora 2 with OMAP4430 in under 2 years. I think that OMAP4440 is probably on a less power optimized process and not good for mobiles, it strikes me as a tablet processor.
 
Last edited by a moderator:
Say, how long did it really take from starting to develop the Pandora until the first finished device got send out? It was way more than just 2 years. The big ordering day was not the start, and it's not where we would start when creating a OMAP 4 device. Now take a look at the OMAP 4, it is some serious high-tech chip which needs a really complex board. It will take much longer to design the board and due to its complexity it will need another company with more specialised machines. The Panda Board has 8 layers! It all adds significant delays even before we reach the same point where we were in October 2008. And from what we know now, once that point is reached it will take much longer than 2 months until we're ready to go, even without an angry Murphy.

Yeah, Pandora took forever to come out. How much of that had to do with board design? The board was done in a few months. Most of the delay was due to working with cheap companies that failed to meet their promised target dates by several months, banking issues, and OP themselves getting the software done. The board manufacturers are slow because OP is not paying for fast production, if they had more money they could get them out much quicker. But that might involve raising funds from more than taking advance preorders.


Any talk of future work aside, OP really needs higher board production capabilities NOW to meet their CURRENT outstanding order. People shout praises every time the board manufacturers pump out a few hundred more after taking several weeks. How long do you think it'll take to get the remainder of the 4000 out?


I dunno what you're talking about with OMAP4 being seriously high tech this and that, Pandaboard has a bunch of peripherals that Beagleboard and Pandora don't so it's not an apples to apples comparison. The SoC itself doesn't require a bunch more support circuitry just to work, afaik the power rails are similar to OMAP3, so why do you think it needs such a complex board? Granted, a Pandora 2 shouldn't be a Pandora 1 with a better SoC, I think an HDMI out is a must, but I say they can just go ahead and replace the S-Video with that and that'd be a good enough upgrade. Why do you think another company is necessary? Do you think 8 layers is some extreme number? I'll ask MWeston but I'm pretty sure Pandora's board is itself 8 layers. Don't you think you're making an awful lot of assumptions?
 
Last edited by a moderator:
gerber.jpg



There you go, 8 layers.


EDIT: Source: http://pandorapress.net/2010/09/15/baking-packing-and-er-gerbering/
 
Last edited by a moderator:
Yeah, Pandora took forever to come out. How much of that had to do with board design? The board was done in a few months. Most of the delay was due to working with cheap companies that failed to meet their promised target dates by several months, banking issues, and OP themselves getting the software done.
Well, assuming the Pandora 2 will actually have something like HDMI which you yourself consider a must they will need to design a new case. Any reasons to believe this will take less than a year this time?

How long do you think it'll take to get the remainder of the 4000 out?
Two Months ;)

I think an HDMI out is a must, but I say they can just go ahead and replace the S-Video with that and that'd be a good enough upgrade.
Well, HDMI does have some nasty licensing costs attached. If they upgrade the processor and add some form of VGA/DVI or better DisplayPort/HDMI out then I'm happy with the upgrade and will gladly order a new unit.
 
Hey Cosurgi,


Do you still have that usb to vga adapter that you purchased? I'm still hoping to see a review of that thing in action once you receive your Pandora.
 
We got the green light from TI to make a new unit with this SoC, yep, but it NEEDS a complete board redesign, it's no drop-in replacement or anything... so it'll take a few years.

My money's there when you get around to making it. I just think of all the possibilities of the first one, let alone an OMAP4.
 
Well, assuming the Pandora 2 will actually have something like HDMI which you yourself consider a must they will need to design a new case. Any reasons to believe this will take less than a year this time?
No need to redesign the case. Repurpose the EXT port, include an EXT->HDMI adapter with every Pandora (a small passive adapter, of course).
 
No need to redesign the case. Repurpose the EXT port, include an EXT->HDMI adapter with every Pandora (a small passive adapter, of course).

I can find two problems with that.


1. The EXT port doesn't have enough pins afaik.


2. Where would the UART lines go? A second EXT port?


Edit: After looking at the pinout, it *might* be possible to get away with that, but that would also break compatibility. But it might be possible to set it to the new "standard" by pulling one of the lines up/down or something?
 
Last edited by a moderator:
Well, assuming the Pandora 2 will actually have something like HDMI which you yourself consider a must they will need to design a new case. Any reasons to believe this will take less than a year this time?

I think that:


1) It wouldn't need a new case design, they could repurpose pins on the ext connector for an adapter. I didn't literally mean an HDMI connector was vital (in the same way it doesn't have an S-Video + RCA audio connector right now)


2) The case moulding consists of several components which had to go through an extensive refinement process - if they did need to add an HDMI port it'd just involve modifying two components and it wouldn't need a ton of refinement and strict tolerances met to get right.


3) The case company chosen took a year because they're cheap and have poor service quality. The iControlPad, for instance, took a fraction of the time.

Well, HDMI does have some nasty licensing costs attached. If they upgrade the processor and add some form of VGA/DVI or better DisplayPort/HDMI out then I'm happy with the upgrade and will gladly order a new unit.

It's $10,000 a year last I was aware. While that does hit a low volume product like Pandora a lot harder than most other HDMI products it's still well worth the price, IMO. It's a pretty big feature for a product like this, especially one sporting OMAP4.
 
2. Where would the UART lines go? A second EXT port?
Part of the magic that MWeston will be able to pull off? When configured one way, the EXT port is HDMI, but change the config in software (or maybe it's hardware keyed, that'd be cool too) and it reverts to UART and GPIO? It's definitely possible, but I think it'll depend on the SoC that is ultimately chosen for the successor. If it has HDMI on it (which isn't likely, but who knows what'll happen next year) then it's easy.
 
Didn't the Beagle Board guys get around the licensing for HDMI by not making it HDMI but instead making it DVI which is the same video, but no audio? Would something like that work if the licensing were a LOT less?
 
Didn't the Beagle Board guys get around the licensing for HDMI by not making it HDMI but instead making it DVI which is the same video, but no audio? Would something like that work if the licensing were a LOT less?

Yeah I read that some place too.. the Beagleboard documentation makes no mention of HDMI even though it has a connector that is just HDMI without the audio and call it DVI-D out.
 
Last edited by a moderator:
It's $10,000 a year last I was aware. While that does hit a low volume product like Pandora a lot harder than most other HDMI products it's still well worth the price, IMO. It's a pretty big feature for a product like this, especially one sporting OMAP4.
I was pretty sure I posted the Wikipedia quote in this thread.


It's 4 cents per unit + $10,000 per year "for high-volume manufacturing"
 
Back
Top