possible soc updates.


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.
First, I would NOT keep the Pandora brand.

The Pandora was a project where the main initiator was Craig, and since I would do a successor without Craig, I would not keep the name.

Second: After all the preorders have finished and the Pandoras have been sold, I have enough money, so I am not looking into investments from Qualcomm or whoever, I just want to buy and use the SoCs.

I've got a reliable company now (GC), I've got the funds needed, so nothing is in the way here.

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.
Yep, that's probably the main issue.

I'm pretty sure it wouldn't be a flop, since we've got a good community, good interest and I'm sure I can improve the current Pandora quite a lot to make it

a lot more attractive (proper case, backlit keyboard, better nubs, etc.)

However, it still won't be easy telling that an SoC manufacturer and convince them.

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.
Yep. I'm willing to try all I can, but if there's someone with experience or some good contacts in that regard, I'm thankful for every help :)
 
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.
That's sad news. Hopefully/maybe they'll be able to come around later.
 
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…
Have you seen the battery life on the new Macbook Air systems? That's well in the range of what a Pandora can provide (yes I know the form factor is completely different, but still Intel has been catching up in terms of power consumption a lot, even with their mainline CPUs).
 
KodeIn, on 13 Jul 2013 - 9:06 PM, said:


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…
Have you seen the battery life on the new Macbook Air systems? That's well in the range of what a Pandora can provide (yes I know the form factor is completely different, but still Intel has been catching up in terms of power consumption a lot, even with their mainline CPUs).
Yeah, it's not like the macbook air has a fucking small mainboard and is packed to the brim with battery cells in the remaining space… <_<  


That's why they dropped the optical media, that they use integrated SSD chips, that the battery cells aren't removable (not without completely taking apart the computer), etc.


READ THE FUCKING TOPIC GODDAMIT! It's been discussed to hell and back x86 processors are a big NO-NO in the pandora form factor!


Too power hungry, running too hot, …
 
Last edited by a moderator:
KodeIn, on 13 Jul 2013 - 9:06 PM, said:


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…
Have you seen the battery life on the new Macbook Air systems? That's well in the range of what a Pandora can provide (yes I know the form factor is completely different, but still Intel has been catching up in terms of power consumption a lot, even with their mainline CPUs).
Yeah, it's not like the macbook air has a fucking small mainboard and is packed to the brim with battery cells in the remaining space… <_<  


That's why they dropped the optical media, that they use integrated SSD chips, that the battery cells aren't removable (not without completely taking apart the computer), etc.


READ THE FUCKING TOPIC GODDAMIT! It's been discussed to hell and back x86 processors are a big NO-NO in the pandora form factor!


Too power hungry, running too hot, …
yeah,yeah, and atom smartphones don't exist/have a form factor and motherboard complexity that is no match for pandora's or probably eat the battery in 5 minutes.

Nobody's proposing to use laptop hardware for a handheld device, there are various x86 SoCs for that (just fyi, there is even non-intel low power x86 SoCs like vortex86 family that is used in various embedded devices). And I can testify myself that tablet atom SoC is no more power hungry then ARM one (I have a Samsung Smart PC and had Thinkpad Tablet before so I can compare battery life myself).

The main problem with current-gen atoms is PowerVR graphics that don't have proper linux drivers but that is going to change with the next generation atoms that are to arrive this year and support for them is already in latest development linux kernel
 
yeah,yeah, and atom smartphones don't exist/have a form factor
Effectively they do not as they have no meaningful market presence or existence beyond Intel assuaging the doubts of its investors regarding a large growing market Intel has no market presence in.

 

Nobody's proposing to use laptop hardware for a handheld device
Grench has been, although he also wants a future Pandora to make an iPhone feel like it needs to hit the treadmill.

And I can testify myself that tablet atom SoC is no more power hungry then ARM one
While the bait thrown at you was hyperbole, this is also nonsense.

A ) The statement is simply false.  To the degree there ever has been a legitimate claim along those lines it's a select few ARM SoCs comparing idles and highs.  The Anandtech report often dragged out to support that claim also showed ARM SoCs performing better in that regard. 

The less fanboyish conclusion associated with that was actually closer to Intel having demonstrated an in-order Atom that had idle and highs in the realm of the Tegra 3*.  Which if you were the typical Anandtech/Intel fanboy led to concluding that Atom would dominate the market and eat ARM alive, last gen.  Something that didn't happen and doesn't posit over to an out of order Atom on a new LP 22nm process versus ARM's big.LITTLE on mature 28nm process in 2014.

*Based on Intel provided measuring equipment, handled by an Intel engineer, and promoted on a website with a pro-Intel bias.  From a company known to... expend resources in garnering favorable results.

B ) They're going from in order to out of order to try to compensate for now having a complete lack of performance competitiveness, which is going to increase the power overhead.  In short all things being equal it's going to have higher idle power consumption.

C) From what we do know of Intel's 22nm process, so far, it's been rather leaky.  Hence Ivy Bridge had worse idling power consumption then Sandy Bridge or AMD's 32nm APU parts.  The 22nm Atom will be on a LP process rather then the HP one, but It's not some kind of foregone conclusion that this amounts to what many Intel fans would think. 

Sure the resistance part of the equation will go down as per normal for lithographic shrinkage, but that's only part of the equation.

D) Based on what they're doing and saying the next gen Atom processor isn't being optimized for phones.  It's basically focused on trying to counter Jaguar and ARM Cortex A15's taking the netbook/tablet market away from it.  Sure they're releasing a token phone variant, again, but Intel might have "supply issues", again, while almost no one's buying.

E) The entire principle behind the publicity stunt called Crystalwell demonstrates rather thoroughly that Intel's GPU tech isn't particularly efficient in terms of transistors or power by desktop/laptop standards.  It's pretty much guaranteed it won't be competitive with ARM space GPU tech.

F) Intel's known to favor protecting its profit margin even over association with long time critical allies.  Microsoft's experience with the original XBox should be evidence enough of that.

G) As already discussed x86 breaks a lot of the work that's been built up, in some cases in ways that could never be repaired, and has little meaningful, if any, advantage over ARM in this power envelope.  The opposite, which is to say ARM's advantages over Intel, however are rather significant.
 
Last edited by a moderator:
Congratulations on entirely missing the point.  Just to be clear, I have long advocated for getting the Pandora's successor down to under 20mm in thickness.

http://www.xbitlabs.com/news/mobile/display/20130705064935_Specifications_of_Intel_Bay_Trail_System_on_Chips_Gets_Revealed.html

My argument is that given the gains in both raw pwerformance and performance/watt associated with the 22nm tri-gate process, the Intel option should be considered and not simply dismissed out of hand.

<2.5W puts the N2805 within the 'habitable zone'.  Cost per unit for the entire SoC, thermal solutions, support and more all have parts to play in this equation.  The next process shrink to 14nm coming in 2014 takes it that additional step at a time roughly coinciding with where a new handheld would be emerging.

These new parts potentially put Intel in the right range of possibilities.  They also need some 'wins' in the handheld market - and that might make their mobile division more likely to collaborate with a project like this one.

The majority of your arguments are moot.

A) Old studies and old benchmarks are irrelevant.  Keeping an open mind, we should be looking at future tech, not old.

B) You're making that claim based on...  nothing.

C) Again - wild negative claims around technology that hasn't arrived.

D) Another completely unsubstantiated claim.

E) And you're basing that statement on...?

F) So you're faulting Intel's independence from Microsoft?  Huh?

G1) X86 breaks the work arounds that were required to work with our current SoC.  Pretty much any new SoC is going to require it's own tweaks.  True, changing architecture would be more dramatic in that load.  I've yet to have heard of any that are mission critical that could not be fixed.

G2) Assumptions of power per watt between ARM (generically applied) and products that aren't available yet is a bit silly.

G3) "ARM's advantages over Intel" - We're not talking about IP organizations or even production companies.  We're talking about actual SoCs that will be coming out inside the next year.

Clearly the ARM fanboys are out in force and it will be difficult to argue for a fair examination of all of the tech that is coming onto the market in the near future.

ED has the right approach at first finding out, "What can we get?"  After he knows that - then the process can whittle down to within the available options, "What will work best?"

It is my humble opinion that approaching one of the largest semiconductor operations on the planet - who just happens to be trying to make in-roads into the mobile space - and has the most advanced lithographic processes ever developed - to find out what they have  in the pipeline that would fit the need - THAT has value that should not be overlooked.

It's too bad Samsung wasn't interested - that narrows the field of possibilities.  Intel may just give ED a categorical, 'No,' too, but that can only happen if they're asked.
 
Last edited by a moderator:
Quote yeah,yeah, and atom smartphones don't exist/have a form factor Effectively they do not as they have no meaningful market presence or existence beyond Intel assuaging the doubts of its investors regarding a large growing market Intel has no market presence in.
It's strange to see a market share argument  in a thread about device that have market share so small that any single so called "failed" phone model outsells it by an order of magnitude.

Please don't use this argument anymore, it makes you look like a fanboy. As a long-time linux user I am tired of this argument being used to suppress an opponent.

Besides I can buy asus padfone in many big local stores nowadays and it seems like lenovo k900 is on the horizon so atom smartphones have enough market presence to actually buy it and that's all that matters.

Quote And I can testify myself that tablet atom SoC is no more power hungry then ARM one While the bait thrown at you was hyperbole, this is also nonsense.   A ) The statement is simply false.  To the degree there ever has been a legitimate claim along those lines it's a select few ARM SoCs comparing idles and highs.  The Anandtech report often dragged out to support that claim also showed ARM SoCs performing better in that regard.    The less fanboyish conclusion associated with that was actually closer to Intel having demonstrated an in-order Atom that had idle and highs in the realm of the Tegra 3*.  Which if you were the typical Anandtech/Intel fanboy led to concluding that Atom would dominate the market and eat ARM alive, last gen.  Something that didn't happen and doesn't posit over to an out of order Atom on a new LP 22nm process versus ARM's big.LITTLE on mature 28nm process in 2014.   *Based on Intel provided measuring equipment, handled by an Intel engineer, and promoted on a website with a pro-Intel bias.  From a company known to... expend resources in garnering favorable results.
I think it's a little rude to just dismiss the actual experience of a long-term device usage of an actual user and outright deny any truth in the vaguely formulated sentence.

I remember I read that Anandtech article and actually liked the wording it used. It is actually a technical article and no clear conclusions can me drawn from it except that there is approximate parity between atom and arm power consumption in similar conditions (for example if it showed that atom SoC used 3 to 5 times more power this conclusion could have been changed to clear ARM superiority). Detailed data from that experiment could only be interpreted together with measurement techniques and testing software (including firmware) and doesn't mean anything for a device user with another set of software he uses and actually can change over time with software updates.

B ) They're going from in order to out of order to try to compensate for now having a complete lack of performance competitiveness, which is going to increase the power overhead.  In short all things being equal it's going to have higher idle power consumption.   C) From what we do know of Intel's 22nm process, so far, it's been rather leaky.  Hence Ivy Bridge had worse idling power consumption then Sandy Bridge or AMD's 32nm APU parts.  The 22nm Atom will be on a LP process rather then the HP one, but It's not some kind of foregone conclusion that this amounts to what many Intel fans would think.    Sure the resistance part of the equation will go down as per normal for lithographic shrinkage, but that's only part of the equation.
Sorry, but unless you are an engineer working in that industry these are blind and uneducated guesses. Probably more trustworthy than outside temperature prediction by a random number generator but not by much.

D) Based on what they're doing and saying the next gen Atom processor isn't being optimized for phones.  It's basically focused on trying to counter Jaguar and ARM Cortex A15's taking the netbook/tablet market away from it.  Sure they're releasing a token phone variant, again, but Intel might have "supply issues", again, while almost no one's buying.
As far as I know Intel is still an engineering-driven company so they don't do anything major just to counter current market situation and I'm positive that there will be a phone low-power atom version that won't be worse then current-gen phone atoms in terms of power consumption. And the problem of SoC supply will be the least important for a pandora successor should Intel get interested in this project.

And, you know most current high-end SoCs aren't optimized for phones anyway, they just can run within a phone power limit.

G) As already discussed x86 breaks a lot of the work that's been built up, in some cases in ways that could never be repaired, and has little meaningful, if any, advantage over ARM in this power envelope.  The opposite, which is to say ARM's advantages over Intel, however are rather significant.
 

The very big intel advantage nowadays is their software support for their devices in linux. No other company (especially ARM SoC vendors) so far haven't provided full hardware driver support in mainline linux kernel with regular updates. This kind of support allows to install any linux distro without any big problems on any intel- (or amd-) powered machine (except some device-specific bits like buttons) and I would really like a pandora successor to be able to do the same. This could reduce P2 driver support need  to some P2-specific bits like nubs and keyboard thus dramatically decreasing the manpower needed to do it.
 
I am just willing to add i am still all for it :)

I use my Pandora every day, and would instantly trow money at a successor.
 
I hope a successor happens, I would pay $1,200 USD for a more modern Pandora type device. Would hope for x86, but could live with ARM with enough software support. 
 
I think we should try to keep it below $1000.

I don't care about the cpu architecture as long as it's well supported in the FOSS world (so x86 and ARM are both OK), it is available (duh), it has relatively good perf (even for single-core stuff it should be at least twice as fast as the current Pandora), and it has excellent perf/Watt (even better than the current Pandora).
 
What exactly is your purpose in linking a tabloid grade news site, which doesn't list a relevant bin?  Even if we took that as legit, the lowest power part there is a 4.5W TDP Celeron dual core clocked at 1.46GHz, which by Anand's estimates make it equivalent to a 730MHz Pentium IV.  In other words significantly worse then my optimistic ballparking on your behalf.

Extrapolation from that to a relevant power point is of course much worse.  The relevant bins are Intel's Z series, the last gen of which were the Medfield BGA617 parts that had supply issues despite Intel being the largest fab and being on Intel's rather mature 32nm process.

You have some serious issues if you're going to take Intel's SDP and pretend that's max draw.  Even Intel's TDP is not max draw, and max draw is what dictates the thermals as much as the Thermal Design Power.  SDP is irrelevant to thermal design considerations, hence why it and the TDP are seperate measurements.  The N2805 bin you're obsessing on would cook itself in a handheld.  Hence why it's an N series as opposed to Z series and why it's branded Celeron, it's a laptop part.  Hence those big heat sinks on the reference boards I showed you.

What is it about basic thermodynamic concepts related this is so hard to understand?  Processors generate heat and that heat has to go somewhere.  In a cramped compact design there's not much room for heat spreaders let alone fan driven air flow, and not enough for the draw you're talking about. 

Intel doesn't claim N series parts are for smart phones or similar profiles.  That's the role of the Z series parts.  Why are you trying to claim N series are suited?

A) Old studies and old benchmarks are irrelevant.  Keeping an open mind, we should be looking at future tech, not old.
B) You're making that claim based on...  nothing.

C) Again - wild negative claims around technology that hasn't arrived.
...

My content under A had nothing to do with benchmarks, so I'm not sure why you feel the need to mention them.  It had to do with power draw and the particulars of the claim of Atom processor being within ARM's range.  You also kind of aren't in a position to pull the "open mind" card when you're trying to sideline actual analysis in favor trying to just declare reality will warp to fit your whims.

B is based off the design choices and rationale of both ARM and Intel.  Not to mention Microsoft with the Xbox 360 and generally being considered a basic fact.  I wasn't aware more management circuitry consuming more power and out of order engines in particular doing so was a controversial subject.  Especially as I showed you the real world pragmatics of that with big.Little and how there are power/performance regimes.

How is talking about what we've seen of Intel's 22nm process, and expressing concerns the existing issues with the process won't necessarily be fixed, "wild negative claims"?  And so on, if I cared to continue but I think my statements stand just fine without reiteration.

It's strange to see a market share argument  in a thread about device that have market share so small that any single so called "failed" phone model outsells it by an order of magnitude.
You make a point to drag out the existence of the Z series parts in an argumentative way, and then call foul when the counterargument of the actual demand and use is brought up.  Do you actually have a point or are you just cheerleading?  I realize the previous poster was hyperbolic, but I'm not them and neither are you.

Sorry, but unless you are an engineer working in that industry these are blind and uneducated guesses.
You're free to think what you want.  You're also free to learn how to setup proper quotes instead of mixing my words with yours.

The very big intel advantage nowadays is their software support for their devices in linux. No other company (especially ARM SoC vendors) so far haven't provided full hardware driver support in mainline linux kernel with regular updates.
Is English your first language?  You have a negative that makes no sense, poor sentence structure, and aren't using proper capitalization.  To address the apparent point though...

Yes Intel has an open source driver, however it's not as good as the one under Windows and AMD largely does the same thing with their Radeon team in addition to the Catalyst team, especially with the kernel 3.11 pull, which is adequate to render your unqualified generalization false.
 
Last edited by a moderator:
@Jebe: A good majority of the people on this forum are people that English isn't their first language. Just be sensitive to it.
 
I think we should try to keep it below $1000.

I don't care about the cpu architecture as long as it's well supported in the FOSS world (so x86 and ARM are both OK), it is available (duh), it has relatively good perf (even for single-core stuff it should be at least twice as fast as the current Pandora), and it has excellent perf/Watt (even better than the current Pandora).
Exactly.  At this point the project should be completely architecture and SoC agnostic.  

  1. Inquire about everything in the pipeline (from every company that makes something that might qualify). 
  2. Find out what can even be attained (Samsung is apparently out - who will even talk to ED?).
  3. From within that population, do a cost/benefit analysis.  (Perf/watt, cost per SoC, features, code availability, etc...)
  4. Make selection.
Excluding any single architecture or SoC before doing 1 & 2 above artificially constrains the options.  All 4 steps are up to ED and whomever his project team is.  All that WE can do is note items in the pipeline for them to include in step 1 so that potential options can get included.
 
It's true that the x86 architecture has shown that a CISC based architecture can still be a viable competitive technology even in the face of RISC based architectures, which at one time were expected to make CISC completely obsolete.

However, it seems to me that the one area where RISC based architectures have really shown their superiority is with power efficiency.  In order to make CISC competitive here you have to use a much more refined process to create the chips, and that requires a big company and large volumes to make it worth it.  So far, it seems even under those circumstances it's difficult for them to compete with RISC based architectures for power efficiency.  How else could all these tiny manufacturers make ARM and MIPS chipsets with passive cooling that have shown up the CISC chips from giant Intel for power efficiency in mobile devices?  Intel has all it can do to mount competition in this area with manufacturers you would normally expect it  to steamroll.

I think that it would be great to have the kind of open source support that Intel has put forth for its CPUs and GPUs.  I still have real doubts that an x86 based chipset is going to be the best choice for a successor to the Pandora.  I'm not completely ruling it out (and I'd love to have support for OpenGL proper), but I wouldn't hold my breath.
 
I wish Intel would use its superior technology to produce ARM SoCs and sell them at a reasonable price in low volumes. One can dream... :)
 
BPdiRdBCMAEZPD9.jpg
http://www.engadget.com/2013/07/18/samsung-new-exynos-5-octa-soc-coming-next-week/

Woooooooooo... if only Samsung changed their mind :)
 
Last edited by a moderator:
^  I read about the new soc the other week, Ive the current one has all kinds of issues, poor performance and heating issue.  The new soc seems to me to be an attempt to fix the wrongs of the current octa soc.
 
Last edited by a moderator:
Back
Top