Which time frame do you prefer for the Pandora 2

[Foxgod]

As we know, rumors are that Craig might want to start with the Pandora 2 in 2013, and have it powered by an ArmV7 A15 with adres extension in order to support more then 4 GB of RAM.


However, other rumors currently state that in 2014 the ARMV8 may come out, if it turns out next year that the ARMv8 Will indeed be released in 2014, would you guys still prefer to have a new Pandora in 2013 with an A15 in it, or would you guys rather wait till 2014 so it can be powered with some of the first ARMv8 socs?

 

[Caine]

Aside from the fact that they will never get it ready in 2013, the spec race is endless. If you wait then there will just be another next best thing the year after, it doesn't really make sense to delay it because of that.

 

[elwing]

In 4 months of course!

 

[Exophase]

Right now we don't know anything about ARMv8 mobile SoCs. We don't even know if they'll be out in 2014. SoC manufacturers have been roughly following a tick-tock like release cycle, where they release an SoC one year then release an updated and shrunk version the next year, then finally release an entirely different one based on new CPU/GPU IP the year after that. 2014 is really soon for an entirely different CPU design.

 

[Foxgod]

Aside from the fact that they will never get it ready in 2013, the spec race is endless. If you wait then there will just be another next best thing the year after, it doesn't really make sense to delay it because of that.

I agree with that, but at best it would be a better ArmV8, its not like Armv9 will be out shortly after.


Question is, would you prefer a pandora with the last and most powerful ArmV7 available, or would you want the Pandora to have one of the first ArmV8's ever to be released.

Right now we don't know anything about ARMv8 mobile SoCs. We don't even know if they'll be out in 2014. SoC manufacturers have been roughly following a tick-tock like release cycle, where they release an SoC one year then release an updated and shrunk version the next year, then finally release an entirely different one based on new CPU/GPU IP the year after that. 2014 is really soon for an entirely different CPU design.

Well, if we do get specs and more info next year, we could have a completely different perspective on it.

 

[Caine]

Well, if we do get specs and more info next year, we could have a completely different perspective on it.

Yes, but design and development of the P2 takes time. You can't design it around a chip that may or may not be revealed next year.

 

[Foxgod]

Well, if we do get specs and more info next year, we could have a completely different perspective on it.

Yes, but design and development of the P2 takes time. You can't design it around a chip that may or may not be revealed next year.

I wonder if we will get a new Pandora every two/three years then


Not a problem to me, i wouldnt mind to buy a new one every bunch of years and trade in the older one to cut cost.

 

[Exophase]

That seemed to be Craig's original plan, he said he wanted to have a cadence more like Apple's than Nintendo's.

 

[Foxgod]

That seemed to be Craig's original plan, he said he wanted to have a cadence more like Apple's than Nintendo's.

I have to agree with that.

 

[ekianjo]

That seemed to be Craig's original plan, he said he wanted to have a cadence more like Apple's than Nintendo's.

Between words and reality...


Well if it was only up to me, a new unit every 2 years or so is better. As long as it stays compatible with the previous ones.

 

[Exophase]

Between words and reality...

Well if it was only up to me, a new unit every 2 years or so is better. As long as it stays compatible with the previous ones.

I agree, especially if OPT wants to keep selling something for top dollar and keep promoting high end hardware.

 

[ekianjo]

Between words and reality...

Well if it was only up to me, a new unit every 2 years or so is better. As long as it stays compatible with the previous ones.

I agree, especially if OPT wants to keep selling something for top dollar and keep promoting high end hardware.

Yeah, and it matches more or less the cycle of my PC renewals (desktop at least): just enough time for the technology progress to become noticeable between one version and another.

 

[albo]

That seemed to be Craig's original plan, he said he wanted to have a cadence more like Apple's than Nintendo's.

I have to agree with that.

Does it mean a Pandora every year?

 

[Batou456]

Okay let's clarify terminology here. ARMv8 and ARMv7 are standards related the instruction set architecture. They're only tangentially related to things how the micro-circuitry is designed and how well gated things are.


It's also a very valid question if ARMv8 processors will even be produced for mobile devices as these 64-bit ARM processors are being developed specifically to address the needs of expanding into the Server market where 32-bit processors aren't really considered acceptable anymore. Operating with 64-bits instead of 32 however involves more power usage and die area.


ARM Holdings is tending to indicate that the Cortex A7 and A15 are intended to be the low and high standard for handsets and handhelds for the immediate future, neither of which is even out yet. If they do, do it I'd expect it to be in time for a Pandora Mk. 3.

 

[mcobit]

And not to mention, that there might be some problems considering, that modules and drivers might not be ready for everything 64bit arm related, when the first socs come out.

 

[Exophase]

ARMv8 for mobile SoCs is inevitable though, using > 1GB of physical memory with the Linux kernel is already a performance compromise. ARMv8 has full support for 32-bit operations in order to facilitate lower power. The die cost is probably pretty minimal since we're talking about fairly wide OoO cores where a lot of resources are spent on scheduling and the ALUs themselves only take tens to maybe hundreds of thousands of transistors. Look at how little die area NEON adds on Cortex-A9, that has 2x64-bit ALUs in addition to 2xFP32 units and a bunch of multipliers.


But I still think we're looking at at least 2015 before any real adoption. Intel might put some more pressure on things since their lowest power Atoms just need to have 64-bit switched on.

 

[Binky]

I'm thinking than if a P2 were to use a 64bit SOC, it would be one of the very earliest available, and would probably be full of bugs and drivers issues. Maybe leave 64bit for the P3

 

[Askarus]

We need P2 at least in 2014.


Else P1 is really to outdated.


Don't wait to long with a new device.

 

[Grench]

Regardless of what processor is available when - use whatever is available to meet the date you're planning to ship on.


I'm in queue to get a 1Ghz Pandora. I'll be ready for a replacement device in ~1.5 years. Others will vary - but for me, January 2014 sounds good.


What will be available for sampling 6 months before then and guaranteed to be in mass production by November of 2013?

 

[FRZ]

First quarter of 2013 would be nice.