Indeed.
Indeed x 2. What Konstantinos is (was?) attempting here is almost suicidal anyway - given the whole worldwide crisis. Money was much easier to find just some years ago. Remember Genesi giving away computers to developers, and stealing shows abroad in the early days of the Pegasos?
Hard on spot.

@all

1. NAS device != desktop motherboard. The one is an appliance with a specialized functionality, where hardware specific needs come second, the other is a computer. I wouldn't even post on amiga/powerdeveloper if I was after a NAS device.
2. a modular design is a very nice concept, but it increases the design costs AND time exponentially. Let's start simple.
3. Consider this an experiment. I could fund the mobo design OR a properly done "market research" (ever asked the costs from a marketing firm to do a market research?). The investor would fund the first (and subsequent production runs). Even as limited as thus, this "market research" did give me an idea of the general interest wrt PowerPC viability.

I'll blog post today as I couldn't do it on Sunday, but the result is that as long as PowerPC gets directly compared to Intel/AMD offerings, performance wise, it just won't cut it. Even Power7 will be trounced in a few months by latest Xeon/Opteron offerings so performance is not the #1 key point of PowerPC cpus anymore. However, it does have some nice points still.

I'm working on a last effort with this, please bear with me.

Hi Konstantinos,
I think this will never change, especially in the eyes of potential investors.

That said, I would like to see an 8610 powered computer, and I hope you can manage somehow.


Johan.

_________________
Johan Dams, Genesi USA Inc.
Director, Software Engineering

Yep, I have a blog... PurpleAlienPlanet

Yes, my point was that users sometimes expect miracles -eg. a super-amiga or a super-ppc mobo for $300 that would beat the crap out of their friends' i7 PC. That's not going to happen. Choice was given out of 3 possible CPUs. It would be dumb to provide NOW complete specs when even the CPU is not decided yet (though even right now most possible choices still remain the 8610 and the P1022).

P1022 seems to have a future and a long roadmap, it is just not desktop oriented. I must admit, that this CPU is my personal favorite, but I'm very interested in servers and appliances...

On the client side ARM seems to have a brighter future. (And ARM is working on the "server" side too, I have a Sheeva plug on my desk, performing nicely.)

_________________
CzP
http://czanik.blogs.balabit.com/

You could make a very nice product out of the 8610, even if there aren't a roadmap for other products following it. Evolutions in SoC designs oftenly calls for a complete new motherboard design anyway, so there will be no benefit from "forward compatibility". The DS107 NAS I mentioned above was PPC based, the DS107+ wasn't. That was no biggie for anyone, neither the users, nor the manufacturer. And the question here was about a new *PPC* motherboard, not which architecture he should choose. And fact is that no released ARM chip I know of can compete with the 8610 in performance. I like ARM and all, but that wasn't the issue here.

Who's asking for a NAS device?
In both the 8641D/8640D and 8610, the PCI(e) is already on the chip, all you have to do is making a motherboard with the connectors. What could be simpler?

My point was that if you make the motherboard small as the Efika and thus leave the PCI(e) expansion slots as an *option* (by connecting a bus board to it), you could find *even more* uses/markets/customers for the design that goes far beyond being "only" a desktop. For example laptops.
But if you're not asking real questions, how could you expect any significant answers? And doing market research without first even knowing *yourself* what you want to achieve and what you want to measure is kind of pointless. You should at least produce some kind of *concept* that one could discuss.
What, was this the moment you discovered that PPC isn't an option for the true, "outside this community" desktop market?

The last PPC Macs were the last PPC desktops, and that was *years* ago.

+1

:-)

To ppc or not? Just some thoughts:

On the live market there are a few nice ppc based systems (ps3/xbox360/wii) and some 'old' macs.
All the consoles are quite capeable, but not alone for their ppc base, but most to their graphic capeabilities (read 3d). For that matter take a look at what can be done with a ps3 with linux on it and you find it a bit disapointing (or that could be me). That brings me to next point. Drivers that support the hardware! E.g. take a 3d graphics card, but there are only 2d driver and partial 3d acceleration on linux. For intel cpu there are binary drivers, but those are not good on ppc.

For me i am now most charmed with the beagleboard that has opengles 2.0 3d acceleration. It boots into x-windows within half a minute next i can start abiword and edit all the text i want. Who needs a powerfull pc (with or without ppc) for that. And with it being 3d capeable (and it realy is) through a binairy (not yet public released) driver. For a matter i have not even missed features like ethernet or harddisk connectors yet or it not being ppc. My compiler makes just as easy i386 code as ppc and even arm binaries.

And for those who want to do massive calculations the gpu is the future way (cuda/opencl) and with opencl it does not even matter if you have nvidia or an ati videocard.

But what is most important for me is the power usage of an system and how that stands to the use of an system. E.g. Running linux on a ps3 just to edit text is a waste of the environment. Same goes for running tux on the efika1.

But the average user does not care less. For them it just has to work. And in a way they are used to. E.g. run windows and word and ie. If it cant it probeable is not good. (question i got from a friend: how does windows run on that beagleboard?) That is where the power of intel and windows lies and where the weakness of ppc/arm and amiga like os's/linux is.

Having another new ppc platform wont change anything and probeably would not do anything that cannot be done on another cpu also. arm can it do more powerefficient and i386 just hast faster models for now.

So what makes the ppc so fantastic that the competitors dont have?

Exactly, no new old-fashioned computer.
I like my beagleboard also, but there are things it cannot do. Eg. play 1080 video. Also, the beagleboard is some sort of a hacker's toy. I'd prefer it if it was in a different form factor with a bit more expandability, builtin ethernet and better sound. Oh wait, sth like the efika mx!! :)
OpenCL/CUDA is fine, but not all problems are convertible to the massively parallel GPU units. Though I guess things will get better as the technology matures. I agree that Cell has lost the game though.
Agree on that.
PowerPC could/might fill a gap between x86 and ARM. It has much better performance than ARM (at least now and probably for a while) and is less power hungry than x86 offerings. It actually is a better fit for a netbook than both ARM or Atom.
I honestly don't know what to reply here.

[sarcasm]Perhaps it's time we change the name of the forum to ARMdeveloper.org or x86developer.org[/sarcasm].

it's not fantastic, it's just pretty darn good at what it is - a 'personal computer' class cpu design. allow me to draw a parallel with arm here*. both arm and ppc are openly licensable, highly customizable design lineups. as such they are subject to fragmentation, 'factioning', or general 'multiplicity' of the architecture. now, the ppc is rooted in mid-performance-class platforms (well, practically in server designs, if we follow the ibm trail, but anyhow), and its fragmentation is spread around a median which is very 'personal-computer'-like. the arm, OTOH, is a design that has been de-facto dwelling at the bottom end of human computational needs till not long ago. only very recently did ARM produce a 'personal computer' class design (referring to the A9** here, the MP in particular). both architectures are fairly free of dead-weight intellectual baggage (e.g. x86_64 lugging along x86) and both are quite power-efficient (being open makes them prone to be pushed to even higher power efficiency, in contrast to intel's 'if it sells windows it must be good', and who realized only yesterday that efficiency matters), and both are quite reasonably positioned at the price/performance curve. one of them, though, sits closer to the 'personal computer' spot on that curve, and the other - to the 'cheap is king' spot on the curve. now, there's a reason why all consoles this generation are ppc designs - consoles always compete with desktops performance-wise in the eyes of public, while at the same time are also extremely cost-sensitive.

so if you want today the best price/performance looking for 'personal computer' performance levels, and you can afford large scale production (clearly markos' project is not that) and you're not willing to play by intel's 'we give you a free atom with each POS chipset you buy from us' policy, then you'll need to seek in the ppc space. maybe tomorrow things will change and the A9 and beyond will be the small-personal-computer chip to have. or, hey, maybe intel will stop being the fat dragon sitting on the treasure and will start thinking about what the consumer wants. or maybe the consumer will decide they've stayed long enough at the dragon's cave attraction, and move on to the next one, whatever it is. *shrug*


* from the position of somebody who enjoys
(a) 4 arm 'pmp/communicator' devices, is expecting another one (the highest-end one - smartbook class), and does on-and-off development on two (soon three) of them. also
(b) somebody who has been running one ppc for a desktop/server and another one for a server/desktop (note the inversion) for sufficiently long time.

** technically, A8 was ARM's first shot at 'high perfromance, low power' and it fell short in a few aspects, in my opinion: its scalar fpu is a waste of silicon (as little area as it is) - your 603e smokes it, its simd is not bad but i'd take altivec, thanks a lot, and last but not last, it's an in-order design, which'd be ok in itself, if it wasn't for the rather longish pipeline (admittedly with branch prediction) - ppc's of this pipeline length offer at least some out-of-order features.

Now that this thread has turned into comparing ARM vs. PowerPC... Why hasn't anybody compared them provider-wise?

I'd say that there's a hell of a lot more healthy variety and competition in ARM than PowerPC. If you can't find a nice ARM chip for your needs, just go to another provider for more options. Take for example, Konstantinos has given us two realistic options for a PowerPC CPU. How many could he give, if the project was ARM oriented?
The largest PowerPC provider (IBM!) pulled the plug long ago, leaving the second largest (freescale!) almost alone... and now commited to ARM.

Of course I'd love a new PowerPC computer (my pick has always been MPC8610), but like others, I believe that it's only the "powerdevelopers" (and only a fraction of them) the people that see some value in it.

None right now. The Cortex A9 chips are not out yet and until we see specific models, there is nothing in the ARM world that compares performance wise with the 8610/P1022, and let's not even think about comparing to the 8640D.
If we are the market itself, then we're doomed... :)

Well, if we still are talking about the 8610, then it's more powerful than any released ARM (AFAIK) and it has better heat characteristics, power consumption, and lower price than any x86 capable of this:

http://www.youtube.com/watch?v=5BT1299hgI0

That's the 8610 in particular, not PPC in general. You can do many interesting things with the 8610.

well, i have an arm pocketable here that can do that. no, it's not using dedicated video circuitry, but it's not doing the heavy-lifting on the arm either. so the arm is not its muscle.