Dreamcast would be awesome


1. Make it emulate more advanced systems such as Dreamcast, Sega Saturn, Gamecube and Playstation 2. Don't worry about anything beyond that, games beyond this generation wouldn't make good portable games anyway, especially with the trend for motion controls.
They don't need to focus on anything. Any sequel will be more powerful practically automatically. However, today's desktop machines with super fast dual and quad cores have difficulty emulating Gamecube and PS2. I think you're hoping for too much, even in the next several years.

2. Make it capable of running PC software. I realise the current Pandora is built using different software, however the next would benefit with being compatable with a lot of the Web 2.0 programs out there such as Skype. This depends of the x86 compatable processors though, which I would say are a priority for the next model... assuming they're viable.


3. Make it capable of running PC games. And extention of point two really, still, having a mobile games console with a full qwerty keyboard is begging for RTS games.


4. Make it compatable with the most popular MMO's if possible. World Of Warcraft, Everquest 1 & 2, Aion, Star Wars: Old Republic and EVE Online... I realise this would mean at least 2GB of RAM to make this a reality for the last two at least, but it's still possible.
x86 gives you the ability to run Windows games, true, but it sacrifices a lot of energy, produces more heat, becomes more complex, etc...


There's a reason ARM processors are frequently chosen for handheld devices. An x86 processor is not likely to happen unless Intel can fix some of the major flaws in the design.

5. Built in webcam, with preferably pre-loaded software. This system is desperately in need of a webcam/camera to make the most of it's portability.


6. Pre-packaged wireless (or even wired) controller. Sure it makes a great handheld, but connecting it to your TV via AV cables and using a wireless controller (much like the 'Yobo FC-16 Go' SNES emulator), would make this a good console for use at home as well.
As Prometheus says, these are already there. The webcam isn't built in, but you can take a great number of modern webcams and it is perfectly plug and play. Adding a webcam built into it will only increase the cost and won't add anything for those that just aren't interested. You may as well suggest it have GPS or accelerometer.

7. This next suggestion would be difficult to implement but it's worth mentioning. Include a phone system. If this was able to read SIM-Cards and act as your phone this would make the system so much more popular. The phone should probably be used via touch screen to avoid cramming yet more buttons on it.
No. You'll just double the cost of the thing for just-another-smart-phone, one that isn't even very suited to making phone calls. With the added cost, you'll drive away existing customers that don't want a phone. You want a phone, buy a USB stick and do it yourself.


THAT is the one and only thing a Pandora 2 should have: continued use of user accessible connections (I would hope USB3 is well established by the time we get there) and the attitude that if you want a feature that isn't available the tools are there to add it yourself! Ok, I guess that's two things.
 
even though most of us dont even have the first pandora....
 
One that has been had many many times before...


Pandora 2 will be an Omap 4 with the same case that it has now... Quote me on that in 2 years when it is officially announced ;)
 
Wow I took a long time to write that. :p Conversation moved too fast.

Honestly though, I don't understand why x86 is such a turn off. As I've said, I don't actually understand processors and the like. I'm not a technical wizz, although I understand enough that I'm considered the "computer guy" by my friends (so you don't need to simplify things too much, I won't go looking for the ANY key). Why are people against the use of x86? Why would it guzzle battery power?
Put it this way: The Pandora has a 4200mah battery and gets between 8 and 14 hours of continuous use with the wifi off.


My laptop has a 7000mah battery and doesn't run more than 3 hours of continuous use.

I've honestly looked, and the only mobile internet device that can run the average MMO is a laptop/netbook. Hense why I thought it would make any potensial next model unique and much more sort after.
Oh, the Pandora has the power to run MMOs of comparable quality, even with it's low processor and seemingly small ram size. It simply can't run the existing ones because of lack of x86 instructions. And no, I don't think this would be a worthwhile tradeoff to lose 10 hours of usage.

a cheap phone costs £10 or less now... and you could always not use that feature. :p
It's not that easy. Cheap phones are usually available at the end of their life cycle, after the company has already made back licensing costs on them. Today's $20 phone was yesterday's $100 phone.

but if this Pandora 2 was released and you're not a fan of x86 processors and mobile phones, you could always stick with the original Pandora. It's not like the community would drop it, they seem vervently proud of it (I'm glad to see).
Maybe not drop, but it would necessarily create a rift. ARM programs can't be run on x86 anymore than x86 programs can run on ARM. If Pandora 2 included an x86, none of the existing programs would be able to run on it. We're talking NES vs SNES here. Anyone that chose to write something for the old Pandora would either have to accept it would only work on the old Pandora, or they'd have to maintain two versions. It seems to me that future compatibility is an important thing to maintain here, unless there is a very good reason to break it.

Pandora 2 will be an Omap 4 with the same case that it has now... Quote me on that in 2 years when it is officially announced ;)
It'll be an Omap 5 and it'll be 3 years from now, mark my words :p
 
Of cause people want and are discussing a possible Pandora 2, even though most of us don't yet have our Pandoras... when people see a good product the first thing they do is pick up on the things they feel would improve it. I agree if there are ways of making it so that people can add additional features then it would be better for those who don't want those feature. It's just for a portable device, having to carry peripherals around for everything, dongle for phone, webcam etc...


Perhaps we're considering this incorrectly. Perhaps we could have 2 different types of Pandora 2.


1. The Pandora 2 Stark - A minimalist design, just a faster sleaker version of the Pandora. Only the basics with ports so you can connect any additional features you want.


2. The Pandora 2 Baroque - An elaborate design, all the bells and whistles, phone, camera, pre-packaged controllers and cables for an "all inclusive" feel. More expensive yes, but more saleable and user friendly.
 
x86 gives you the ability to run Windows games, true, but it sacrifices a lot of energy, produces more heat, becomes more complex, etc...
There's a reason ARM processors are frequently chosen for handheld devices. An x86 processor is not likely to happen unless Intel can fix some of the major flaws in the design.

Another thing is that you lose any hope of compatibility with the current generation (not to mention platforms like Android), and you lose out on a lot of ARM-specific optimization done for Pandora.


In the context of this thread, this has a really big implication.. while Pandora is out people can work on difficult projects like Dreamcast emulation that the Pandora might not be fast enough for. But if Pandora 2 comes around then all the work will hopefully prove fruitful - but not if a lot of work was done in hand writing and emitting ARM, particularly NEON code.


Now, as far as power consumption is concerned, the latest Atom CPUs + GPUs (Moorestown) at 800MHz shouldn't consume drastically more than Cortex-A8 + SGX 530. However, even with Moorestown you still need 3 chips to get the same functionality something like Pandora gets in two, and I don't think Moorestown itself has a PoP connection with RAM. So you need the space for, RAM, NAND, and the third chip. This makes routing much more complex and uses a lot more board surface area. I think there's a reason we still haven't seen Moorestown phones.


That aside, 800MHz Atoms are hardly anything to write home about. I'd rather have the Cortex-A8 personally, even if Intel claims that the Atom is performing so much better clock for clock even on a similarly sized memory bus (me, I think that its biggest advantage is in better compilers, and that goes away when people write their own ARM ASM). For those wondering, the 1.6GHz Atoms that are very popular on netbooks aren't comparable here - in order to get that clocking they need to be ran on a much less energy efficient process, meaning considerably less perf/watt. In fact, Cortex-A8 and Scorpion are currently being more aggressively clocked on their low power consumption processes than Atom is, and Cortex-A9 as well, despite having a much shorter pipeline than Atom. Here's hoping Global Foundries delivers on its 28nm and 20nm claims for Cortex-A9... things are going to heat up and Intel's process advantages are going to get slimmer and slimmer, all the while still weighed down by architecture...
 
Last edited by a moderator:
Without a doubt, ARM is the way to go. Consider the current OMAP3 chip, pulling 10+ hours of battery life when overclocked to a ridiculous degree (upwards of 1ghz), off a 4000mAh battery. Intel is working on better mobile chips, but they are lightyears behind ARM in this department.


The thing to keep in mind is that ARM is growing extremely fast, reaching ever further market penetration in the smartphone and MID department. While currently, you're often looking longingly at your x86 apps/games, I absolutely garauntee you're going to find more and more apps and games written specifically for the ARM platform.


Look at the millions of ARM handhelds in circulation today, and the pace of development on them, and the rate of evolution: mainstream folks tend to hold onto full PC's for ~3-5 years, and even then they're buying consumer PC's with well behind cutting edge hardware. Hell, go into an average big box store right now and likely the best video cards you'll find on a shelf will be around the ati 5750 range; which is considered entry level at best. Conversely, people often exchange their smartphones every year or two (as carrier upgrade plans allow), keeping millions of regular people with top of the line smartphones.


The small form factor, low power consumtion device market is an enormous growth market.
 
I just ordered my Pandora! I hope the 2nd doesn't come out for awhile lol I just spent 370 on the thing ;) . I want to at least be able to hold it before i have to get a new version of it... oh wait this isn't Sony right? ;)
 
The Pandora has been a few years in the making, so you don't have to worry about seeing a Pandora 2 any time soon.


Investment notwithstanding, it would yet again take a few years to plan for what the second instalment would be about (hard/soft).


Of course, the earlier it starts, the better it would be in the long run (that's while being optimistic).


Hmmm, maybe we ought to make a "Pandora 2" thread, if there's not one already...
 
Oh no, x86 Pandora 2 again
dry.gif



Before everything else, I want my Pandora 1 software to run on Pandora 2.


x86, as for now, isn't viable for handheld gaming system : it runs too hot, eats battery like there is no tomorrow, is awfully expensive and is not as compact as an ARM SoC.


I don't want the pandora team to make another windows platform, another generic, albeit small, computer.


The Pandora still have a lot of unexploited/unexplored power, let's make games specifically designed for it, even MMO games!


To have internet everywhere, I'll either use my GSM via usb or bluetooth, later maybe a 3G or 4G dongle (you just need a small USB cable and a velcro patch to hide it under the pandora). I never used any webcam on any computer, I don't need a webcam in my Pandora.


Why would they package controllers? Just pick any bluetooth or usb controller, connect it to the Pandora, and voilà. If you really want a Pandora joypad, just wait a bit, they already work on getting the Pandora and the iControlPad in the wild
tongue.gif
. Maybe, if the Pandora home console (without screen or battery) goes in production, they will make openPandora licensed joypads
wink.gif



As for Skype, there are many others voice over IP protocols, but if you really insist, Skype is currently trying to make an API called SkypeKit, that will allow softwares like Pidgin to connect to Skype services. Just wait some more, SkypeKit is currently in Beta phase.


Post Scriptum : If someone is making an mmo, please make it less "generic". Every time I try a new mmo, I feel like I'm playing the same game with slightly different graphics. Got that feeling when I tried another mmo after playing the Lineage 2 beta.
 
Last edited by a moderator:
@TornadoCreator


It really sounds like you would be better off buying a netbook and a gamecontroller rather than the Pandora.


I'm hoping that the Pandora 2 will be based on the Cortex-A9. When I want to play x86 games, I'll do so on my desktop.
 
http://drk.emudev.org/blog/?p=143


Interesting. He is claiming Dreamcast games execute 120-200 MIPS. 40% of that is movs and saveT, which can be combined with other instructions by the dynarec, but still that leaves around 100 million per second. That's about 5 times what N64 emulation requires, and probably out of the range of current ARM hardware.


I'm guessing that "static" refers to PC-relative constant loads. I don't quite understand what "inline" refers to. Still, 20 million non-const loads/stores per second is a lot, about 4x what N64 emulation requires. I wonder what the L2 miss rate would be like on an OMAP3530.
 
Well it seems like even the Windows Dreamcast emulator is a mess right now with little support. NullDC or whatever and someone ported it to PSP but never released the code because it was in beta and prob. will never get out of beta since there is no way the PSP can run a dreamcast emulator. the PSN announced that Sega is porting some of its Dreamcast titles to be down-loadable content at the PSN store which is pretty cool but that might only be for the PS3 =/ and deff. won't support the network functions and some of the 3 party games :( . If the Pandora 2 comes out in 2012 hopefully we will have enough time to use it before the world blows up ;) jk.
 
http://drk.emudev.org/blog/?p=143


Interesting. He is claiming Dreamcast games execute 120-200 MIPS. 40% of that is movs and saveT, which can be combined with other instructions by the dynarec, but still that leaves around 100 million per second. That's about 5 times what N64 emulation requires, and probably out of the range of current ARM hardware.

I wonder how he gets this number though. SH4 is superscalar, but not with two true ALUs, and I think there will be many internal stalls and cache misses to direct mapped instruction cache and data cache. I really doubt it can sustain anywhere around 1 IPC like this, but the biggest question I have is how this is measured without some really expensive emulation of cache and bus activity that is more timing correct. I think this is more a matter of allowing the game to run this many, not that the game necessarily would on real hardware. That it varies at all shows there's probably some stall emulation, but just for relatively simple things.


It's possible he's profiling on the real hardware, but there's no way he'd have gotten this sort of detailed per-instruction histogram, just at best number of instructions vs number of cycles. I'd actually be surprised if the hardware has performance monitors at all.


What I've heard is that Dreamcast games run fine limited at 100 million instructions per cycle.

I'm guessing that "static" refers to PC-relative constant loads. I don't quite understand what "inline" refers to. Still, 20 million non-const loads/stores per second is a lot, about 4x what N64 emulation requires. I wonder what the L2 miss rate would be like on an OMAP3530.

I thought he was saying something like 27 million out of 33 million loads were constant. If L2 misses are bad on an OMAP3530 then there'd be even worse L1 performance on a Dreamcast.. and I doubt the 100MHz SDRAM is that fast at servicing cache misses (but better than the RDRAM in N64)
 
Last edited by a moderator:
Post Scriptum : If someone is making an mmo, please make it less "generic". Every time I try a new mmo, I feel like I'm playing the same game with slightly different graphics. Got that feeling when I tried another mmo after playing the Lineage 2 beta.
Have you tried Eve Online, Dungeons and Dragons Online, Horizons: Empires of Istaria, Dark Age of Camelot, Neocron 2?

Edit: ARM is really awesome.


:p


Just wanted to add that I don't care too much for processing power, as long as certain stuff runs. Longer battery life is the really desirable factor for me. As if we were discussing ARM Pandora vs. X86 Pandora, which we are not. Maybe I should go to bed.
 
Last edited by a moderator:
I wonder how he gets this number though. SH4 is superscalar, but not with two true ALUs, and I think there will be many internal stalls and cache misses to direct mapped instruction cache and data cache. I really doubt it can sustain anywhere around 1 IPC like this, but the biggest question I have is how this is measured without some really expensive emulation of cache and bus activity that is more timing correct. I think this is more a matter of allowing the game to run this many, not that the game necessarily would on real hardware. That it varies at all shows there's probably some stall emulation, but just for relatively simple things.
Almost all modern emulators have idle loop detection, I assume that's what drk||Raziel is doing. I doubt he's modelling the internal stalls and cache misses.

I thought he was saying something like 27 million out of 33 million loads were constant. If L2 misses are bad on an OMAP3530 then there'd be even worse L1 performance on a Dreamcast.. and I doubt the 100MHz SDRAM is that fast at servicing cache misses (but better than the RDRAM in N64)
If the Dreamcast RAM is really 64-bit and 100MHz as often claimed, the cache misses probably aren't that severe. 60ns DRAMs were pretty common in 1999, and if it can really transfer 64 bits per cycle, it could probably service a cache miss within 15 cycles.
 
Back
Top