the open pandora 2 must use an amd apu!

[Jonttu]

I personally wouldn't mind an x86 Pandora...in a few years. In a few years I have a feeling that Intel will be on top perf/cons. wise. Honestly they are heavily in the lead in nm chomping with their dyes. With the Motorola Razr I, they made an incredible leap in competing with arm equivalents perf/consumption wise in a ridiculously small amount of time. I mean, what can you acclaim that to? Is it luck? Or is it just because Intel just happens have shit tons of engineers and money to develop towards the platform. I would scoff at this if they simply hired ex-rival engineers (similar to that which Apple did with the A6) and would just be stuck at the intermediary gap of just catching up, but Intel hasn't hired any new micro architecture engineers. They haven't been heavily focusing on the ultra-mobile platform. Medfield, when it was introduced performed better than arm equivalents at the time. Yeah it was poor in battery life and Intel being the cockers that they are waged people would prefer power to endurance. They were loosing ground space after netbooks and umpc's started to flop off the market ( :,( ) and started focusing on ultrabooks ( ) good move. My point is, now that the focus is on the right thing, Intels SoC are going to start intimidating competitors. I might sound like an x86 loyalist, but in all honesty it's just a logical deduction, and the evidence is quite astounding. But yeah sorry for the bump, just had to put that out there.

 

[FaeMinx]

^ I feel you will be proven right. I'm sure we will see some amazing developments in the next few years.


Arm, however is jumping forward in leaps and bounds and wouldn't you say it has the *fastest growing* software catalogue of any architecture *ever*.


Thus, it is here to stay.


Quite possibly in the not too close future we will have devices with both arm and x86 chips inside.


What we really need is a dramatic breakthrough in battery technologies - or some miraculously effective form of passive charging.

 

[_wb_]

Yes, Intel has an impressive army of engineers. And if they change their focus from horsepower to power consumption (I have no idea how big of an "if" that is), they might be able to come up with something better than ARM-based low-power cpus.


One huge thing that is keeping them back though is the need for backwards-compatibility to a big old ugly legacy instruction set. This is also their main advantage / selling point to people who are interested in keeping big old ugly legacy code running without needing to recompile or emulate.


I am not one of those people.

 

[sswam]

I have a feeling that Intel will be on top

Yeah? Well, I vote ARM.

or, maybe some nice new parallel system will take over


Yeah, ARMH is up 1176% over ten years. INTC is up ~26%. (AAPL is up 6721%, but they are not really in the CPU business!)


Here's another pretty picture:

So, over the last 10 years, Apple > ARM > MIPS > NVIDIA > Intel > Microsoft > AMD.

 

[Exophase]

Mobile is only one of many businesses for Intel (who has a much, much larger market cap than ARM) so of course it's not going to dramatically grow the whole thing. ARM has been growing all over the place, its Cortex-M series has grown its presence in embedded by a lot as it pushed down on other microcontrollers..


But it looks to me like ARM's growth has hit a pretty big plateau as of 2011, so I wouldn't expect them to hit huge new spurts. Maybe they'll gain a little more from servers.

 

[_wb_]

Low-power low-heat server farms sure sound a lot better than having to build your server farm near a river to have cooling water available and getting electricity bills that go through the roof. Wikipedia says: "The performance of the very largest server farms (thousands of processors and up) is typically limited by the performance of the data center's cooling systems and the total electricity cost rather than by the performance of the processors."


So it makes a lot of sense to use ARM there.

 

[sswam]

it looks to me like ARM's growth has hit a pretty big plateau as of 2011

Yeah, maybe due to "global recession" or whatever. According to my 2nd graph there, NVIDIA and MIPS also had some trouble starting 2011, and it bit them worse than ARM. ARM has been on the up again for >6 months, while Intel and AMD look to be declining.

 

[gunrock]

I love my ARM devices and their cunningly low power drain cpu's, but if Intel, as _wb_ says can sort out their power consumption, the idea of portable Xbox 1 emulation is enticing!

 

[Exophase]

XBox emulation on PC has been very limited and what little you get requires Windows, is that really that enticing?

 

[_wb_]

Maybe someone should start a ClosedPandora project dedicated to bringing some locked down Windows on a low-power x86 chip to the world of handheld consoles.


I wouldn't be very enthusiastic about such a project, but it looks like some people are really addicted to their chains and balls.

 

[Exophase]

If Pandora 2 uses a modular CPU board then hopefully OPT releases the interface specifications, then someone else can make boards for stuff like this.

 

[craigix]

Actually if we supported something as big as the Pi board for example you could make an x86 one.


We have ordered a Replicator2 to test these designs.

 

[Foxgod]

Actually if we supported something as big as the Pi board for example you could make an x86 one.


We have ordered a Replicator2 to test these designs.

Cool, looking forward to some nice home cooked cases

 

[craigix]

Should allow for getting things out to potential customers to try in hand too, quite exciting.

 

[GizmoAgent]

http://formlabs.com/


Shame you couldn't get one of these

 

[gunrock]

XBox emulation on PC has been very limited and what little you get requires Windows, is that really that enticing?

I've not been following Xbox emulation much, but what's the issue? Given that softmods and mod chips have replaced the host OS with their own bios and dashboards since the year dot, I can only guess that lack of interest is holding it back? Running unsigned code seems to be easy, so is it running the signed game executables that's the issue?


My interest is in being able to play my backlog... That's what makes the idea enticing!

 

[Foxgod]

XBox emulation on PC has been very limited and what little you get requires Windows, is that really that enticing?

I've not been following Xbox emulation much, but what's the issue? Given that softmods and mod chips have replaced the host OS with their own bios and dashboards since the year dot, I can only guess that lack of interest is holding it back? Running unsigned code seems to be easy, so is it running the signed game executables that's the issue?


My interest is in being able to play my backlog... That's what makes the idea enticing!

I think the next step in emulation will be getting n64, dreamcast and saturn running at good speeds.


I dont see a lot of hope for anything thats stronger then the dreamcast, maybe Gamecube games will be running at 4 or 5 fps on an A15 SOC.

 

[WizardStan]

I think the next step in emulation will be getting n64, dreamcast and saturn running at good speeds.

Gunrock is talking about theoretical XBox emulation some theoretical Pandora 2 can get an x86 module. The idea goes that the XBox is an x86 processor with basically DirectX libraries, so emulation should be a piece of cake, little more than a bunch of library wrappers and some validation code. The reality is that it isn't nearly so simple and the offerings available are very limited.

 

[Exophase]

I've not been following Xbox emulation much, but what's the issue? Given that softmods and mod chips have replaced the host OS with their own bios and dashboards since the year dot, I can only guess that lack of interest is holding it back? Running unsigned code seems to be easy, so is it running the signed game executables that's the issue?

My interest is in being able to play my backlog... That's what makes the idea enticing!

The issue is that the WINE approach is hard and limited. There are lots of different versions of the OS and libraries used, and they have to all be recognized and in some cases handled differently by the emulation. Plus not all games are limited to playing nice with the OS/library abstraction, some will access hardware in a way that's difficult to emulate.


To this date only a few games are really playable on XBox emulators. One of the people who has worked with XBox emulation a lot over the past few years is convinced that the approach is a dead-end and that conventional emulation techniques need to be pursued.

 

[gunrock]

The issue is that the WINE approach is hard and limited. There are lots of different versions of the OS and libraries used, and they have to all be recognized and in some cases handled differently by the emulation. Plus not all games are limited to playing nice with the OS/library abstraction, some will access hardware in a way that's difficult to emulate.


To this date only a few games are really playable on XBox emulators. One of the people who has worked with XBox emulation a lot over the past few years is convinced that the approach is a dead-end and that conventional emulation techniques need to be pursued.

Actually, I've never advocated the DirectX and stubbed windows kernel emulation approach. I would have thought traditional hardware level emu is the way forward. I'd guess (never having done it myself) that often, there's a lot of effort emulating the cpu and custom hardware, but in this case the CPU emulation part is essentially gone, thus emulating the Xbox 1's GPU would be the focus and that would be at least a little understood, I'd guess, given it's an Nvidia part.


I never, Stan, thought it'd be easy, but a degree or three easier than PS2 emulation, surely? I think basically, the desire isn't there. Emulating the PS2 makes sense as it would create a large degree of kudos and it has a better library than the Xbox 1. This latter point probably accounts for the GameCube emulation effort, too.