possible soc updates.

[vcoleiro1]

Yes I do think those HDMI sticks are high volume for Rockchip who dominate that market in terms of the SOC used in most of them. However, Rockchip SOCs are in more than PC Sticks  , they are in tablets like the Archos, Cube, Pipo etc etc. They are also in gaming handhelds. Rockchip sell in excess of 40 million tablet SOCs PA - not including the SOCs for the PC Sticks or  gaming handhelds and other devices. They are not small fry.  That said I take your point about different distribution set ups etc.  

But as far as people trying with Qualcomm, other than smartphone and tablet makers who do produce in large volumes, I can't see anyone else who would have tried to approach them.  My point is lets not die wondering, it doesn't hurt to ask , you don't know what Qualcomm are willing/unwilling to do if you don't ask. At the end of the day you would have at worst lost the cost of a few phone calls.   

 

[Exophase]

Yes I do think those HDMI sticks are high volume for Rockchip who dominate that market in terms of the SOC used in most of them. However, Rockchip SOCs are in more than PC Sticks  , they are in tablets like the Archos, Cube, Pipo etc etc. They are also in gaming handhelds. Rockchip sell in excess of 40 million tablet SOCs PA - not including the SOCs for the PC Sticks or  gaming handhelds and other devices. They are not small fry.  That said I take your point about different distribution set ups etc.

You totally missed what I was saying. It's not that they're a small company, it's that they service a huge number of relatively small companies. It doesn't matter if they're in several HDMI sticks if those all represent a bunch of smallish orders. That's what their market is like now.

But as far as people trying with Qualcomm, other than smartphone and tablet makers who do produce in large volumes, I can't see anyone else who would have tried to approach them.  My point is lets not die wondering, it doesn't hurt to ask , you don't know what Qualcomm are willing/unwilling to do if you don't ask. After all, it's just a phone call or two.

Your point was duly noted already. I already said I didn't suggest not trying. Why would I say something so stupid?

But if you think only OPT was interested in making a low volume mobile device...

 

[vcoleiro1]

The real question on following up SOCs for a P2  is when/if ED will be ready and willing to do so. 

ED, any light on when you would go down a Dragon Box path.  Would it be after all Craigs P1 pre-orders are cleared and the investors paid off?

 

[EvilDragon]

None of these SoCs are available through normal distributors channels.


Not TI, Samsung, NVidia or Qualcomm (and AFAIK not even Rockchip).


It works like this:


You get in contact with the company, let them know your business plan, get the technical handbook (if you're successful) and get drivers and support.


Then they tell you what distributor to use, usually the same one who sells all their stuff (I.e. for TI it's EBV, for Samsung it's MSC (at least in Germany)).


When you order the parts via the distributors, you have to tell them the project name. If the project has green light, you can order the parts.


So it's not possible to simply buy these SoCs.


Usually, these companies only give green light if the expected quantity is huge enough or if it's somehow else helpful to them (stuff like Ouya, that is expected to gain a huge popularity and is therefore advertising for that SoC as well).


They need to give you support for their SoC when building the hardware, which can eat a lot of ressources. If there's not enough to expect back, they won't waste these ressources and won't allow you to use their SoC.


They often have open hardware developer boards (like the BeagleBoard, the PandaBoard, the Arndale, etc.). This is a mixture of advertising and offering a reference design to others to lower support costs.


That's how it works. If we can't get a company interested in something like a Pandora successor, we don't have a chance getting a next-gen SoC.

 

[Asiyura]

My interpretation of what he told on the forums: yeah, he first wants the investors to get their money back.

Then get more than two hours of sleep at night.

And once he'll be more relaxed, he'll start thinking about making a pandora successor.

He already have some leads as to who to contact for hardware design (EvilDragonette for the casing, the people that made GTA04 for the circuitry IIRC).

But that's far too early to have any serious talk about the hardware.

 

[EvilDragon]

The real question on following up SOCs for a P2  is when/if ED will be ready and willing to do so. 


ED, any light on when you would go down a Dragon Box path.  Would it be after all Craigs P1 pre-orders are cleared and the investors paid off?

No clear path yet. I AM already gathering contacts to companies as that can take a long time, so I can be ready when the time comes.


I'm even contacting SoC companies right now. Even if we don't need it right away, but if you're already in good contact with them, it's easier to get their next-gen SoC, whatever that may be in the future.


The investors don't need to be paid off upfront, the development costs for the successor was planned with that invested money.


Samsung was easiest to get hold off and while Samsung Germany loves the idea and wants to help me as good as possible, they couldn't convince the korean mother company, so there won't be an Exynos for us.


TI would be our best guess, but they're retracting themselves from mobile business.


So there's Qualcomm, NVidia and Rockchip left right now.


Another way would be to use modules - they can be bought through normal distributor channels. They are basically the SoCs setup on PCBs ready-to-use.


However, most modules are for embedded systems and way too huge for our needs.


I only know Jorjin for mobile module solutions, but their best one is a TI CortexA9 right now. It's unknown whether an A15 will follow or not.


So yes, if anyone has any good contacts into one of the above companies OR knows any suitable modules we could use, let me know.


I'll be happy to contact them.

 

[Exophase]

You can buy most TI SoCs (including DM3730) from many distributors who have them in stock, although they may only have hundreds or even tens instead of thousands. Same with Freescale. Both also sell them directly on their website. But for TI this doesn't apply to OMAP5 or even OMAP4 and they didn't make distribution versions like they did with OMAP3 (DM3730 essentially being the same product as OMAP3630 but sold in different channels)

There are also Chinese warehouses with Chinese SoCs, you can find them on AliBaba and ask them for a quote, although I'm sure not all of them are legitimate.

But most other SoC makers don't do this.

 

[vcoleiro1]

ED , would you not hold off on a P2 until Craigs Pre-orders are fulfilled?. OR would you envisage P2 profits then paying off Craigs P1 orders -  with P2's then being paid and delivered to them.

 

[EvilDragon]

ED , would you not hold off on a P2 until Craigs Pre-orders are fulfilled?. OR would you envisage P2 profits then paying off Craigs P1 orders -  with P2's then being paid and delivered to them.

Without even getting detailed information from Craig about the remaining preorders, I could very well wait for an eternity.


I'd rather offer upgrades to the new handheld instead of waiting forever.

 

[EvilDragon]

You can buy most TI SoCs (including DM3730) from many distributors who have them in stock, although they may only have hundreds or even tens instead of thousands. Same with Freescale. Both also sell them directly on their website. But for TI this doesn't apply to OMAP5 or even OMAP4 and they didn't make distribution versions like they did with OMAP3 (DM3730 essentially being the same product as OMAP3630 but sold in different channels)There are also Chinese warehouses with Chinese SoCs, you can find them on AliBaba and ask them for a quote, although I'm sure not all of them are legitimate.But most other SoC makers don't do this.

The OMAP-Series was planned to be the professional series of their SoCs, therefore, no OMAP could be bought right away.


The DM-Series was the series planned for low-quantity devices as well, everyone could buy it.


Their plan was to release the DM SoC about half a year after the OMAP.


They skipped that with the OMAP4 but wanted to do that with the OMAP5 (but they're restructuring now, so it's all unknown...)


The DM-Development team actually is a totally different one than the OMAP team.

 

[rohezal]

They are selling very small  Exynos4412 Prime servers with Exynos4412 boards attached to the maiboards. Maybe ask them if you order 10K of this boards, if they would sell them alone?

http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G135235611947&tab_idx=2

 

[EvilDragon]

They are selling very small  Exynos4412 Prime servers with Exynos4412 boards attached to the maiboards. Maybe ask them if you order 10K of this boards, if they would sell them alone?

http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G135235611947&tab_idx=2

That's the standard Hardkernel-Module, yes. You can buy it from them without the server board.


You can get it here:

http://www.hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G135270682824&tab_idx=1

That would be a solution, yes. Price is pretty high, but I'm sure that can be discussed.


I've also sent them a mail and asked about an A15 module, but no reply for weeks yet.

 

[rohezal]

I want to purchase big quantity. What is the best price for 10K or higher units ? 

Price is all the same not depending on the quantity. 

Why? Are they are soldering the stuff by hand? Maybe a nice email from  ED or GC can convince them that a huge order number is attractive and can lead to a discount. AMD is very happy for the PS4 and XBOX One.

 

[Exophase]

The OMAP-Series was planned to be the professional series of their SoCs, therefore, no OMAP could be bought right away.

That plan was probably put into effect after OMAP3530 because you could easily buy it from random distributors and TI themselves. And that's probably why that's the one that was available for Pandora.

The DM-Development team actually is a totally different one than the OMAP team.

Maybe that's true for some other product but DM3730 and OMAP3630 use the same die...

 

[fusion_power]

So it's not possible to simply buy these SoCs.

So they don't want to make max. profit with selling their chips to everyone wo just pays? Strange Companies.

That's how it works. If we can't get a company interested in something like a Pandora successor, we don't have a chance getting a next-gen SoC.

 The chances are better these days imho, thanks to Kickstarter, see OUYA, GCW-Zero, Game Stick &co. . AND you can show them the Pandora, so these Chip Companies can see that there is a potential profit project incomming (at least for them ).

In theory, this should make hardware aquisition much more easy than it was with Pandora 1.

 

[CrazyIvan]

That's how it works. If we can't get a company interested in something like a Pandora successor, we don't have a chance getting a next-gen SoC.

Have you tried contacting Intel? I don't think contacting sales will do any good but probably intel OSTC guys will be interested and could help.

Intel have the best opensource support for their hardware (except PowerVR crap in current atoms that will go away with next generation) and I would prefer better support to better chip performance since OpenPandora don't have a large hardware support team anyway.

Next generation atoms won't be available until the end of this year but I doubt that P2 will be in active development by that time.
 

 

[Asiyura]

Yeah, that way we can have an Intel x86 processor in a palmtop computer connected to a large backpack containing the 35Kg of batteries able to maintain him powered for 5 hours…

 

[urisma]

So it's not possible to simply buy these SoCs.

So they don't want to make max. profit with selling their chips to everyone wo just pays? Strange Companies.

That's how it works. If we can't get a company interested in something like a Pandora successor, we don't have a chance getting a next-gen SoC.

 
The chances are better these days imho, thanks to Kickstarter, see OUYA, GCW-Zero, Game Stick &co. . AND you can show them the Pandora, so these Chip Companies can see that there is a potential profit project incomming (at least for them ).


In theory, this should make hardware aquisition much more easy than it was with Pandora 1.

You're forgetting about overhead. 5000-10000 chips may seem like a lot, but it hardly is. A chip company can't just send the wafers off and call it a deal. There's support (which is a huge portion), manufacturing, coordination, making the contract go through, etc. All of these take time and money. Second, is all the money going to be up front, or will Pandora be looking to SOC manufacturers themselves for investment? If it's the first, where is Pandora going to get the money? If the second, is the Pandora brand a safe investment? It's had an incredibly rocky history, which is bad for investment. Potential profit for a company is very dangerous. They need to make absolutely sure that it does bring profit.

I can offer some perspective. Take Qualcomm, with their Snapdragon line. It seems like it'd be a great idea to sell the DragonBox out of the DragonShop powered by QUALCOMM Snapdragon. No brainer, right? Qualcomm doesn't fab, which is fine in this case, since creating the whole console will be taken care of by other people. All Qualcomm has to do is provide some chips. Here's the problem. Chips don't make Qualcomm its money. For the money Qualcomm does make off chips, a whole lot of that still isn't SOCs. Chips have high overhead, meaning that for every dollar worth of chips sold, a high percentage of that is not profit. The overhead of fulfilling a small order like Pandora would take would make the profit margin even more narrow.

The even bigger problem is perception. The Snapdragon is Qualcomm's marketed brand, meant to become a household name (like how "Intel Inside" is for Intel). If the Pandora2 were to be powered by Snapdragon and flop, that'd be very very bad for the Snapdragon platform.

I'm not saying its impossible for the P2 to obtain a low volume order from a big company, I'm just saying its quite hard.

 

[CrazyIvan]

Yeah, that way we can have an Intel x86 processor in a palmtop computer connected to a large backpack containing the 35Kg of batteries able to maintain him powered for 5 hours…

well, a conservative power density estimate for current Li-Pol battery technology is 100Wh per 1kg (I have a fully assembled external battery with 50Wh capacity that weights a little less than 0.5 kg so this is a conservative estimate). So a 35kg battery will give us about 3.5kWh of power capacity. If we account for large recharge currents we can assume that at least half of that capacity can be used in such operating conditions (5hr discharge). That gives us 350W power consumption of a palmtop device. This kind of power trying to dissipate through such a small device will melt it. To put it into bigger perspective the typical high-end single card gaming desktop system can reach this number if this card is a dual-chip one and for example my desktop with overclocked c2d e8600 and gtx 670 could hardly reach 250W of peak power consumption. So even if we throw in monitor power consumption (about 50w for 27'' led one) we could say that 350W is quite big even for a desktop.

And typical max power consumption for a mid-range notebook (normal voltage processor, mid-range discrete graphics, 15.6'' screen) hardly reaches 50W.

 

[slaeshjag]

^ Just an fyi: The prefix k is for kilo. A kilo is 1000 of something