Personally, it looks like AMD is gaining ground on Intel thanks to their ownership of ATI. AMD has demo'd two different families of APUs already (combined CPU/GPUs).

ARM is at a disadvantage architecturally when compared to X86. Its a cruder processor and some of its features lead to significantly less dense code.
But ARM is gaining ground constantly. However, if you were to compare current ARM offerings and processor like those I referenced earlier, X86 still has a performance edge.

I don't see any clear winner here yet except the consumer. When there is competition, increased innovation occurs and prices are lower.

[ Edited by Jim on 2011/1/21 20:40 ]

Hey did you see the news about IBM working with ARM to create processors at the finest process yet?
First it was Global Foundries demoing 28nm processes by make ARM processors. NOW IBM has created 14nm ARM processor.
What disadvantages ARM may have, it looks like they're going to keep their low draw lead over the competition.

Quote:

The reason I don't think that Denver will be a complete Soc (or that at least the server Models won't be) is that the Southbridge functions built into a Soc (like disk i/o, usb, etc) do not need to be duplicated across a multiprocessor platform. Look at the AMD processors I referenced. They combine CPU and GPU functions and some traditional Northbridge functions into an APU but they still rely on an additional IC for support.
This is closer to what I picture Denver looking like.

> http://www.hpcwire.com/blogs/NVIDIA-ARMs-Itself-for-Heterogeneous-Computing-Future-113025584.html
> "As part of this strategy, the company has obtained rights to develop its own
> NVIDIA-designed high performance CPU cores using ARM's future processor
> architecture. Presumably this will be based on a future 64-bit implementation
> of the ARM ISA, given that 64-bit computing is the accepted standard outside
> of the mobile space."

From the same author, 6 days ago:

"At this point, it's not clear if NVIDIA intends to use the A15 as the basis for its first CPU-GPU processors or wait for a full 64-bit capable ARM architecture, which at this point is still under wraps."
http://www.hpcwire.com/features/ARM-Processors-Set-to-Challenge-x86-On-Its-Own-Turf-115153179.html?page=2

Seems he's gotten confused over time ;-)

Update:

> recent BlueGene system is built around 8192 "Power BQC" processors
> at 1.6 GHz each, which I assume have PPC A2 cores (like PowerEN).

Confirmation:

"IBM's hardware people cleared up the mystery about the processor at the heart of the BlueGene/Q super, saying that it is [...] a 16-core Power A2 processor"
http://www.theregister.co.uk/2011/02/08/argonne_ibm_bluegene_q_super/

More info:
http://power.org/wp-content/uploads/2011/03/IBM_March_29_Webinar_-_Dr._Luigi.pdf (pages 7 and 26)


Edit: PDF link updated

[ Edited by Andreas_Wolf 18.09.2012 - 23:15 ]

Update:

>> The speculation that ARM will be used as a basis for a design that
>> interprets X86 code (similar to Transmeta)? No, I can't see that.

> Thanks for addressing this speculation. I'm with you on that one. It's hard
> to see that such a solution could make sense performance-wise. Even the
> author of the article circulating that speculation agrees with us on that.

News from that author regarding x86 compatibility of Denver:

"T50/Tegra 5 is now known as Denver, but the x86 part has been dropped for legal reasons."
http://semiaccurate.com/2011/04/06/nvidia-in-full-philosophical-retreat-for-tegra-3/

> Nvidia's Denver might be 64 bit [...]. [...] I am unsure
> about whether or not Denver will be 64 bit or not

CEO of nVidia on Denver two months ago:

"To take the ARM processor, partner with them to develop a next-generation 64 bit processor to extend it so that all of computing can have the benefits of that instruction set architecture. It is backward-compatible with today’s ARM processors."
http://venturebeat.com/2011/03/04/qa-nvidia-chief-explains-his-strategy-for-winning-in-mobile-computing/

Quote:

Interesting link! So it will be 64-bit. And it is backward-compatible with today’s ARM processors.

"ARM is now the only CPU in the world that will have deep penetration in the mobile devices, the PC, servers and supercomputers."

ARM is the future!

[ Edited by takemehomegrandma 12.05.2011 - 06:51 ]

"ARM is now the only CPU in the world that will have deep penetration in the mobile devices, the PC, servers and supercomputers."

Propaganda. You could add PowerPC there too. At least there are more PCs, consoles, cars, servers and supercomputers using PowerPC than ARM.

Most ARM cpus have poor performance per mhz compared to PowerPC or Intel cpus.

Quote:

No, not really. ARM rules the mobile world, while Atom and others tries to scale the x86 downwards into this segment. PPC is nowhere to be seen.

x86 rules the desktop and workstation segments, while ARM tries to scale upwards into these segments. PPC is nowhere to be seen.

Quote:

PC's based on PPC are a thing of the past. Probably Consoles as well. Anyway, you are saying "currently" as a response to what was said about "future". ARM and nVidia is aiming to do an x86 killer, with a clear ambition to compete with x86 on desktop, server and supercomputer markets. Microsoft has joined in, which clearly adds weight to the effort.

Despite this you want to equal PPC to ARM's future? Sure, feel free to call it propaganda until you see some real results (especially if you distrusts nVidia's competence in creating high performing designs, despite them being one of the leading entities in this field), but don't claim that PPC has a better course laid out for it, when all applications that's left for it is in various boring embedded gadgets and cars.

> Probably Consoles as well.

"01net also claims to know some of the technical specifications of the new console (translation from Develop): 'CPU is custom IBM PowerPC with three cores [...]'"
http://en.wikipedia.org/wiki/Project_Caf%C3%A9#April_2011

> all applications that's left for it is in various boring embedded gadgets and cars.

You forgot gaming consoles, servers and supercomputers. Power Architecture is still big there currently. Whether ARM(64) will supersede PPC(64) there remains to be seen.

64bit ARM is about two years into the future, Current ARM processors, in their current state of development, still aren't as powerful as PPCs.
Where the future will lead is anyone's guess, but for the time being ARM is no threat to PPCs or X86 processors.

Quote:

Could very well be that also the next generations of game consoles will be based on some PPC derivate one more time, nobody really knows until it gets public. Rumors about everything are everywhere. I wouldn't be entirely surprised if Microsoft and nVidia already have something ARM-based cooking for the new Xbox though. And why not a Xbox handheld as well while you're at it, using the same ISA, with all the latest wireless stuff included...?

Quote:

And in the context of this thread (if ARM would be where MorphOS could go now that there are (and won't be) no more suitable PPC HW), it's completely irrelevant. Likewise in the context of Crumb's post to which I replied. *I know* that PPC and Power are still used today. But denying the fast expansion of ARM and the buzz surrounding it because of PPC still being used doesn't make sense. Crumb dismissed nVidia's chiefs words ("ARM is now the only CPU in the world that will have deep penetration in the mobile devices, the PC, servers and supercomputers.") as propaganda, and said that PPC is big in cars and consoles (none being an area addressed by Jen-Hsun Huang's statement) and PC's (Huh?! Where?) and Servers (Well, at least the POWER CPU's still have a market share for certain server segments) today. OK, but *it won't* compete with ARM in mobile devices, and neither will x86. Thus I think Jen-Hsun Huang's words may hold ground and shouldn't be dismissed as "propaganda" quite so easily; he may actually know what he is talking about regarding his own product development; if he says they will release CPU's that will compete with x86 in PC, server and supercomputer markets, chances are kind of good that they will do just that. If so, that would make his statement quite true, right?

> nobody really knows until it gets public.

Yes, that's exactly what I wanted to convey. You say the next generation gaming consoles probably won't be based on PPC whereas others claim to know the Wii's successor will be based on PPC. We'll know more about that in less than 4 weeks time.

> why not a Xbox handheld as well while you're at it, using the
> same ISA, with all the latest wireless stuff included...?

I'm not aware of any announcements or even rumours regarding an "Xbox handheld". Wait, you just started it ;-)

> in the context of this thread [...] it's completely irrelevant.

Yes, as irrelevant as "boring embedded gadgets and cars". Yet you mentioned it. I just corrected you on your claim regarding "all applications that's left for" PPC as you swept three of the true applications Crumb mentioned under the table.

> Likewise in the context of Crumb's post to which I replied.

...and claimed that of the applications he mentioned only "boring embedded gadgets and cars" were "all applications that's left for" PPC.

> denying the fast expansion of ARM and the buzz surrounding
> it because of PPC still being used doesn't make sense.

Yes, that's why I don't do this.

> Crumb [...] said that PPC is big in cars and consoles [...]
> and PC's [...] and Servers [...] today.

...and supercomputers ;-)

> If so, that would make his statement quite true, right?

Yes, it's just that I neither responded to Huang's statement nor to Crumb's response to Huang's statement but to your incorrect claim about "all applications that's left for" PPC.

Quote:

Well, "the future" is kind of the essence of this thread, not the past, not the current!

Quote:

Nobody has ever claimed that there aren't faster PPC CPU's than what's currently available from ARM, but I actually think my Efika MX would kick at least Sam440's butt. Especially so when they get the SW properly tuned. There are numerous other PPC CPU's that certainly would get a fight as well. And does Cortex A9 (multi core (up to 4), 2GHz+) count as "current"? Well, at least there are products out based on the Tegra 2, and Apple has its A5 used in iPad 2, both will probably put a great deal of PPC CPU's to shame.

Nobody is building consumer products based on PPC anymore that you could surf on, chat on, use e-mail on, edit your photos on, etc. There are no more PPC PC's or laptops. I have two ARM based devices in my possession (Efika MX Smartbook, and Smarttop) who both does that just fine. And look how well Windows 8 and MS Office performs on *current* (not future) ARM HW: http://www.youtube.com/watch?v=xKc_XGuvNIk (FF to 1:10 into the clip)

Quote:

...currently no (and neither PPC nor x86 is currently no threat to ARM processors), but point is, ARM *will* challenge x86 on its traditional markets (it will kind of start with the release of Windows 8, not with the release of "Denver"/"Tegra 5"), and frankly, I have very difficult to see x86 challenge ARM on its traditional markets. And this is what Jen-Hsun Huang says; ~ "ARM will be the only CPU that will exist in *all* those segments".

> I actually think my Efika MX would kick at least Sam440's butt.

Rather than guessing why not perform some benchmark tests between your 800 MHz Efika MX and an 800 MHz Sam440? I'd be interested in the results. But mind you, the Cortex-A8 is just an in-order execution core whereas the PPC440 can operate out-of-order.

> There are numerous other PPC CPU's that certainly would get a fight as well.

Of course. Current PPCs scale from as low as 40 MHz ;-)

> Nobody is building consumer products based on PPC anymore
> that you could surf on, chat on, use e-mail on, edit your photos
> on, etc. There are no more PPC PC's or laptops.

THTF/MTC ceased production of LimePC devices based on PPC? I ask because they still offer those in their web stores:
http://www.iseeuon.com/web/shop/shop.php?app=goods&goods_id=14
http://www.iseeuon.com/web/shop/shop.php?app=goods&goods_id=15
http://www.thtfit.com/cpe/shop.php?app=goods&goods_id=14
http://www.thtfit.com/cpe/shop.php?app=goods&goods_id=15

And also Cherrypal:
https://www.cherrypal.com/secure/product_info.php?products_id=1

@Andreas_Wolf

Just out of curiosity about your own personal opinion (since you obviously travels the net to collect facts about so many things):

Which one of PPC and ARM do *you* think has the brightest future (in a 5-10 year perspective)?

What option do you think would be the best for MorphOS, provided the developers wants to keep going for another 5-10 years:
1) Stay forever on PPC
2) Migrate to ARM (doesn't necessarily mean abandoning PPC)
3) Migrate to x86 (doesn't necessarily mean abandoning PPC)

Quote:

Which kind of was my point - there are a great deal of flavors of *current* PPC and ARM chips, and since not all of them (in fact, not a single one of them) has the ambition to be the fastest CPU on the planet, but rather be a suitable solution for a certain application, then Jim's statement that "Current ARM processors, in their current state of development, still aren't as powerful as PPCs" doesn't really hold ground; there are *many* areas where *current* ARM CPU's can (and does) compete with *current* PPC CPU's, of which the CPU used in LimePC is a very good example of! Genesi jumped to i.MX as soon as their LimePC developments with THTF crashed and burned, and I don't think they have regretted that one single second. The i.MX family has long lasting road map, the "mobileGT" kind of reached its end station there and then (in the context we are speaking about at least); if more CPU's will follow in that series, they probably *won't* be anything you could use to make a "LimePC 2" of. If a new version of "LimePC" will follow, it will definitely be based on ARM, not PPC...

> (in fact, not a single one of them) has the ambition
> to be the fastest CPU on the planet

One could get the impression that POWER7 is very close wouldn't it have strong in-house competition from z196 ;-)

> Jim's statement that "Current ARM processors, in their current state of
> development, still aren't as powerful as PPCs" doesn't really hold ground

Yes, a fair comparison is not that trivial. Best performing ARM processors currently are based on cores implementing ARMv7-A ISA (i.e. ARM Cortex-A9, Qualcomm Scorpion, Marvell Sheeva PJ4) and it's clear that their performance is only a fraction of POWER7's (which is best performing PPC currently) performance. But for a fair comparison we'd have to take also wattage and price into account, calculating ratios like performance per wattage or performance per price.

> The i.MX family has long lasting road map

Like the 4+ GHz quad-core Cortex-A9 based i.MX63 processor for 2011? ;-) I'm glad Freescale removed this nonsense roadmap from its website.

> the "mobileGT" kind of reached its end station there and then
> (in the context we are speaking about at least)

Don't tell that to Zylesea. He says he'd like to have a cheap product based on MPC5125 (yes, in the context we are speaking about here):

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7001&start=563
https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7713&forum=3&start=8

> If a new version of "LimePC" will follow

I wasn't talking about new versions but about the *current* ones that you chose to neglect in your previous statement on "consumer products based on PPC".

> Which one of PPC and ARM do *you* think has the brightest future
> (in a 5-10 year perspective)?

If you're talking about desktop computing and brighter (not brightest) future then I'm inclined to say this would be ARM, considering recent developments and announcements.

> What option do you think would be the best for MorphOS, provided
> the developers wants to keep going for another 5-10 years:
> 1) Stay forever on PPC
> 2) Migrate to ARM (doesn't necessarily mean abandoning PPC)
> 3) Migrate to x86 (doesn't necessarily mean abandoning PPC)

I would want to keep backwards compatibility (to m68k at least), thus option #3 is out for sure. Whether I consider option #2 viable depends on the answer to a yet unanswered question about a certain feature of current (or future) ARM processors which would be required for backwards compatibility reasons:

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=11&topic_id=6726&start=87

More thoughts:
https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7771&forum=3&start=15

>> Which one of PPC and ARM do *you* think has the brightest future
>> (in a 5-10 year perspective)?

>If you're talking about desktop computing and brighter (not brightest) future then I'm inclined to >say this would be ARM, considering recent developments and announcements.

And of all current ISAs?
Personally I don't see either of these supplanting X86 any time soon.

> And of all current ISAs?

x86(-64) of course.

Thanks Andreas. As much as I'd like to see an alternative, for the foreseeable future 64bit X86 does seem to have the dominant position.

And while ARM is attractive (especially from a price perspective), its unlikely that an ARM port of MorphOS would have enough power to run PPC apps.
68K apps, obviously, could be run. But a great deal of the software I have is MorphOS specific.

Quote:
Why are we talking about the past ? "back"ward... A port to a new architecture would be a reason to look in the future, and forget about the past.

It would be like constructing new train lines, but running old trains at low speed over it: what's the point ?

So, not thinking about running previous apps (ppc,68k): what about new perspectives ? x86 ? arm ?

Don't wanna go arm/x86 just to run 68k/ppc apps...

[ Edited by Leo 13.05.2011 - 06:51 ]

Quote:

...wasn't really the issue here.

Quote:

Sigh.

Most people in this community who casually makes comparisons of various CPU platforms, usually does that based on one single point of view; Traditional Desktops. Like Crumb and Jim above. Because in their view, this is what Amiga/MorphOS is about. Problem is that nobody has been developing PPC desktop CPU's since 2007. x86 is the only desktop CPU today, so when desktop-oriented guys makes casual comparisons about architectures, it all really sums up to what's available from Intel and AMD. A new, viable, competitive PPC based desktop motherboard will *never, ever* be seen again! PowerPC has been "reduced" to be application CPU's after Apple moved away from the architecture, which is what ARM has been from the beginning. This is where these two architectures compete today (if anywhere), in various specific applications, not on a general desktop arena, because PPC simply isn't there anymore (even if some people tends to think so, based on the second hand mac market or whatever) and ARM hasn't got there yet (but is about to).

So if the application is, for example Nettop's and Netbooks, then the PPC camp would probably put forward the 5121e, and the ARM camp could choose from, say, the i.MX51, i.MX53, Tegra 1 or Tegra 2. All of those ARM chips performs much better than the PPC and has far better features. One other of the the target application areas of the mobileGT 5121e is automotive infotainment, but I'll tell you what - many car manufacturers are working with ARM and Android now for exactly this area. Many of Freescale's ARM CPU directly targets applications previously owned by PPC. So yes - they do compete, and ARM most often comes out on top in comparisons (performance, features, etc). Home servers/NAS units are other kinds of consumer applications where ARM would compete (and beat) PPC, and then there are *numerous* of embedded areas as well.

POWER7 doesn't really fit the picture...

Quote:

Freescale is obviously at least a year behind the competition in the ARM market; while various companies are now pushing Cortex-A9 CPU's, Freescale is only now pushing their i.MX53 which is still Cortex-A8 (albeit faster than i.MX51 in many ways). But to suggest that Freescale wouldn't have a future planned for their i.MX effort is nonsense; this would probably be their most important segment now for potential growth. Roadmaps are snapshots of how future plans look at that moment in time, nothing else, and future targets are *moving* targets, so much can change after such a snap shot has been taken, so they are only relevant for a limited amount of time. Your whole perception of reality seems to be based on links and web references, but that gives a very poor view of the reality, and it's a good thing that at least Genesi isn't relying on linked images on websites to get info; they are friends with the people who are making those road maps in the first place, so they (and others) have a better source or info...

Quote:

You don't understand why I choose to neglect it, do you? I choose to neglect it because it's *painfully irrelevant* and not because I don't know about it, so there is no need for you to try to "enlighten" me. It's a parenthesis at best. Or rather: it *was* a parenthesis, it's quite old now and nothing new based on PPC followed it, so you know what - I'll happily *continue* to neglect it and stick to my statement that nobody is building any (for MorphOS) interesting products based on PPC, and nobody will do it in the future either.