Pandora Needs To Go X86 (eventually)

[second exodous]

This tread is kinda pointless, the Pandora is going to be ARM. If it does well the next revision will be ARM. If it does poorly then it might switch to another architecture. If it does poorly then maybe there will not be a next revision.....

The point is no one knows how well it will do, if it will be a big success or a big failure. No one even knows if it will depend on the chip it uses.

One has to go on what they know and feel personally. Me for instance have been waiting for something like this for a very long time, I am buying two units in case no one else I know gets it so I can play multi-player games.

If I go on I'll miss what I'm trying to say. I'm basically saying that what you say right now at this point of time won't affect what becomes of this Pandora and probably not the next. Time will tell. I don't really see anyone taking your side of your argument, you're here to argue with an entire community of Pandora supporters. You've let us know what you think no wait a few years to say 'I told you so' if you believe this so strongly and point to this thread.

Edit: I spelled success wrong.

 

[PoisonedV]

Topic starters need to get brains (eventually)

 

[Tozarian]

Why does everyone keep talking about Pandora 2 or Pandora 2.0? Talking that is like talking about an XBOX 720 or PlayStation 5- it makes you sound like an idiot.

On topic: X86 will murder battery life. That is all.

 

[White Flame]

MD: You keep saying that Pandora needs to move to x86. You should instead be thinking that more other things need to go cross-platform. That's where the new momentum is anyway.

 

[reyawn]

MD: You keep saying that Pandora needs to move to x86. You should instead be thinking that more other things need to go cross-platform. That's where the new momentum is anyway.

I agree. The cross-platform software library is growing strongly right now, so the momentum is not going to x86 - it merely appears like it is because it is backwards compatible with more existing COTS software products and thus more "mainstream friendly".

 

[cb88]

Chances are the pandora 2 will run a OMAP 4 ARM processor with a GB or 2 of ram and 10GB nand an so on

It will outright stomp whatever X86 offering will be out then... how you ask power usage

a Watt on an ARM is worth 10 or more on an x86

who knows we might not even be running Linux at that point viva Haiku!!!

 

[musicalwoods]

Everyone keeps talking about YouTube needing flash, and I find this highly comical. I found the secret to accessing their mp4 format (intended for the iPhone, I believe) quite a while ago, and a video download is just one easy bookmark click away anymore.

Also, ARM is aiming for the netbook sector, and is in partnership with Adobe to get Flash 10 out this year (probably for the freescale netbooks). TI is one of the OEMs that they will give it to free of royalties, but I don't know what it will take to get into the Pandora.

News Article:
http://www.adobe.com/aboutadobe/pressroom/...AdobeFlash.html


Oh, and the YouTube trick...
Create a bookmark containing the following (in 'Location' for firefox):

java script:if%20(document.getElementById('download-youtube-video')==null%20&&%20!!(document.location.href.match(/http:\/\/[a-zA-Z\.]*youtube\.com\/watch/)))%20{var%20yt_mp4_path='http://www.youtube.com/get_video?fmt=18&video_id='+swfArgs['video_id']+'&t='+swfArgs['t'];%20var%20div_embed=document.getElementById('watch-embed-div');if(div_embed){var%20div_download=document.createElement('div');div_download.innerHTML='<br%20/>%20<span%20id=\'download-youtube-video\'><a%20href=\'+yt_mp4_path+'\'>Download%20as%20MP4</a>%20'+%20%20((navigator.userAgent.indexOf('Safari')!=-1)?'(control-click%20and%20select%20<i>Download%20linked%20file%20as</i>)''(right-click%20and%20select%20<i>Save%20'+%20(navigator.appName=='Microsoft%20Internet%20Explorer'?'target':'link')%20+'%20as)</i>'))+'</span>';div_embed.appendChild(div_download);}}void(0);

 

[truekaiser]

Starcraft, Morrowind, Diablo series, Fallout series, Neverwinter Nights series, Battlefield series... all of that would fit on an 16 gig SD card, and you'd still have more than enough room left over for every single NES, SNES, and Genesis rom.

nope, every snes rom adds up to about 10gb. genesis roms about another 10, nes is close to 1gb. morowind is about 1 gb, nwn is at least 4gb.

 

[lulzfish]

...There's no easy way to get a legal copy of XP...

Oh really?

CODE
http://www.newegg.com/Product/Product.aspx?Item=N82E16832116511



I should have just said, "Windows is too freaking expensive"

 

[Kekerot]

I've been researching x86 vs ARM for a few months now because I want a more powerful handheld device and right now the choices are

10 hours pandora
limited software

2-4 hour intel MIDS
huge software available


I think pandora/ARM wins this generation of devices but when the next intel system on a chip comes out I think most consumers would rather get the x86 variant.

http://www.engadget.com/2009/02/16/intel-a...-hot-mid-action

Right now this is the best x86 UMPC available but it is modded, check it out

http://www.youtube.com/watch?v=S8YNkZ4pfYA

What would be awesome is if someone made an ARM/x86 hybrid chip or board. If we had the option to boot either that would be great.

 

[EdCa22]

So you like an hour and a half of battery life? Because even if you put an Atom in something the size of a Pandora, that''''''''s probably what you''''''''d get, max. Then you''''''''d have a much more difficult to design board because you need a lot of extra chips in an x86 (e.g. GPU) so you need more layers on the PCB as well so the whole thing will be even more expensive and take longer to design and test, and it may even be impossible to make it that small. Then you''''''''ve got heat issues. I''''''''ll place money that a machine as small as the pandora but with an x86 will get hot, and your hands will get sweaty. Or worse, burn.

Most of the size of the pandora seems to be the battery And as x86 mobile platforms get smaller and more efficient, there would be less radiant heat and all necessary extra chips would be incorporated into the soc design. I believe this is already being done and the products will only get smaller, faster and more efficient as time goes by.

Anyway, why argue about the hardware of the future? If software developers and publishers release source code for all their entire back catalogues, that would eliminate the backwards compatibility/legacy pc argument and let the hardware speak for itself.


It is true that x86 chips are becoming more integrated. If you read down I've then said "the performance of ARM chips with each iteration of the architecture is increasing at a rate that x86 can't beat". There have always been integrated x86 chips for embedded though, they just haven't really been that popular. I have an old intel print server that has virtually nothing in it but a modified 386 processor, RAM, ROM and an ethernet IC.

The fact is, despite the deflation of the desktop PC market, x86 processors as a whole are still not aimed at portable computing. They will get better but architecture developments will probably always be focused on high performance computing, meaning that any truly mobile (i.e. pocketable) version of a processor will always be somewhat crippled in terms of running the latest desktop software, which will continue to be written sub-optimally to run on these hugely powerful machines. We still don't have a really good portable x86 with a decent battery life that you can put in your pocket and can match the performance of PCs of 5 to 10 years ago, how will they catch up when ARM is already so far ahead? I think the movement is currently the other way - ARM has taken over the markets where you might expect to find (and used to find) x86 SoCs - think set top boxes, routers, and now netbooks. For the most part x86 will probably continue to be used for performance computing and is even likely to move into even higher performance computing (e.g. GPGPU - though admittedly that isn't really x86). It's speculation, but I think the new trend towards truly efficient, integrated x86 processors will not last.

Perhaps Material Defender raises a good point here in that there may be a small market for a pocketable device that is equivalent to the circa 2002 era PCs, just to play old PC games and some emus, but that market is roughly equivalent in size and demographic to the GP32 market, and most of them will have pandoras, and the usefulness of such a device will decrease with time as software houses continue to become more cross platform aware.

 

[Kekerot]

So you like an hour and a half of battery life? Because even if you put an Atom in something the size of a Pandora, that''''''''''''''''s probably what you''''''''''''''''d get, max. Then you''''''''''''''''d have a much more difficult to design board because you need a lot of extra chips in an x86 (e.g. GPU) so you need more layers on the PCB as well so the whole thing will be even more expensive and take longer to design and test, and it may even be impossible to make it that small. Then you''''''''''''''''ve got heat issues. I''''''''''''''''ll place money that a machine as small as the pandora but with an x86 will get hot, and your hands will get sweaty. Or worse, burn.

Most of the size of the pandora seems to be the battery And as x86 mobile platforms get smaller and more efficient, there would be less radiant heat and all necessary extra chips would be incorporated into the soc design. I believe this is already being done and the products will only get smaller, faster and more efficient as time goes by.

Anyway, why argue about the hardware of the future? If software developers and publishers release source code for all their entire back catalogues, that would eliminate the backwards compatibility/legacy pc argument and let the hardware speak for itself.


It is true that x86 chips are becoming more integrated. If you read down I've then said "the performance of ARM chips with each iteration of the architecture is increasing at a rate that x86 can't beat". There have always been integrated x86 chips for embedded though, they just haven't really been that popular. I have an old intel print server that has virtually nothing in it but a modified 386 processor, RAM, ROM and an ethernet IC.

The fact is, despite the deflation of the desktop PC market, x86 processors as a whole are still not aimed at portable computing. They will get better but architecture developments will probably always be focused on high performance computing, meaning that any truly mobile (i.e. pocketable) version of a processor will always be somewhat crippled in terms of running the latest desktop software, which will continue to be written sub-optimally to run on these hugely powerful machines. We still don't have a really good portable x86 with a decent battery life that you can put in your pocket and can match the performance of PCs of 5 to 10 years ago, how will they catch up when ARM is already so far ahead? I think the movement is currently the other way - ARM has taken over the markets where you might expect to find (and used to find) x86 SoCs - think set top boxes, routers, and now netbooks. For the most part x86 will probably continue to be used for performance computing and is even likely to move into even higher performance computing (e.g. GPGPU - though admittedly that isn't really x86). It's speculation, but I think the new trend towards truly efficient, integrated x86 processors will not last.

Perhaps Material Defender raises a good point here in that there may be a small market for a pocketable device that is equivalent to the circa 2002 era PCs, just to play old PC games and some emus, but that market is roughly equivalent in size and demographic to the GP32 market, and most of them will have pandoras, and the usefulness of such a device will decrease with time as software houses continue to become more cross platform aware.


I think you are wrong about x86 handhelds. Intel seems to be pushing them pretty vigorously lately. Watch this video http://www.youtube.com/watch?v=S8YNkZ4pfYA it's the best handheld umpc experience out there IMO and it costs about $500-600 and I've seen them as low as $300 on ebay. In one or two generations I bet the battery life will go up to 5-6 hours for a device that size and then why should a general consumer get an ARM device?

 

[lulzfish]

Well, of course the general consumer is going to want x86.

But ARM is still better in general, so I'd still want it.

 

[socket]

MD, it's this kind of thinking that keeps forwards development backwards! The more platforms out there the more platforms will need to be supported. You claim to hate windows but are obsessed with X86. I don't get it. Give it a rest. The ARM is overall just plain better. Instead of defending how good it WAS and WILL BE let's get things developed NOW for the BETTER chip.

 

[maciek_urbanski]

I agree that for the time being OMAP3/CortexA8 has better capabilities and performance/watt than Atom. But don`t underestimate weight of Intel, smartphones & MIDs are too big market to let go...

So the question is: "If there would be a Atom chip that is directly comparable (feature- and perf/watt-wise) to OMAP would be there ANY difference what`s integrated in Pandora ?"

And remember - most of users don't care about ISAs, SIMDs, cache sizes, etc...

(For the clarification - for all things that does not matter to user: engineering elegance, 3-argument ISA, instruction predication, multi-register memory transfers and their SIMD(maybe except lack of saturation arithmetic) - I'd choose ARM). ;]

 

[mali]

Wow, four pages of replies.
I'll put my two cents in, too.
Flash 10 will come soon(it's already there and should ship with the first ARM Netbook in summer) for ARM v7, just check MWC or prior press releases.
There will be a huge amount of ARM netbooks coming out this year, Pegatron(separated part of Asus) and Winstron to be the first.
There are much more ARM chips out there than x86.
Ubuntu for ARM will come out in April.
The biggest advantage of ARM Netbooks/MIDs will be their low price, that will inevitably bite off more and more of x86 marketshare in Netbook, MID, UMPC segment while simultaneously pushing software development/porting for ARM architecture.
ARM is developing hybrid designs, that will allow to integrate ARM cores in x86 devices and switch x86 off to do light tasks(like browsing, hardware accelerated media playback) while conserving battery life.
Intel should have a good next Atom in the works, because Cortex A9 and his SoC friends are serious competition. Intel will soon make an attempt to push Atom into Smartphones, which is in my opinion just the logical consequence to tha fact that ARM will invade in x86 territory. My prediction is, that it will become a huge failure because ARM technology is still at least one generation ahead in power consumption and will keep this advatage for a while. The x86 compatibility turns from an advantage into a drawback in the handheld/smartphone area, because you have to have more horsepower to accoplish the same tasks.
You have to take into account, that the x86 compatibility is tied to Windows, which is a huge resourcehungry beast. If they drop Windows on handhelds/smartphone the only thing that distinguishes x86 from ARM is worse battery life.
Intel has no option they must go to smartphones to at least try to sell something because they will slowly but steadily lose sales beginning in MID/UMPC/netbook and even desktop space(Nettops are coming, see Pegatron).
Another big problem is, that the conventional desktop is dying. The big sales-engine in the past was the gaming industry, which drops x86/desktop gaming design in favor of console development.

So my conclusion, x86 is dying for a few years already and this process will gain more momentum this year. Putting an x86 in Pandora is backwards.

It's absolutely the right thing to have an ARM core in Pandora, because of technical reasons and also to benefit from the ARM hype that is being created now and will have its full outbreak in summer 2009.

 

[fiveseven]

No, it doesn't.

Reading this topic just wasted 5 minutes of my life. I want them back.

 

[Kyo]

welcome to you completely missed the point. X86 would kill the pandora. The price would be doubled, the battery life 1/4th, the heat production multiplied and the audience nullified.

 

[Kramy]

Ok, so to clarify:

EVENTUALLY Pandora needs to go x86. Eventually means "not right now", because the embedded x86 blitz is just beginning. I am saying, if you have been paying attention, if you have struggled HARD and tweaked for weeks on end with Linux compatible ARM devices (e.g. Zaurus) like I have, you will understand just how big a deal x86 is when it comes to development and usability.

From my understanding, the current problem isn't with the CPUs so much as the accompanying chipsets. I will wager that within two years we will see x86 devices with over 50% of the battery life of ARM, easy.

You crack me up man. :lol:

Your experience with the Zaurus was probably quite painful. I know mine, trying to get java working on the GP2X, definitely was.

But I don't think the Pandora devs are like that. In the past 6 months, I've seen the amount of ARM-compatible software go up by about... 25x? It's crazy. In another year, almost everything linux-compatible may have ARM versions.

Your argument is pretty much that you want x86 because power consumption might go down and you need ActiveX. :unsure: I vote no. By the time x86 has a feasible alternative to ARM SoCs, ARM CPUs will be faster than Atoms, still consume way less power, have even more stuff integrated into them, and compatible software will be abundant. Your arguments only hold water if you compare today's ARM SoCs to x86 SoCs developed in 2010, which isn't exactly fair.

In other news, my Athlon XP is slower than a Core 2 Duo, and it doesn't run some software.

 

[ldesnogu]

If ARM is so darned "forwards", then will you please help me make my Zaurus C1000 awesome like it should be? I've been waiting for three years now. Been through five distinct projects, all of which utterly failed to produce anything approaching x86 usability.

You are talking of a device that uses a more than 6 years old CPU.
If you base your conclusions on ARM based chips on your Zaurus, there's no doubt your point of view and comments are wrong.