Power to the PowerPC clones

Over the last few months, I've heard a ton of weird rumbles all pointing to the PowerPC scene being a really interesting place about a year from now -- and things like Genesi releasing the specs and plans to their ODW hardware are a sign of that. Chances are you've never heard of Genesi, but I've followed them for awhile, primarily because there aren't a lot of people making PowerPC hardware you could throw on your desk if you had a mind to...
They've had a somewhat weird history -- and sometimes a shaky one -- going back to the Amiga OS. You basically pick up their machines if you're developing for the PowerPC in an embedded market or you're an enthusiast looking to play with custom kit. Hell, before the Mac Mini (which changed things), they were a cheap way to hook a Mac into your system for using Linux on PowerPC, etc.
By itself, their current hardware won't set the world on fire, but this why the above announcement is interesting:
- This is basically setting up PowerPC cloning, just not from Apple. Anyone who wants to -- primarily any company -- will be able to download the specs and implement their own PowerPC solution without having to go through the R&D of building everything around the chip.
- This is about Linux and other *nix's in a big way, as the PowerPC has basically been ceded to them, and the PowerPC isn't just going to go away overnight once Apple moves to x86.
- We know Genesi is going to be supporting IBM's PowerPC 970 variant (G5), which is interesting as they'd hitched their train to Freescale in a big way.
- IBM has a hand in a lot of this stuff, it's just not quite clear how much. I've heard from multiple people that:
- IBM isn't done with the PowerPC outside of the embedded market, and while they're making a big push with the CELL architecture -- perhaps because they control so much of the IP -- they're seeing Linux as a disruptive technology that allows the specifics of the hardware architecture to become less important, giving the PowerPC a new opening for Desktop hardware.
- Everyone I talk to says IBM is very, very unhappy with Apple over what happened with the PowerPC, but annoyingly they either conflict as to specifically why, or its just vague enough that inferring too much would be dangerous. The only real commonality in the stories that they feel they were screwed/blindsided, that they're pissed, and that they're seeding cash into places to see the PowerPC on the desktop get life under Linux.
Oddly enough I haven't been able to nail down how them being pissed factors into them doing anything they wouldn't otherwise be doing, outside of say, seeding some cash to see Linux on PowerPC optimized for some areas Apple has traditionally targeted, but the fact that they're pissed keeps coming up probably means something.
- We're going to see more of this type of thing over the next year, as IBM is pretty serious about it and is putting their money where their mouth is to give it momentum -- albeit indirectly and discreetly. I.E., I can't say that they're bankrolling some of what Genesi is doing, but they're slipping cash around.
If you're confused about the Power.org connection, the old Expanding the Power ISA's Influence post with the included Powerpoint file should fill in some of the blanks on how the licensing can work.
Some of what's going on with the rumblings around the PowerPC raise more questions than they answer -- but while they're weird, they're all pointing towards an interesting year for the PowerPC, just with free operating systems instead of OS X.
Comments (41)
Posted by: at November 18, 2005 11:36 AM
Uh, that's great for IBM and Linux, but if it doesn't run OS X (or Windows) who gives a flying fuck? They might sell some cloned PowerPC thin clients, but the desktop and PowerPC aren't fated.
Posted by: Ben Reubenstein at November 18, 2005 11:38 AM
I want to know what OS my G5 will run in a couple of years. I bought a brand new tower right before the Intel announcement. How many itterations of OS X until they drop PPC completely?
Posted by: mikeash at November 18, 2005 11:52 AM
How many itterations of OS X until they drop PPC completely?
Probably very many. They're going to be releasing new PPC machines until at least 2007, according to the announcements at WWDC this year. Beyond that, it costs them almost nothing to keep supporting it in their OS releases. Adding support for a new processor architecture can be a pain, but merely keeping support for an architecture that's already supported is not very tough.
Posted by: JHM at November 18, 2005 12:10 PM
"Beyond that, it costs them almost nothing to keep supporting it in their OS releases."
Except new hardware sales! :-)
JHM
Posted by: Anthony at November 18, 2005 12:49 PM
Everyone I talk to says IBM is very, very unhappy with Apple over what happened with the PowerPC, but annoyingly they either conflict as to specifically why, or its just vague enough that inferring too much would be dangerous. The only real commonality in the stories that they feel they were screwed/blindsided, that they're pissed, and that they're seeding cash into places to see the PowerPC on the desktop get life under Linux.I'm sure different people at IBM are pissed for their own reasons. It can't be a total surprise though.
I imagine their future plans involve Cell, while it might not make a great desktop or laptop chip I'm sure it'll be a great chip for certain types of workstation.
Posted by: Stern at November 18, 2005 12:53 PM
I had a Pegasos 2 machine (until I managed to break it), and while a bit underpowered it was a nice enough machine.
I think many big companies would be interested in cheap Linux desktops running OpenOffice, both for the cost and security benefits. It's not sexy, but there's money in that racket.
The rumours about Genesi's future plans hint at PPC970-based machines built around Marvell's newest system controller which supports PCI-E and other more modern goodies. This could bring the specs up to modern iMac/PowerMac level, all you would need after that would be someone to pack it up in a nice box and there you go. Obviously not a choice for the Siracusas of the world, but an option for those who got Macs as (relatively) inexpensive PPC-Unix boxes.
Posted by: mark at November 18, 2005 01:17 PM
IBM started power.org before Apple announced their decision to switch. It was just after Lenovo bought IBM's PC division. The intriguing thing at the time of the power.org founding was why Apple was not involved with power.org.
So as you say, this IBM move to Linux on PPC was not driven by Apple's decision to leave, unless IBM had mucho advance notice. But I think even the seeding of cash exception that you note doesn't have much to do with Apple's departure.
The whole Linux on PPC thing does seem to be very much aligned with the sale to Lenovo - and it could simply be IBM's desire to provide businesses with a real alternative to Microsoft on the desktop; one that would allow for IBM's software and services division to flourish even more.
Posted by: bbrv at November 18, 2005 01:40 PM
@drunkenbatman...:-)
Thanks!
http://static.flickr.com/32/64539431_7809180a33_o.jpg
BTW, the Evening at the Adler video was great.
Signed,
drunkenbatmanfan
Genesi
Posted by: at November 18, 2005 02:56 PM
Anonymous commenter wrote:
"Uh, that's great for IBM and Linux, but if it doesn't run OS X (or Windows) who gives a flying fuck?"
Honestly I don't give a flying fuck if a machine runs Windows or OS X. The real questions are: does it run Linux? and how well does it run Linux?
Posted by: LaoTzu at November 18, 2005 03:11 PM
The key word here is Desktop. IBM wants to license Power to Anyone, so they get to fab it. (a division still losing $$$ @ IBM - but I'll note most $$ is from services, as per Gerstner's business plan - OS, CPU, we don't care, we'll give a solution )
Desktops are bad biz now. Even Dell is losing a lot on them. IBM obviously is in the server market, and database, workstations... but they have no plans to ever make a portable CPU or use one...
Laptops are good, no white box clones, you can't built a $200 AMD laptop at CompUSA! Portable is the only growth area, desktops have plunging ASP's - heck Asia will ship $450 laptops now ( maybe $100 MIT Laptop is likely!)
Power was never meant to be in a Portable. Yet, console, blade, X Serve, Quad G5, it's great in... but even if Intel has dual 64 bit, virtualization, the chips will be slower than PPC - but 99% of code & compilers are much more optimized on x86. Apple spent a LOT trying to explain say a 500MHz iBook ~ 1.1GHz PIII + building their own compilers, AltiVec is great, but SSE tools & Intel/ MS make all code far more optimized for x86.
It's far easier to keep up with x86 is you are using it. Plus Intel advertises for you, pays billions EVERY year to optimize compilers ( and spec ;-) And for Apple, missed CPU ship dates have hurt sales more than any other factor...
Thrown in, a Power chip is actually NOT PPC. It has micro ops, and CISC like stuff. A 21 stage FP pipeline, vs Yonah/ P M @ 14 stages. x86 decodes x86 ISA - then uses a RISC chip to execute - then decodes back to x86 - so, not really a different architecture.
IBM did put some cash into GCC compiler, but mostly their unfree optimized libraries in fortran. Apple worked on GCC a lot, but Intel will let 'em use their compiler ( at least 15% faster ) but how cool is it switchers can install Windows on an eMac for legacy apps - no VPC required!
5 Years ago, I had this ? document, which discusses when Apple should switch to x86 - it ends in consensus - We just started developers on X - and cocoa was a developer dud, so we did Carbon, we need to wait.... Then, when Palladium, and Longhorn ship w/ TPM - we compete with them as less draconian, more secure, higher quality, in every area....and it mentioned VPod for same time frame....
Kills me, because five years talking to Apple guys who have secret BORG tech to work on, but must lock office door, people ask...you can't tell... so a relief it's out now.
Steve said @ WWDC 1st the LOW END ( think G4 Consumer!) next year, then hopefully by END of 2007 will complete ( Intel would have quad core 64 bit w/ HT, Virtualization ( running different OS simultaneously....
In closing, IBM is mad, they want "Power Everywhere"! Chip fab deals, plus some help on compiler optimization, etc. Apple has switched to a mostly portable maker ( and G4 w/ 167MHz bus is a bit rough ) Yet, desktops are the worst - anyone can assemble from shelf parts - the store helps you! For under 300 bucks? Over 50% of desktops are white boxes - no names - Dell reported a bad Q - the $299 PC was just to get people there - then upgrade DVD, Screen, etc. Losing $$ because - people bought when cheaper than built at store - and buying upgrades @ Microcenter....
I love the brilliance of PPC but at least Intel is switching to a more efficeint ( down from a 36 stage pipe in Prescot!!! ) so I see everyones point......but look foreword to watching what happens when steve isn't fired as soon as they launch a new platform - Gates will be destroyed....hahaha!!! And we will have a better, spam/adware/virus free world....finally!
DB rules!
Posted by: Chris Ryland at November 18, 2005 03:39 PM
Apple is smart to abandon the PowerPC. IBM is no longer a product company but a consulting company with big (but shrinking) pockets, so their PPC interest will eventually wane.
Posted by: William at November 18, 2005 04:42 PM
IBM is mad? Where did I put down the world's smallest violin, I know I have it around here somewhere.
So IBM, where is that 3ghz chip? Wasn't that scheduled for 2004? Why did it take so long for dual core? We knew about those dual-core chips for a couple years but Intel AND AMD still beat IBM to market with dual-core. Of course those AMD dual-core chips are too expensive, why it would probably cost $1000 to put together one of those machines.
Oh, those jokes about Pentium IV chips running hot were funny. Snails on fire ha ha ha. Very funny, until Apple had to use GIGANTIC heat sinks or LIQUID COOLING! to keep a G5 from burning up.
Mac users understood real well that Mhz were not everything, and Ghz weren't either. Real world results are what matter, right? Too bad we got toasted on a lot of those tests too.
I feel for you IBM, I really do.
Posted by: Anthony at November 18, 2005 04:51 PM
LaoTzu:
IBM's POWER and PowerPC chips are POWER and PowerPC because they implement the spec. Nothing else is required. They're not a classical RISC design, but classical RISC isn't competitive anymore so I don't see why that's a problem.
Intel's chips, even the Pentium 4s, do very well against IBM's PowerPC chips. When the dev kits came out, with one processor and lacking many optimizations, people still commented that it performed very well compared to dual-G5s. This is going to improve over time for Intel because they're abandoning netburst.
IBM is capable of making very nice chips but they are not motivated to make very nice chips for Apple. It takes too much specialization these days.
Posted by: Stevew at November 18, 2005 07:18 PM
They're going to be releasing new PPC machines until at least 2007, according to the announcements at WWDC this year.
Take anything Jobs says with an enormous grain of salt. Not long ago he was assuring the company's shareholders that the PowerPC had legs for the future. What he failed to mention was that those legs were not moving fast enough. Do we even need to mention the comments on a video iPod?
And who can forget the class action suits from their own users (that those users won) with regard to 'promises' that were never kept by Apple.
Intel Macs at the low end will probably be faster and cheaper than their closest PPC equivalent cousins and in all likelyhood will be able to run intel coded apps (through Windows or whatever) to boot.
However, I NEVER buy Rev A Apple products which are always buggy in some areas. It would be foolish to think intel Macs won't be twice as buggy. If we see MWSF announcements for intel macs then they will surely have been rushed out the door.
Posted by: Anthony at November 19, 2005 04:16 PM
I doubt they'll support PowerPC as a first-class architechture for that long. I think it's more likely that they'll do several major releases that support PowerPC, and then keep doing security patches to the last release for a while after that. I think 5-7 years isn't an unreasonable length of time for that to go on total, but I think most of that will be in minimal patch mode.
Also, I think Rosetta will be supported indeffinitely, until x86 is replaced (could be a long time now that x86 has consolidated the laptop/desktop/small server).
Posted by: Matthew Butch at November 19, 2005 09:59 PM
IBM being pissed helps support my theory that Apple decided to go Intel about a month before the announcement, and that not a whole lot of thought was put into it.
Posted by: at November 19, 2005 10:28 PM
IBM being pissed helps support my theory that Apple decided to go Intel about a month before the announcement, and that not a whole lot of thought was put into it.I imagine they has predetermined criteria and milestones that if they weren't met they'd switch. Apple doesn't exactly go out of their way to announce their plans to everyone.
Posted by: Pegasos II User at November 20, 2005 02:10 PM
Multipleboot on Pegasos II G4 machine is possible!
1) MorphOS
2) Linux
3) MacOS X !!!
Posted by: Pegasos II User at November 20, 2005 02:12 PM
PowerPC AltiVec is really really POWERFUL!
http://pegasos.jinak.cz/clanky/powerpc/ppc/rc5regression.gif
Posted by: Anthony at November 20, 2005 02:25 PM
PowerPC AltiVec is really really POWERFUL!http://pegasos.jinak.cz/clanky/powerpc/ppc/rc5regression.gif
Isn't that the benchmark that uses altivec but not SSE?
Posted by: VL-Tone at November 21, 2005 12:24 AM
What is it with people not wanting to believe that Apple will stop producing PPC Macs in 2007?
Apple will dissolve it's PPC I/O chip division before the end of 2007, they want to get rid of this weight as fast as they can.
Do you realize how much it cost Apple to design all the custom I/O chips in PPC Macs? It will be even worse while the PPC Mac market shrinks, it would be stupid for them to invest in new I/O chip designs for the PPC at such a small scale. It would be very costly to Apple to extend this transition beyond 2007.
For the intel Macs, Apple will take the intel I/O chips, they are not dumb enough to design custom I/O chips when intel can provide them the same chips in bulk and for cheap, and without any investment from Apple.
Also, answer this question: "At what speed will the XBox 360 PPC chip run in 1 year? and in 2 years? 3 years?" Ask yourself the same about the PS3. Then, some neurons maybe able to switch in your brain and make you realize that they will be running at the same speed and provide the exact same amount of power in 3 years... Also remember that Sony and MS had to invest Billions in IBM for them to build them these chips.
I've seen the PowerPC getting stuck at various stages of its evolution over the years (ie: stuck at 500 mhz for 18 months, stuck below 1GHz for an eternity etc.) and I've had enough.... Too many broken promises, I don't trust IBM/Freescale, they don't care about the desktop CPU market...
Posted by: Pegasos II User at November 21, 2005 07:49 AM
> Isn't that the benchmark that uses altivec but not SSE?
No! DNetC RC5-72 use SSE x86 too! PowerPC AltiVec is simply the BEST! My Pegasos II G4 1GHz machine is much better than PC x86 3.6GHz :-)
Posted by: Pegasos II User at November 21, 2005 07:56 AM
Reply to VL-Tone: It's not about MHz & GHz!!!
PowerPC has GREAT PERFORMANCE, see above... and very low power consumption.
Posted by: skTrdie at November 21, 2005 07:57 AM
I love PPC and I hope it will stay up as much as possible...I mean who wants a world with all x86 machines? Diversity is beautiful, and technically PPC is much more productive than x86, a company like GENESI and more could work with it and really come out glorious, especially in the open-source world (Linux).
Posted by: Anthony at November 21, 2005 01:22 PM
Pegasos:
Even if that's the case, RC5 isn't representitive of most desktop tasks. It runs almost entirely out of cache and good performance in a niche area (altivec running out of cache) doesn't make up for poor performance on everything else on a desktop.
G5s are a viable desktop processor, but they have their own problems (eg the laptop versions suck). Also, Intel sells Pentium Ms as low as 5 watts. With that little power, other parts of the system that both architechtures share start to dominate the power usage anyway.
Posted by: Peter da Silva at November 21, 2005 02:18 PM
The real questions are: does it run Linux? and how well does it run Linux?
The real answers are "Yes" and "As well as anything else that runs Linux". You'd have to really work hard to build something with an open architecture today for which the answers wouldn't be the same for any free UNIX ported to the hardware.
The real question is why anyone cares about a desktop that doesn't run commercial software. That's something I've been trying to figure out for, well, about the past ten years now. Windows NT on the Power PC and MIPS and Alpha had the same problem, and they had more chance of getting commercial software than Linux on Power PC does. Heck, even BeOS was clearly marginal at best... right from the start... and that had way more commercial developers behind it.
Power PC on the desktop lives and dies with Apple. Linux on the desktop is tied to the intel architecture because that's where such commercial Linux applications as exist run, so Linux on Power PC is irrelevant to the desktop no matter how good and how cheap it is.
Posted by: VL-Tone at November 22, 2005 02:29 AM
Pegasos II,
It doesn't matter than the PPC can deliver more power from the same MHz than x86 CPU's. I was complaining about PPC platform routinely getting stuck at some speed/power for a long time. When the G4 got stuck at 500Mhz for 18 months, performance didn't improve either, the same chip was produced for 18 months. Those things hurt Apple very badly over the years.
I defended the PowerPC platform over the years for being a much much cleaner architecture, that could provide more power per MHz. I defended Apple's choice because I thought the switch to intel was unfeasible, and because I was under the impression that IBM and Freescale would get their act together. Times and times again they failed to deliver. I don't see how it will be different this time with the PS3, XBox 360 and Revolution. Video-game console CPU needs are different than consumer PC needs. Apple can't afford 3-4 years before the next CPU revision.
Posted by: Peter da Silva at November 22, 2005 12:53 PM
Apple can't afford 3-4 years before the next CPU revision.
How about a few months?
Posted by: Anthony at November 22, 2005 03:41 PM
How about a few months?There always seems to be something on the horizon, but when it finally arrives it's usually late and usually not as good as it was supposed to be. And when it is (eg G5) it doesn't last.
Posted by: Peter da Silva at November 22, 2005 06:52 PM
Anthony... there's nothing wrong with the G4 core that a faster bus won't fix. Once you allow for the memory bottleneck, it's faster clock-for-clock than the G5... and a lot better watt-for-watt.
And over the last couple of years the G5 has actually ramped up in clock speed faster than the P4. Both Intel and IBM seem to have hit the same problem with the clock boost they get from longer pipelines having run out of steam. Which is why Intel's moving towards putting the faster bus from the P4 on the PIII core and calling it the Pentium M.
The MPC8641 is the same approach, and while I wouldn't be surprised to see Freescale dump it now that Apple's dumped them... it was on track to come out sooner than Yonah. Which is also "something on the horizon" right now...
Posted by: Anthony at November 23, 2005 12:46 PM
Anthony... there's nothing wrong with the G4 core that a faster bus won't fix.That remains to be seen. Athlon XPs used to do okay agaist Pentium 4s with dramatically slower busses, and a G4's disadvantage isn't that much bigger. Indeed, G4s haven't increased in speed that much since Athlon XPs were dramatically faster with their relatively small bus speed advantage.
I don't think the bus is the only thing wrong with those chips.
Which is why Intel's moving towards putting the faster bus from the P4 on the PIII core and calling it the Pentium M.There's more to Pentium Ms than that. There has been a lot of work done on the core, and it has no trouble beating G4s, G5s and Athlon64s at similar clock speeds. Note that they also have a significantly slower bus than G5s and Athlon64s and they still manage to be faster most of the time.
The MPC8641 is the same approach, and while I wouldn't be surprised to see Freescale dump it now that Apple's dumped them... it was on track to come out sooner than Yonah.With all the stuff they had on the die, I doubt it was just for Apple. I also don't think Apple dumping them is required to explain Freescale being late on delivering a chip.
Posted by: Peter da Silva at November 23, 2005 04:38 PM
Let's see. G4 - 166 MHz bus (well, new chips support 200 MHz but Apple's still running it at 166). Athlon XP - 333-400 MHz bus. Since when is twice the bus speed (even if Apple was using it at full speed) a minor difference?
With all the stuff they had on the die, I doubt it was just for Apple.
Let's see, they're doing a chip that just happens to have precisely the features you'd need in an integrated laptop. What market are they aiming for?
I also don't think Apple dumping them is required to explain Freescale being late on delivering a chip.
Errr... Jobs wasn't trashing Freescale for being late when they dumped the Power PC, he was blaming IBM for being late. So far as I know there was no plans for any Intel-on-desktop-Freescale-on-laptop policy, so (of course) Freescale got dumped along with IBM. If you know something more, I'd be interested in details.
And in any case Freescale was saying 2H05 for samples, so where's this "explain Freescale being late" coming from?
Posted by: Anthony at November 23, 2005 05:56 PM
Let's see. G4 - 166 MHz bus (well, new chips support 200 MHz but Apple's still running it at 166). Athlon XP - 333-400 MHz bus. Since when is twice the bus speed (even if Apple was using it at full speed) a minor difference?It's a relatively minor difference, AthlonXP's disadvantage compared to a Pentium 4's bus was larger than a G4's disadvantage compared to AthlonXPs. I'm not saying the bus isn't a problem, I'm saying they'd get better performance than they do if the bus were the only problem.
Let's see, they're doing a chip that just happens to have precisely the features you'd need in an integrated laptop. What market are they aiming for?Their primary market which is many times larger than their sales to Apple: embedded.
If you know something more, I'd be interested in details.You're the one that said it was set to come out before Yonah. I don't follow Freescale's activities much anymore, but I do know Yonah is sampling now for a January release.
Posted by: Peter da Silva at November 23, 2005 08:12 PM
I'm sorry, I must have missed something. It sounded like you're using the performance difference between the AthlonXP and the P4 to estimate how much difference the bus speed makes to the G4. I must be misunderstanding you because that makes no sense at all.
I apologise for being mistaken about the release date of Yonah. That one's my fault. Sorry. Then... the MPC8641D was set to come out close to the time Yonah came out... and making a new motherboard would have been a lot easier than what they're having to do now.
And, again, the G5 hasn't hit the 3 GHz target but it's ramped up more than competitively with Intel's processors.
Which is why I'm still skeptical of Jobs' explanation of why the intel switch now. I suspect it's at least as much to do with burying the corpse of OS 9 as with problems in the PPC lineup.
Posted by: av at November 23, 2005 08:33 PM
>Also, answer this question: "At what speed will the XBox 360 PPC chip run in 1 year? and in 2 years? 3 years?"
>Ask yourself the same about the PS3. Then, some neurons maybe able to switch in your brain and make you realize
>that they will be running at the same speed and provide the exact same amount of power in 3 years...
No, because the 65nm shrink is going to happen.
The Cell architecture will go to 45nm later.
Posted by: Anthony at November 23, 2005 08:34 PM
Which is why I'm still skeptical of Jobs' explanation of why the intel switch now. I suspect it's at least as much to do with burying the corpse of OS 9 as with problems in the PPC lineup.Assuming for the sake of argument that MPC8641Ds will be able to keep up with Yonahs, it doesn't address the overall problem: PowerPC laptop/desktop/small server performance is characterized by brief periods at the top separated by long periods of decline. If MPC8641Ds can keep up with Yonahs but Freescale doesn't make any serious upgrades for years after they're released, Apple is just as screwed. PowerPC has never given them a significant, prolonged advantage, and I agree with them that it's time to let it go.
Posted by: av at November 23, 2005 10:51 PM
>If MPC8641Ds can keep up with Yonahs but Freescale doesn't make any serious upgrades for years after they're released, Apple is just as screwed.
Even if Freescale won't make any serious upgrades anymore, there is this upcoming PWRficient architecture as replacement.
Posted by: Anthony at November 23, 2005 11:25 PM
Their release is a long way off and they're a relative unknown. If IBM and Freescale are unreliable, I don't see any reason to assume PA Semi will be any better.
Also, from the performance numbers they're claiming, it were available tomorrow it might be viable competition for Yonah, but it's unlikely that that will be as impressive a year from now.
Posted by: at November 25, 2005 09:05 AM
@VL-Tone
Mac I/O chips aren't designed by Apple!
Agere Systems produces de G4 northbridges and Broadcom the G5 ones.
So please, stop spreading FUD
Posted by: Peter da Silva at November 28, 2005 01:04 PM
I'm not arguing that maintaining a unique instruction set is going to be competitive over the long haul, I'm just questioning why they made the cut right now. The "3 GHz G5" and the "G5 laptop" lines are great sound-bites, but Intel is having the same problems with overheating laptops and CPU speed topping out, and both IBM and Freescale had promising chips in the pipeline (and Apple's since adopted IBM's dual core chips), and Freescale does have a roadmap past the MPC8641D.
So this looks to me like a bad time to jump ship.
So, what is it. Does Jobs know something really bad about problems at IBM or Freescale that he's not allowed to talk about and their own employees aren't in on?








Big Endian, Little Endian, can't we agree that all this bickering should Endian? All I'm saying is, give bits a chance!