the open pandora 2 must use an amd apu!


I'm sorry, your point is what exactly? Yes, this can run some Windows games Pandora 2 wouldn't be able to, I admitted that. Yes, Intel processors can emulate ARM Android apps, at less than 50% native speed - probably a similar hit to what you'd get with an optimized qemu + wine port. Do you even have any word that Intel's software that accomplishes this is even available for AMD platforms?


Your old UMPCs don't come anywhere close in performance to what a Pandora 2 would, except in those Windows games. Notice how like usual the battery life tests are strictly relegated to web browsing and video playback instead of really using the CPU.


And I really don't think you're getting it - that UMPC used a much slower CPU/GPU than what you're asking for in this thread. Right now, and for another several months, there are really going to be two options for x86 - either Medfield level with a CPU and GPU well below what the soon to be released Cortex-A15 SoCs are, or Hondo (or perhaps Clover Trail) level with TDPs too high for Pandora's form factor.
 
Last edited by a moderator:
yeah, I am not sure if intels arm emulator will work properly on amd parts but some older umpcs from 5 years ago would be the p2 if its arm...if its omap4/5 remember the sony ux series that used culv core 2 chips and the wibrain that used the via c7 or even the asus eee pc 7" that used the celeron chips...older atoms and z-series atoms aren't my fav either but the z-series AMD parts aren't too bad...see the msi windpad and acer w500. As for software...is the pandora built around the idea of an opensource computer in the palm of your hands then why limit it with arm chips that you guys still don't have proper documentation(mostly talking about drivers for graphics) to and if x86 fits most criteria why not make the switch?
 
Sony UX is a Core Solo around 1.2GHz, not a Core 2. VIA C7 is a joke that's substantially weaker than Atom. I'm confident that a 2GHz dual core Cortex-A15 would have no problem beating either of those, and I'm also confident it'll have no problem beating this CPU (1GHz dual core Bobcat) or anything Intel is currently offering in this TDP range (currently just Medfield, 1.3GHz turbo to 2GHz single core Atom, but perhaps will include Clover Trail - dual core up to 1.8GHz Atom). In fact, with AMD if you want a higher clock speed you have to take on the 18W Zacate line. For Intel you're stuck with Atom, with the nearest better uarch being the 17W Ivy Bridge - the ULV Core Solo days are over. Want to put something like that in something like Pandora? Tough shit.


Yes there were lots of little netbooks, UMPCs, and tablets that used x86. These are not Pandora's form factor and these are definitely not Pandora 2's form factor. And you know what? UMPCs never caught on. I wonder why. But you want Pandora to try to recreate this? The closest you came with something like Sony UX is still larger than the original Pandora, and lasted 1.5 hours if you used the CPU intensively. But more to the point it cost a small fortune, not in any small part due to the CPU which would have been cherry picked lowest voltage of the lowest voltage, probably something no more than a tiny percentage of the chips could hit. And you paid for this. But these days Intel doesn't do that anymore, because they'd rather sell you an Atom, and AMD doesn't do that because they don't have the market for it. So again, want to recreate your old UMPC processor? Tough shit.


I know you have strong special interests for x86 but they don't apply to most people, least of all not most people on these forums. Pandora 2 isn't using a CPU + I/O chip that needs 6W, and showing me some tablet that runs on it changes nothing. Now stop arguing with me about hardware specs, you don't have enough idea of what you're comparing. Your argument is a waste of time because x86 does NOT fit most criteria, you can't hand wave parts that have an order of magnitude worse perf/W like your Bobcat chip.
 
Last edited by a moderator:
I think it would be pleasant to have a pocket PC that can run modern or at least 21st century PC + Windows + x86 games directly.


But the Pandora is not that machine. It runs open source Linux, on the ARM processor.


It would be possible to build a pocket Pandora look-alike based on the x86, but I don't recommend it. Perhaps Microsoft will build an x86-based phone that can run windows games, then you can attach an iCP2 to it. When some company clones that phone, and someone else ports Linux to it, I might even buy one (of the clones). But I prefer ARM.


I reckon a pluggable x86 "second processor" for ARM Pandora might be interesting, following the Acorn tradition. Acorn made 6502, Z80, ARM and other 2nd processors for their BBC Micro in the 1980s (in addition to developing the ARM chip, of course).


Just about any game developer who wishes to do it can port their games from x86 windows to ARM / SDL / GLeS, and enter into the phone / tablet market. Or, they can write decent portable software in the first place; which is probably easier than mucking about with Windows APIs and Direct X fugliness. ARM is the processor that keeps x86 from an almost complete monopoly. It's also simply superior in many ways. We should continue to use ARM.

Jesus wept.

Every time someone say "AMD x86 Pandora", raptor-Jesus eats a baby kitten. ;)
 
Last edited by a moderator:
Umpcs didn't work because of price...those things were $500-900 and the prices should have fallen for x86 chips compared to 4-5 years ago. also why must pandora only stick with arm?
 
Last edited by a moderator:
monstercameron: I would also love an ultra-portable x86 device - even if it only has about 2hours under full load... In that case I would buy 2 - 3 spare batteries and probably use it mostly while it's plugged in.


But then I would look at the costs and realise that I could get a very capable laptop for much less that would give me much more use overall - albeit slightly less portable than the aformentioned hypothetical UMPC and far far less portable than the Pandora / 2.


There simply is no way that x86 can even come close to the portability / battery life we enjoy so much with our dear device.


It's a nice dream, but totally impractical unless there occurs some miracle in battery technology sometime in the future.
 
Something that would be awesome:


an x86 CPU in a P2


Some other things i'd like:


the CPU being pasted to a solid aluminium P2 case with heat dissipation fins to stop it burning a hole in my pocket


battery technology to keep the little pocket-warmer running


enough money to afford a solid aluminium case
 
The Z-60 is intended to be used in tablets

Binky said:
Some other things i'd like:
the CPU being pasted to a solid aluminium P2 case with heat dissipation fins to stop it burning a hole in my pocket


battery technology to keep the little pocket-warmer running


enough money to afford a solid aluminium case
you know what, thats awesome! I tend to have cold hands. In winter that'd be a nice feature to have. These chemical hand warmers just don't have that stamina of a 4000mAh battery : (

monstercameron said:
so why must pandora only stick with arm?
because ARM is the obvious choice for portable devices with small from factors. x86 processors just have reached the point where they can compete with ARM in the smartphone/tablet market. YES! ARM performes much better but the RAZR i (x86 based) is not bad eigther!

CraigiX said:
X86 is dying. Watching it flailing about as it goes down is sad.
x86 also improves. I'm not seeing them getting canibalized by ARM in the future. Instead Intel and AMD are clearly putting up a fight.


Don't you guys think thats rather intersting? ; )
 
why is arm the most obvious(architeturally speaking and not sourcing/pricing) mips has their aptiv line which they claim beats the equivalent arm sku(?) in ipc and they are just as frugal, then why not mips?
 
I agree there ARM has been optimised superbly thanks to underpowered phones/consoles/handhelds from the past.


X86 suffered from the 'just put in a faster CPU' mentality for way too long...

I don't get the whole 'optimized for ARM' thing.


Yes, there are some applications that are optimized for some specific ARM instruction set but there is not that much of them because of the nature of smartphones/consoles. They are small and specialized as opposed to PCs.


Sure there are a lot of generic optimizations in some libraries for low-power devices but they are generic and can work on x86 too.


Another problem is that 'put a faster CPU' mentality is now fully transferred to arm space where is now mhz and core race is going on.


This race already passed the point where you can't just put the most powerful arm chip into device and expect great battery life from a good battery, my galaxy note which have a battery more than half the size of pandora's works much less then pandora under load, about twice less, so i don't expect a very powerful pandora 2 with good battery life under load unless it will have a battery like twice the capacity or more.


There should be a low-power atom with great linux driver support near an end of 2013 but it's too late for pandora 2.


until then there will be no real low-power (<2w for full SoC under load) x86 solution that will work well under linux.


By the way I think we will see 5w or more ARM solutions by then
 
The only benefit of x86 in a handheld is the ability to run some closed-source software from the short time interval between what can still be emulated full-speed on ARM (probably anything released before, say, 2000) and what needs the full power of a recent CPU/GPU, not just a low-power x86 chip intended for web browsing (probably most things released after, say, 2008). Also note that some of this software (e.g. game engines by id software) tends to be released as open-source software after some time, which makes it easy or at least possible to port it to other platforms like ARM (unless they heavily rely on x86 asm, but I suspect that to be very rare).


I'm not saying that this is not an important benefit for some, but e.g. for me it is no benefit at all, since I don't even have any software of that kind. We're talking about a relatively small time window (maybe 8 years or so, probably less) and only a specific subset of software from that time window: closed-source, x86-only software.


The disadvantages of x86 cpus for handhelds are huge: they get hot and they consume ~10 times as much power.


So this decision is really a no-brainer: forget x86. Instead, use your energy to put pressure on those software companies to re-release their stuff on ARM (this could be quite profitable for them) or release the code.
 
or so a Board inside:


Its Nvidia Ion Pico Itx Prototype Boards


motherboard2.jpg



nvidia_pico-itx_ion_graphics_reference_platform_2.jpg
 
who is to say that there can't be multiple pandora devices?

Probably the people who can barely scrape the resources to do one Pandora.

why is arm the most obvious(architeturally speaking and not sourcing/pricing) mips has their aptiv line which they claim beats the equivalent arm sku(?) in ipc and they are just as frugal, then why not mips?

It's not just about architecture. x86, ARM, MIPS, they all have their pros and cons. You aren't seeing the bigger picture. Manufactures using ARM have a much more position in the industry right now. OPT doesn't pick a processor, they pick an SoC. This SoC has to be available (to them in particular, not just to anyone), affordable, designed for the appropriate TDP, and have suitable features. Right now they know something like OMAP5 will check the proper boxes by the time Pandora 2 is ready to be released - what MIPS Aptiv SoC would they use? Has there even been a single one announced? Or are you running with it just because MIPS made a press release that the IP is available, and made their own claims about its capability? Guess what, processor IP by itself is worthless until someone picks it up. MIPS' position in the market right now is to end up in controllers for networking equipment and set-top boxes.


Maybe it'll be totally different in a couple years, maybe even sooner, but Pandora 2 needs to be designed now.


Seriously though, I don't really get you. First you say that OPT should use an SoC with an integrated FPGA, which would need mostly new development work to utilize it. Then you say it should use x86 to avoid having to do new development to utilize it. Now you're saying MIPS. I'm starting to think you're just enamored with bucking the status quo, or enjoy being contrary.

I don't get the whole 'optimized for ARM' thing.
Yes, there are some applications that are optimized for some specific ARM instruction set but there is not that much of them because of the nature of smartphones/consoles. They are small and specialized as opposed to PCs.


Sure there are a lot of generic optimizations in some libraries for low-power devices but they are generic and can work on x86 too.

It has nothing to do with smartphones. Emulators optimized for ARM started over a decade ago with GP32 and GBA. That's what we're talking about.
 
Last edited by a moderator:
Back
Top